Home

Rebuilt and Modified Altec 1567A: A Technical Report

Contents

1. lt Vintage Unit Output i g amy F E Faders FF ae oe lt Standby site Pre triode Attn Pad Z Pol Pad witches Bal Inputs Outputs CH1 CH2 CH3 ne ons Output Microphone Input Transformers Top Brace yfmr CH5 CH4 CH3 CH2 CH1 Right V4 V3 V2 Side ge Panel Vintage Vintage Unit Pwr Xfmr Left Side Panel 120 VAC Power Shielded Jack Hi Z Area of Toroidal Auxiliary Pwr Xfmr Panel Power Su Board Bal Output Driver Boards 18V PRY 12V Bottom Brace CHI CH4 Mounted Under Plate Figure 1 Front upper image and rear lower image views of modified Altec 1567A with some controls assemblies and parts labeled After briefly describing the original stock Altec 1567A will outline the modification s main features After that will describe the modified unit in detail and finally provide some observations on its performance and applications 2 The Original Altec 1567A 2 1 General Description of Original Design This initial description applies to all vintage Altec Lansing model 1567A microphone preamp mixers Notes on the configuration and condition of the particular unit that was modified are in Section 2 2 For your convenience a schematic diagram of the original stock Altec 1567A is reproduced in the Appendix at the end of this PDF document Figure 2 is a block diagram of the audio signal path of a stock Altec 1567A configured as a m
2. s 60 Q maximum spec but 14 greater than the nominal 50 Q These solid state devices emulate transformers by using feedback to control the two complementary outputs legs As a by product of this approach the driver effectively doubles the input voltage 6 dB gain at high load impedance or nearly doubles it 5 5 dB gain working into a 600 Q load the difference is due mainly to voltage drop across the output impedance This is because the voltage swing of each complementary output leg matches that of the input so across a floating differential load the difference voltage is twice that of the input In single ended mode where one output needs to be grounded exactly as a transformer balanced output is used in single ended mode feedback automatically forces the opposite output to nearly double the input voltage Using a single output leg without grounding the opposite leg is not recommended since this lets common mode noise normally squelched by feedback appear at the output this increases the chip s normal noise by over 40 dB Normal driver noise is described further in Section 5 14 Similarity to transformer balanced outputs breaks down at the clip thresholds not just because of the distortion type hard clipping but because the clip threshold is about 5 dB lower for single ended mode than differential mode at 600 Q load That s because an individual output cannot swing beyond a limit set by the power supply voltages and feedb
3. capacitors Some modification or repair had been performed around the fader pots and mix bus on the front panel but it was left wired according the original schematic According to the original manual an optional assembly provides XLR input jacks and other connectors this particular preamp was not so equipped and screw terminal strips were the only input output I O choice Modification will eliminate these strips and provide I O jacks on the auxiliary panel The VU meter was also optional equipment but this unit was equipped with a working one and a single working 44 lamp half the number required Lamp type is not specified in the schematic or manual so used two 47s which run cooler than 44s The unit had four model 4722 input transformers and one model 15095 output transformer The shield can on three of them seemed too loose so glued these to their bases with small dabs of epoxy cement at their crimp points For the 4722 s marked the channels they came from 1 4 upon extraction All tested good for continuity and no shorts using an ohmmeter but 3 had significantly lower than expected impedance on AC performance tests more on this in Section 5 2 The two secondary windings of the 15095 output transformer did not have equal DC resistances see Section 5 20 While later tests showed that audio sounded good using it its function may be sub optimal As described in Section 4 10 decided to add a rear chassis mounted switc
4. 1 4 Frequency Response Channels1 4 Unbalanced Output Impedance and Applications Channels 1 4 Balanced Output Characteristics Channels 1 4 Understanding and Adjusting Clip Alerts Channels 1 4 Types of Distortion Channels 1 and 2 Input Transformer Saturation Threshold Channels 1 2 and 5 Pre Triode Attenuator High Z Pad Applications and Effect on Bandwidth Channels 1 4 Noise Channel 5 Applications and Input Characteristics Channel 5 Keeping Track of Knobs Channel 5 Variable Feedback Feature Channel 5 Gain and Bandwidth Channel 5 Noise and Distortion Channel 5 Output Impedances and Transformer Characteristics Channel 5 VU Meter 6 Appendix Original Altec 1567A Schematic ODDO OMDAO W 1 Introduction This report describes the extensive re build and modification of a 1960 era Altec 1567A microphone preamplifier mixer It is a form of breakout modification in which individual input channels of a vintage preamp mixer are given separate buffered outputs This suits today s multi track studio ecosystem better than a dedicated monaural mixer While such a modification need not eliminate the original mixer function in this particular case it does it can still be reconstituted externally if desired The old master channel is made completely independent a fifth channel A main goal here was to maximize options for and control over distortion and coloration by allowing the channels to be linked i
5. 5 dBV output from CH5 is much more distorted on order of 3 THD than that of CH1 CH4 near 0 5 THD see Figure 25 will discuss CH5 s distortion characteristics in general terms next Naturally SNR improves as input signal amplitude increases at any given fader setting Of course the trade off is that distortion also increases Given CH5 s multi triode signal path the relationship between input level fader setting feedback setting and output distortion is 57 complex Unlike with CH1 CH4 I did not examine distortion in CH5 systematically but can offer a few general comments and sporadic observations The immediate post fader triode stage V3B has limited headroom and is easily saturated notice in Figure 14 that its cathode bias is only 83 mV This is related to the triode s saturation threshold for signal peaks at the grid although the amount of negative feedback applied to the cathode dynamically affects the bias more feedback effectively increases the saturation threshold In any case input headroom at the V3B stage is not nearly as high as it is for the solid state fader followers used in CH1 CH4 but in exchange driver stage clipping in CH5 is soft rather than hard Thus CH5 offers multiple but interacting ways to tailor distortion effects to emphasize input stage V3A distortion make the channel s input signal relatively high and set the fader relatively low To emphasize output driver distortion do th
6. AC operated heater circuit shared with the VU meter s two lamps The power supply uses solid state rectifier diodes 2 2 Pre Modification Condition of This Project s Particular Altec 1567A This section describes some specific issues with the unit noticed before its disassembly and how some of these affected re build and modification plans The serial number of the vintage unit is 1188 Distinctive blemishes on the front panel are a few deep scratches above and to the left and right of the power switch knob and two small holes drilled under the Altec logo on either side of the words MIXER AMPLIFIER Old repairs to the unit were fairly easy to identify none appeared recent within the last many years Green dye factory quality control was visible on unperturbed solder joints Although the AC power cord was a replacement there were multiple cracks in its outer sheath This was not a concern because modification will provide a single standard IEC style AC power jack behind the auxiliary panel Referring to part numbers in the original schematic see Appendix of the four chassis mounted multi section can electrolytic capacitors C17A B and C20A B were replacements attached by 4 40 screws and nuts instead of the factory rivets Although C19A B C and C1A B C D were original C1B had been disconnected and replaced with a pair of 25 uF axial units in parallel All other parts appeared original including all signal coupling
7. Bandwidth Unless noted otherwise all of the figures reported in this section were with variable feedback set at the design feedback level see preceding section the tone controls neutral and the pre triode attenuator full clockwise The channel fader was full clockwise for gain measurements At the balanced output which was set for nominal 600 Q secondary windings linked in series and terminated with a 600 Q resistor channel gain for a low amplitude 1 KHz sine wave input measured 54 5 dB with the balanced output still under load gain at the unbalanced output measured 70 1 dB into a 1 MQ load or a calculated 70 9 dB if the balanced output were not loaded The difference in channel gain seen at the balanced versus unbalanced outputs represents the voltage difference between the 55 secondary and primary sides of the output transformer As detailed in Section 5 20 this combines the effect of a 4 90 1 voltage step down ratio and resistive losses in the windings For the nominal 150 Q setting secondary windings in parallel the transformer step down ratio was 9 94 1 after accounting for resistive losses at that setting channel gain at the balanced output measured 50 1 dB with a 600 0 load and was calculated as 48 5 dB for a 150 Q load Unlike for CH1 CH4 did not measure the gain of CH5 s individual stages just the overall channel gain as given above However have made the following estimates Since CH5 s first triode stage is ve
8. CH1 CH4 are driven by solid state buffers linked to the wiper of each channel s fader pot more on this below The direct wiper signals are available at unbalanced high Z outputs 3 3 Special High Z Circuitry in Channels 1 and 2 Absent in CH3 and CH4 the additional high Z circuitry in CH1 and CH2 is built in a shielded enclosure of the auxiliary panel internal partitions help limit crosstalk between channels Important performance limitations of this build out are given in Section 5 13 Trade offs aside it adds the following features to CH1 and CH2 High Z Unbalanced Input This 14 inch jack bypasses the input transformer to patch unbalanced signals directly to the triode grid circuit At 1 KHz input impedance is 840 KQ when the pre triode attenuator is full clockwise and the high Z pad is off This is a normalled jack meaning it normally links the input transformer secondary to the triode Inserting a 4 inch plug there automatically opens that connection and replaces it with the ground referenced signal at the plug s tip Pre Triode Attenuator Pot The main practical application of this pot and the pad switch described next is to reduce a signal s amplitude when driving the input transformer to saturation for distortion effects Without attenuation the signal would be much too high to isolate the effect of transformer distortion due to simultaneous triode super saturation The normal attenuator setting is full clockwise but w
9. CW 108 3 74 3 10 103 3 69 3 20 94 3 60 3 30 84 4 50 4 40 74 4 40 4 1 Channel gain 54 5 dB design feedback level used nominal 600 Q balanced output terminated with 600 Q 2 With input amplitude giving 20 5 dBV balanced output into 600 Q at OdB fader attenuation for easy comparison to CH1 CH4 s noise figures given in Section 5 14 However CH5 s harmonic distortion at this output level is much higher than that of CH1 CH4 see text CH65 s 108 3 dBV EIN figure for the full clockwise fader is very close to the corresponding figure for the unbalanced input of CH1 or CH2 106 7 dBV see Section 5 14 This reflects the similarity of the front end triode stages and the full fader minimizing the output stage s relative noise contribution A more practical difference in CH5 s noise compared to CH1 CH2 becomes apparent with fader attenuation applied In addition if one compares the signal to noise ratios SNRs of the channels for a given output amplitude CH5 s higher driver stage gain causes a lower SNR even at full fader To illustrate the above table shows SNR computed for a 20 5 dBV output so it may be directly compared to CH1 CH4 s SNR given in the table in Section 5 14 At full faders CH5 s 74 3 dB SNR is 12 4 dB worse than that of CH1 CH4 which is 86 7 dB From there the SNR difference increases even more as faders are turned down due to CH5 s much noisier output driver as mentioned previously Finally a 20
10. LINDBERG MODEL Y2365 2 16 VA TOROIDAL TRANSFORMER PRIMARY NOT SHOWN DIODES D1 D4 1N5401 D5 D20 1N40a4 CAPACITORS R2 C1 C4 C9 C10 1uF 50V Stacked Film 150 C5 C8 47 uF S V Electrolytic C11 C14 47 uF 35V Electrolytic All RESISTOR Values in ohms gt 12y Figure 6 Schematic diagram of power supply in the modified Altec 1567A s auxiliary panel See Figure 4 for transformer s primary circuit The secondary windings of the toroidal power transformer each rated nominally 18 Vays are hooked in series for 36 Vams Output with a grounded center tap D1 D4 form a full wave bridge rectifier with bi polar DC outputs referenced to ground and filtered by C5 and C6 Bypassing each rectifier diode C1 C4 are snubber capacitors intended to silence RF switching noise The 12 V outputs are unregulated and depend on the voltage drops across R1 and R2 they are filtered to low ripple by C7 and C8 Correct voltage here requires invariable loads in the relay control circuits served As detailed in Section 4 6 six of the relays those of CH1 and CH2 operate on 12 V while the other six for CH3 and CH4 use 12 V each relay control switch routes current to either a relay or an equivalent dummy load resistor to maintain constant current in the 12 V circuits The regulated outputs 18 V and 15 V use heat sink mounted one amp voltage regulator ICs U1 and U2 LM7818 and LM7918 Each of them has three diodes D5 D10 i
11. OUTPUTS INPUT 7 Xx Polarity Input Transformer Balanced Line Driver Altec 4722 Treble Su Pre Triode Daa CHANNEL 5 CHANNELS a Impedance OUTPUTS INPUT Line Output Transformer Altec 15095 Figure 3 Block diagram of audio signal paths in modified Altec 1567A See Figure 2 s inset for key to the symbols used 3 Overview of Modified Unit 3 1 General Architecture Figure 3 is a block diagram of the modified Altec 1567A s signal paths Compared to the original unit Figure 2 notice that the modification trades the original mixer function for channel independence Channels one through four CH1 CH4 are based on the original unit s four microphone preamps Channel five CH5 is derived from the former master channel Each channel has a balanced low impedance low Z output at a male XLR connector and an unbalanced high impedance high Z output at a inch female jack 3 2 Channels 1 Through 4 These channels share a common basic layout except that CH1 and CH2 have additional circuitry in the high Z link between the input transformer and triode stage see next paragraph All transformer coupled low Z balanced input circuits have pad polarity and Z select switches The normal setting for each toggle switch is the down position These switches operate relays installed in the vintage chassis near the input transformers so sealed relays with gold contacts handle the signals for enhanced reliability The balanced outputs of
12. PRE TRIODE VARIABLE ATTENUATOR ALL RESISTOR VALUES IN OHMS R4 PRE TRIODE MF METAL FILM RESISTOR 10M CF ATTENUATOR 1 TOLERANCE WHERE 174W 2x TOLERANCE WHERE 172W CF CARBON FILM RESISTOR T 5 TOLERANCE 174W OD POLYPROPYLENE FILM CAP ACITOR SPRAGUE ORANGE DROP HIGH IMPEDANCE ux UNBALANCED INPUT TIP Vx STAR GROUND FOR VINTAGE UNIT 2 dB PAD SHOWN IN AX STAR GROUND FOR AUXILIARY PANEL NORMAL POSITION NORMALLY CLOSED 174 INCH FEMALE JACK SHIELDED COMPARTMENT ON AUXILIARY PANEL Figure 10 Schematic diagram of triode stage used in CH1 CH4 of modified Altec 1567A CH1 and CH2 have a high impedance grid circuit build out that includes an unbalanced input jack attenuator pot and pad switch bottom of diagram Instead CH3 and CH4 simply use 1 MO resistor R1 retaining the vintage design The dashed lines depict the placement of the two alternative high Z circuits The nominal 50 KQ secondary winding of the Altec type 4722 input transformer needs a load resistor to reflect the proper impedance to the primary circuit this is provided by 1 MQ resistor R1 Figure 10 for CH3 and CH4 exactly like the stock unit or the high Z build out for CH1 and CH2 also representing a 1 MO resistive load see below At 1 KHz voltage step up in the input transformer is about 25 dB when the full primary winding is used The signal at the secondary is applied to the grid of one triode of a 12AX7 either V1 or V
13. SECONDARY NOT SHOWN 120 VAC 1A MAIN IEC ee POWER JACK FUSE Yx Star Ground for a Vintage Unit AX Star Ground for SWITCH Auxiliary Panel AUXILIARY PANEL lt gt VINTAGE UNIT Figure 4 Schematic diagram of the modified Altec 1567A s line AC connections to the power transformer primary windings Secondary circuits are in Figures 5 and 6 The ground link between the vintage unit and auxiliary panel is also shown 4 2 Power Supply Line AC and Power Transformer Primaries Figure 4 shows how the modified Altec 1567A s single 120 VAC power jack feeds the primary windings of the vintage unit s power transformer and the auxiliary panel s toroidal power transformer The power jack fuse holder and main power switch a DPDT toggle wired in parallel making an SPDT with higher reliability share a small enclosure AC box behind the upper right corner of the auxiliary panel This is near the vintage unit s power transformer so only short wires are necessary to 12 reach the latter s primary via a hole punched in the vintage chassis directly over the AC box The toroidal power transformer is mounted directly below the AC box so its leads are also as short as possible its primary windings are hooked in parallel for 120 VAC use only A metal oxide varistor MOV serving as a transient surge suppressor lends some overvoltage protection mostly for the regulators in the solid state power supply but the old vintage power
14. bottom see text 5 3 Channels 1 4 Modeling Balanced Pads Figure 16 models source coupling to CH1 CH4 s balanced inputs with the low Z pad off Model A or on Model B Importantly these simple models ignore reactive components of the source or load impedances therefore please 31 consider their accuracy as limited to the audio mid band around 1 KHz Each diagram shows source impedance Zsource aS two equal resistors one in series with each wire of the balanced line The circled waveform icon is a hypothetical AC generator voltage source with zero impedance so the individual resistors for the balanced Zsource are in series and simply sum together Zsource could be diagrammed as a single resistor without altering the math With the pad turned off Model A in Figure 16 Equation 1 simply says that a channel s input impedance Zn equals Zioap Zoan is the impedance of a channel s transformer balanced input circuit including the 220 Q shunt resistor in the case of CH1 and CH3 and is symbolized by a resistor in these models remember that model accuracy is restricted to the audio mid band The voltage loss in dB across Zgource in this case is given by Equation 3 in which Zsource and Zioap form a voltage divider Note how matched source and load impedances result in a 6 dB voltage loss as mentioned in Section 5 2 ag cues ge oe Pn A RER A i AEE ee ee eee AGES TE PESTS Tg TAT TES EET PES TAG IU E RT PASTE S
15. in the lower left corner of Figure 14 the inch unbalanced input jack occupies one compartment of the auxiliary panel s shielded high Z area see left hand photo in Figure 11 26 This input is AC coupled by C1 to a switchable pad variable attenuator network like the ones used in CH1 and CH2 see Figure 10 and its description in Section 4 7 As in those channels this network interacts with the following triode to set CH5 s input Z to about 840 KQ at 1 KHz for the normal attenuator and pad settings full clockwise and off respectively Also as in CH1 and CH2 stray capacitance restricts the most useful range of pre triode attenuator R2 to its relatively extreme settings and only the counter clockwise region with the 20 dB pad on Normal settings for most applications should be pad off and attenuator fully clockwise see Section 5 13 for more information Via a minimum length of shielded cable passing into the vintage unit the signal at the wiper of R2 is applied directly to the grid of V3A one triode of a 12AX7 Unlike the corresponding triode s hook up in the original unit the mix bus amplifier or summing mixer here V3A uses cathode bias provided by R5 Compared to the original design this should increase the maximum input amplitude this stage can handle Electrolytic capacitor C3 bypasses R5 to prevent degenerative feedback and maximize gain The chosen value of R5 680 Q optimizes the triode operating point while retaining plate
16. its three triode signal path my distortion tests sometimes changed the input amplitude and or fader position as well as the feedback knob see also Section 5 19 However they suggest an effect whose magnitude is reasonably close to the theoretical one in which percent distortion is halved for each 6 dB decrease in gain due to feedback Since the vintage unit was originally designed for a specific fixed feedback level frequency response was probably carefully tweaked for that condition C14 in the original schematic is a good candidate for such a tweak Making feedback variable by simply swapping a variable resistor for a fixed feedback resistor is a somewhat crude technique And the extra physical length added to the feedback loop using shielded cables to from the vintage panel mounted control invites effects of stray capacitance Therefore was not surprised to measure frequency response effects of varying the feedback As the feedback knob is turned clockwise increasing gain a high frequency emphasis around 7 KHz and above is present at low gain decreasing in intensity until just clockwise of the design level mark flattest frequency response is between 1 and 2 o clock beyond that setting high frequencies roll off with increasing gain until at full clockwise response relative to 1 KHz is 1 dB at 6 7 KHz and 3 dB at 16 5 KHz To summarize mid range feedback settings yield the flattest frequency response 5 18 Channel 5 Gain and
17. load resistor R4 s original value 100 KQ this should keep this stage s output Z similar to the original s for driving the tone control channel fader network which follows The original pre tone network recorder output connection see Appendix is omitted in the modification The tone control network itself including original bass and treble pots R8 and R9 is one of the two vintage sections that were left original and not re built completely during the modification the other is the VU meter network It thus retains the original ceramic disc capacitors C4 C8 such capacitors age better than film capacitors used for DC blocked signal coupling and carbon composition resistors R6 R7 and R10 these read within tolerance on an ohmmeter The tone network s output feeds channel fader pot R11 via the original short length of shielded cable At the fader s wiper coupling capacitor C9 is anew SBE Orange Drop 716P series polypropylene film and foil unit as are all coupling capacitors throughout the modified vintage unit except C14 in this channel see below From C9 the post fader signal passes via shielded cable to the grid of triode V3B which is the first stage of CH5 s two stage output driver called a line amplifier in the original Altec 1567A manual see link in Appendix This stage is re built exactly like the original and no attempt was made to correct an observed minus twenty percent difference in measured
18. model shown in Figure 18 One needs to bear this in mind when using extremely high Z sources such as piezoelectric pickups 33 N Q Capacitive Reactance Xy Xin EQ 1 1 2nFCep Where Cap Effective grid to plate GP Capacitance 104 pF F Frequency Magnitude of Input Impedance Z 1 Zin je EQ 2 Re Where Ro Grid Resistor 1 MQ 2 Magnitude of Input Impedance Z 10 100 1K 10K Frequency F Hz Figure 18 Input impedance model for the unbalanced high Z inputs of CH1 CH2 and CH5 with pre triode attenuator full clockwise and high Z pad off Equation 1 shows how the triode s capacitive reactance depends on frequency and Equation 2 places that reactance in parallel with the 1 MQ grid resistor The curve is the solution to Equation 2 Red index lines indicate performance at 1 KHZ With minimum length patch cables recommended the unbalanced inputs are compatible with the outputs of the modified units other channels as well as most external low level ground referenced sources External sources should either share a good common ground with the modified Altec unit or receive their ground reference via their output patch as in an effects pedal or an electric guitar or bass Line level sources on the nominal 10 dBV standard should usually be compatible but if peaks cause unwanted distortion decrease output level at the source if possible avoid the temptation to turn down t
19. mostly obsolete chassis mounted can types This also calls for laying out new terminal strips somewhat differently from the originals Also modification will upgrade the grounding system to a more robust star ground based strategy Not surprisingly the as is unit had poor low frequency response probably due to the elderly signal coupling capacitors My plan included replacing all of these with high quality polypropylene film units While working all of the original tube and transformer sockets were old tired and dirty My preference when rebuilding old gear is that sockets should be replaced with new ones for highest reliability decided to use ones with ceramic insulators for the best long term stable performance Happily all of the potentiometers operated without scratchy spots once exercised a little so none of them required replacement MODIFIED ALTEC 1567A Block Diagram of Audio Signal Paths Pre Triode Attenuator Output CHANNEL 1 Fader INPUTS Pad Impedance Sn nd T A Polarity CHANNEL 1 OUTPUTS Input Transformer Altec 4722 Pre Triode CHANNEL 2 CHANNEL 2 INPUTS Pad Impedance Attenuator OUTPUTS V7 bX po Polarity Input Transformer Altec 4722 CHANNEL 3 CHANNEL3 pad OUTPUTS Balanced Line Driver INPUT C Dp Polarity Input Transformer Balanced Line Driver Altec 4722 Output CHANNEL 4 CHANNEL 4 Pad Impedance Fader
20. nearly constant regardless of the combination of relays activated Shunting relay coils diodes D1 D3 suppress back EMF spikes when relays change state 4 7 Channels 1 4 Triode Stage The triode circuits in CH3 and CH4 are essentially identical to those of a stock Altec 1567A see Appendix with the omission of the wire from the pre fader 19 output to pin 1 of the transformer socket this was a negative feedback connection needed only with the phono equalizer plug in accessory However the grid circuits of CH1 and CH2 are built out in shielded compartments of the auxiliary panel so that each includes a high Z unbalanced input pad and variable attenuator Each alternative grid circuit is included in Figure 10 will describe the common aspects of the triode gain stage first followed by detailing the grid circuit elaboration specific to CH1 and CH2 ALTEC 4722 INPUT TRANSFORMER CHANNELS 3 AND 4 SIMPLY USE A Re 1 M LOAD RESISTOR IN XFORMER 220K SECONDARY TRIODE GRID CIRCUIT 172W AS IN ORIGINAL ALTEC 1567A MF ci OD 1 uF 400V TO TOP OF FADER 172 12AX7 MEASURED DC VOLTS CHANNEL TRIODE CATHODE 174W MF PIN NO i us ce Ux 47uF VOLTAGE MEASUREMENTS DONE R3 ON DMM WITH 1 M OHM INPUT PRIMARY CIRCUIT NOT SHOWN IN THIS DRAWING QO0QQVQ00 R1 1 0M 35V Rg 174W MF UNITS IN SERIES CHANNELS 1 AND 2 ARE BUILT OUT TO 180 1 69K PROVIDE A HI Z INPUT OPTION SWITCHABLE Vk PAD AND
21. omitted to avoid clutter Notice that CH1 solid curve has slightly greater bandwidth than CH2 dashed curve the presence of a 220 Q shunt resistor across the primary of CH1 s input transformer is 36 the only difference between these two channels This may help explain why Altec included a similar resistor in the original design see Sections 4 6 and 5 2 t Respons To Lk ELATIVE 4 46 pk _ a See N b Se e N AA WANA 4 _ t Figure 19 Measured frequency response of CH1 solid curve and CH2 dashed curve relative to response at 1 KHz 0dB under conditions listed in the drawing Sine wave input was from a function generator with 50 O source impedance Balanced channel outputs were terminated with 600 Q in single ended mode and measured with RMS voltmeter in the Hewlett Packard 331A instrument Results for CH3 and CH4 were the same as that of CH1 The channel fader pots present a resistive load to the triode outputs see Sections 5 5 and 5 8 Additionally stray capacitance in the shielded cable connecting the pot s wiper to the output driver board and high Z outputs apparently causes wiper position dependent impedance changes that affect frequency response slightly At lower settings 32 dB to about 12 dB as painted on the vintage panel there is a broad response peak centered on about 12 18 KHz depending on channel tha
22. plate voltage here compared to that published in the original schematic 97 V here versus original 120 V see Appendix This was the largest plate voltage difference versus the original schematic The final vacuum tube stage uses the two triodes of V4 the 6CG7 hooked in parallel to act like a single triode with lower output impedance Note that the original schematic omits the 6CG7 s internal shield at pin 9 although unimportant in this application it was grounded in the original unit as it is in the re build as shown in Figure 14 The original cathode bias resistor R20 value of 470 Q caused V4 to conduct a little too much current in my opinion 10 7 mA versus the 9 33 mA deduced from voltages in the original schematic this made plate load resistor R19 dissipate about 1 7 W closer to its 2W rating than was comfortable R19 is one of the cases where recycled the original component like me wanted it to spend its golden years doing a 27 little less work than perhaps it had in the past Increasing R20 to 604 Q made V4 s plate and cathode voltages more closely match those given in the original schematic and made R19 dissipate a cooler 1 36 W It s unclear whether this bias tweak was prompted by an aging 6CG7 or an outlier I had no other 6CG7 for comparison or some other cause but when it s time to replace V4 please re check the new one s voltages and re bias if necessary The stock vintage unit used a 47 KQ fixed
23. saturation by slightly more than one octave when driven by one of the modified unit s other hybrid channels Therefore Figure 22 suggests that CH2 is a better candidate than CH1 for hearing what transformer saturation offers as a distortion effect Only the low frequencies of an audio program are subject to this effect Perhaps it could be tried on a kick drum or bass guitar to see if this type of distortion does anything aesthetically useful in the context of a mix 5 13 Channels 1 2 and 5 Pre Triode Attenuator High Z Pad Applications and Effect on Bandwidth In CH1 and CH2 the pre triode attenuators and pads allow compensation for the very high amplitudes at the input transformer s secondary when experimenting with transformer saturation In CH2 for example when the full coil primary is driven at 25 dBV in the mid range band the 25 dB voltage step up by the transformer makes the amplitude across the secondary 50 dBV a shocking 316 Vams To safely avoid triode saturation at least 58 dB of attenuation is required before applying this signal to the grid Additional specialized experimental uses for these attenuators may include directly interfacing speaker level outputs from power amps to the unbalanced inputs of CH1 CH2 or CH5 However for all routine applications attenuators should be set full clockwise and the high Z pads turned off input amplitudes causing unwanted triode saturation should be turned down at their source whenever p
24. source of cross talk The second source is addressed by shielding the most sensitive high impedance low input amplitude high gain circuits and maximizing their distance from outputs For pairs of channels with their faders set full clockwise and no pre triode attenuation for CH1 and CH2 evaluated cross talk by comparing amplitudes at the balanced outputs On one channel the balanced input was fed a 1 KHz test signal with amplitude sufficient for maximum un clipped output into a 600 0 load called 0 dB for this test With the other channel s input open its measured output level was expressed in dB relative to that of the active channel could detect cross talk only between pairs of channels that share a twin triode i e CH1 CH2 and CH3 CH4 suggesting electromagnetic coupling within and near tubes as the dominant path The worst pair was CH1 bleeding into CH2 which read 45 dB most of this susceptibility appears related to the relatively high impedance of CH2 s balanced input see Section 5 2 reflected to the grid circuit because simply engaging CH2 s input pad which shunts the transformer primary with a 150 ohm resistor reduced cross talk to 64 dB Cross talk for other 12AX7 sharing pairs tested as follows CH2 into CH1 72 dB CH3 into CH4 78 dB CH4 into CH3 74 dB While not specifically measured cross talk increased with increasing frequency When cascading channels to get distortion effects due to excessive gain b
25. swing between about 15 V and 15 V without clipping Importantly when feeding U2 U1A has slightly more headroom than U2 itself also U2 s clip threshold decreases as output loading increases as described in Section 5 9 Thus by monitoring the output of U1A the clip alert circuit is assured of receiving a clean signal when the balanced channel output has just exceeded its clip threshold threshold adjustment range of the clip alert therefore brackets all conditions in which clipping at the main outputs can occur The clip alert uses inexpensive yet effective TLO72 dual FET input op amps U3 and U4 run at 15 V Setting the circuit s input Z at nearly 1 MQ voltage divider R4 R5 halves the signal s amplitude i e makes a 6 dB pad so it can never exceed U3A s headroom C9 blocks any DC offset of U1A s output which is no greater than a few millivolts Working as a unity gain buffer U3A drives an absolute value circuit built around U3B This is literally a cookbook circuit by Walter G Jung IC Op Amp Cookbook 3 Ed 1986 ISBN 0 672 22453 4 p 245 Fig 5 148 to whom refer you for a detailed description Basically like a full wave rectifier it inverts only the signal s negative voltage swings and doesn t affect the positive ones This lets the following stage threshold detector respond to both positive and negative peaks of the original signal Configured as a voltage comparator U4A serves as the threshold dete
26. the pad and source impedances when a pad is engaged But this is not the difference one hears when switching the pad on because there is already some hidden loss across Zsource with the pad off given by Equation 3 To model the perceived loss when a pad is turned on the solution of Equation 3 must be subtracted from that of Equation 4 as stated in Equation 5 Figure 16 bottom The black curves in Figure 17 represent Equation 5 solved for a range of Zsource values the dashed green horizontal line is the special case of Zsource 134 Q which yields a perceived pad loss of 19 4 dB at all ZLoan values All loss curves converge on 19 4 dB when Zoan is 1258 Q which is the Zoan value at which Zn does not change when a pad is turned on dashed blue lines Summarizing Figures 16 and 17 the effect of CH1 CH4 s low Z balanced pads depends on both source and load impedances If you infer that designed these pads targeting 20 dB loss at ZioapS in the 1 2 KQ range you would be correct lowered the impedance of in CH1 and CH3 by adding the 220 O shunts after installing the pads Also recalling the way each channel s Zioap depends on frequency Figure 15 bandwidth restriction due to pad engagement will occur under many conditions even though again caution you against using my scalar impedance measurements and models for accurate bandwidth prediction Unfortunately no single pad design can give consistent performance when source and loa
27. their Document 600078 Rev 04 PDF available on line Non polarized bi polar electrolytic capacitors C7 and C8 AC couple feedback in the driver s servo loops to minimize DC offset at its outputs The THAT1646 emulates transformer balanced outputs in many respects will describe its characteristics more completely in Section 5 9 Note that U2 s non inverted output feeds pin 3 of the male XLR output jack and the inverted output feeds pin 2 You may initially think this conflicts with the conventional standard of using XLR pin 2 for or hot however remember that the channel s triode stage inverts the signal so this crisscrossed output connection makes the balanced output jacks for CH1 CH4 match the polarity of their respective balanced inputs and the high Z inputs of CH1 and CH2 However the high Z unbalanced output remains inverted with respect to standard XLR polarity and the high Z inputs of CH1 and CH2 Of course setting a polarity switch to invert affects only balanced input signals not the high Z inputs 4 9 Channels 1 4 Clip Alert Indicator Circuit The clip alert indicator Figure 12 bottom half works by detecting whether a pre set instantaneous peak amplitude is exceeded at the balanced line driver IC s input Powered at 18 V the output of U1A and individual output pins of U2 i e measured in single ended mode not differential mode double this for differential mode can
28. transformer may also benefit The MOV is located in the vintage unit for ease of replacement should a large surge cause it to short out or explode The fuse is a 1 A slow blow type Once warmed up the overall modified unit dissipates about 35 watts which is 0 29 A at 120 Vans the vintage unit accounts for about 0 16 A and the auxiliary panel 0 13 A But the inrush current upon power up briefly exceeds 1 A which a standard fast acting 1 A fuse as used in original vintage unit doesn t always withstand This surge is caused by filter capacitors charging and also the considerable magnetic inertia of the big 160 V A toroidal transformer Also shown in Figure 4 is the link between the star grounds for the vintage unit and the auxiliary panel It s included in this schematic because the low resistance heavy braided conductor making this connection passes through the same hole in the vintage chassis as the line AC connection The star ground terminal for the vintage unit is a short heavy gauge solid copper wire anchored at its ends with two of the bolts mounting the vintage power transformer to the chassis see Figure 9 for photo inside vintage unit also see Section 4 4 The star ground terminal for the auxiliary panel is a piece of thick brass foil mounted on one edge of that panel s power supply board see Figure 7 for photo of this board The original schematic of the vintage power transformer Appendix shows an electrostatic shield b
29. 106 7 dBV there this reasonably assumes that transformer noise is insignificant compared to the triode s Triode stage noise decreases as a channel fader is 50 turned counter clockwise while output driver noise does not change EIN therefore increases with fader attenuation as the relative contribution of driver noise increases see table below Of course EIN is independent of input signal amplitude but the signal to noise ratio SNR is not SNR at a given output amplitude is easily found using dB units subtract the fader attenuation and the channel gain from the output amplitude to get the input signal amplitude then subtract the EIN corresponding to that fader setting to find SNR Along with EIN at different fader settings the following table shows SNR when the balanced input signal is 45 dBV this input signal yields the maximum un clipped output amplitude 20 5 dBV see section 5 9 and Figure 20 into 600 Q in single ended mode when the fader is full clockwise Fader Attenuation dB EIN dBV SNR dB 0 full CW 131 7 86 7 10 131 6 86 6 20 131 1 86 1 30 128 0 83 0 40 120 1 75 1 1 At balanced inputs channel gain 65 5 dB Add 25 dBV for EIN at CH1 CH2 unbalanced inputs 2 With input amplitude that yields 20 5 dBV S E output into 600 Q at OdB fader attenuation At any given fader setting SNR decreases without limit as signal amplitude decreases However improving SNR by increasing the amplit
30. 2 Plate load resistor R2 and cathode bias resistor R3 have the same values as their counterparts in the original design This gives a triode operating point close to the original design s measured DC potentials on the plate and cathode listed in Figure 10 are within about 10 percent of those given in the original schematic 20 As in the original C2 shunts the cathode bias resistor to eliminate negative degenerative feedback at audio frequencies maximizing gain at the cost of some bandwidth loss As in the original design C1 AC couples the output of the triode stage to the top clockwise most terminal of the channel s output fader pot see Figure 12 R1 High quality SBE formerly Sprague 716P series Orange Drop polypropylene film capacitors are used for signal coupling throughout the modified vintage unit with one exception noted in the CH5 description Section 4 10 Voltage gain in this triode stage loaded by a 250 KQ fader pot is 35 dB Accounting for transformer step up 25 dB using the full primary coil as noted above transformer plus triode gain is therefore about 60 dB see Section 5 5 for details of CH1 CH4 gain structure Note that common cathode gain stages such as this invert the input signal Instead of R1 to load the high Z link between the transformer and triode CH1 and CH2 are each provided a normally closed unbalanced input jack pad and variable attenuator as diagrammed in the bottom portion of Fi
31. 2 Distortion threshold and type versus frequency with CH4 s balanced output driving the balanced input of CH1 circular data points solid curve or CH2 square data points dashed curve CH1 and CH2 settings are listed in drawing At each sine waveform frequency tested distortion thresholds were judged using an oscilloscope on the output of CH1 or CH2 RMS voltage at CH4 s output was then measured The type of distortion was also noted When due to transformer saturation distortion appears at the waveform s zero crossing phases Distinctly clipping at the output driver manifests at peak phases both positive and negative In any case Figure 22 shows that transformer saturation gives way to driver clipping at 250 Hz for CH1 and 550 Hz for CH2 This is consistent with the differential mode output driver clipping characteristic shown in Figure 20 and CH1 and CH2 s input impedance versus frequency shown in Figure 15 The 220 0 primary shunt resistor in CH1 lowers the input impedance by providing a pathway for current to bypass the transformer s primary The driver cannot source 45 sufficient current to saturate the transformer at frequencies greater than 250 Hz and still heat up the shunt resistor The only difference with CH2 is that CH2 lacks a shunt resistor This buys 8 dB more mid range headroom note in Figure 20 that the driver is current limited when clipping into CH1 at 1 KHz It also widens the band susceptible to transformer
32. 4 stage based on this tube s published characteristics used General Electric spec sheet ET T941B dated November 1956 As shown in Figure 14 and described in Section 4 10 both triodes of the 6CG7 are wired in parallel to behave as a single triode effectively halving the impedance compared to a single triode Since the published characteristics are for individual triodes drew a load line on the plate characteristic chart representing current through one triode as limited by a 30 KO plate load resistor double the parallel configuration s 15 KO value Also assuming a doubled cathode bias resistor value the predicted operating point correlated well with the expected voltages at B plate and cathode for the parallel combination The analysis yielded a dynamic plate resistance rp of 8 19 KQ for a single triode rp for the parallel combination should be one half this value or 4 095 KQ In turn placing this in parallel with the 15 KQ plate load resistor predicts the V4 stage s open loop i e no feedback output impedance 3 22 KQ However negative feedback effectively reduces this output impedance because it tries to keep the output voltage constant should load conditions change Imagine suddenly decreasing the load impedance The output voltage would tend to drop but this decreases the negative feedback signal making the gain increase to compensate Theoretically each 6 dB decrease in gain due to voltage feedback halves th
33. 40 30 20 10 0 Fader Attenuation dB Figure 26 Left chart Signal to noise ratio SNR at CH1 CH4 s unbalanced high impedance outputs as a function of fader setting which makes little difference in this case see text Assumes very high load impedance 2 10 MQ say on these outputs Each curve corresponds to conditions giving the labeled RMS output amplitude at full clockwise fader 0 dB attenuation add fader attenuation value for actual output amplitude Red shading indicates conditions causing clipping by diode clamps that protect the solid state line driver circuits see Sections 4 8 and 5 8 Right chart Percent harmonic distortion versus triode stage output same as shown in Figure 25 see that figure s legend Alignment to the left chart places full clockwise fader output amplitudes in register with the right chart s triode stage output scale as indicated by arrows 5 15 Channel 5 Applications and Input Characteristics Since CH5 is based on the original master channel one obvious application is using it to complete a vintage Altec 1567A signal path simply patch the high Z output of one of the other channels into CH5 s input With some adaptor assembly effort one might even reconstitute the original four into one mixer function branch the 1 4 inch plug that feeds CH5 s input to four 1 4 inch female jacks via 330K resistors see R18 R21 in the original schematic shown in Appendix Use a small shielded break out box and m
34. EE EERSTE A ST a LOSS Cuetw FAA TATS J gt 5 tH D n A 20 50 0D 200 500 IK 2k Figure 17 Performance of balanced input pads used in CH1 CH4 of modified Altec 1567A as predicted from models shown in Figure 16 The horizontal axis Z o p is the input impedance when pad is switched off The input impedance with a pad engaged is shown by the red curve plotted against red vertical scale at right which is the solution to Equation 2 in Figure 16 The perceived loss when a pad is engaged is given by the black curves using black scale at left which emerge from Equation 5 in Figure 16 loss depends on Zoan except when Zgource 134 Q dashed green curve Engaging a pad changes the input impedance except when Zoan is 1258 Q dashed blue index lines 32 The pad shown in Model B Figure 16 has the same component numbers R1 R3 and values as used in CH1 CH4 of the modified unit see Figure 8 Equation 2 says that a channel s balanced input impedance Zn when the pad is switched on is the sum of series resistors R1 and R3 plus the parallel combination of ZLoan and R2 This equation generates the red curve in Figure 17 which shows that the pad converts a two decade Zioap range 20 Q to 2 KQ into less than a ten percent range of Zn Engaging the pad increases Zn when Zoan is less than 1258 Q emphasized by dashed blue index lines in Figure 17 Equation 4 in Figure 16 expresses the combined loss across
35. ER CO MADE IN USA 9 Figure 24 Effect of CH1 s pre triode attenuator setting on bandwidth with high Z pad off circular data points solid curve or on Square data points dashed curve This is derived from Figure 23 s dataset here each curve shows the frequency at which response is 1 0 dB relative to the response at 1 KHz as a function of attenuator setting Since my tests used CH1 s balanced input the source impedance seen by the pre triode attenuator network was probably on the order of 50 KQ the nominal secondary impedance of the Altec 4722 transformer Expect additional high frequency loss in the attenuator s clockwise range with pad off when source impedances greater than 50 KQ are inserted at CH1 CH2 or CH5 s unbalanced inputs the severity of this and the need for short patch cables will increase as the source Z increases At least this is suggested by my informal analysis which follows Here is how stray capacitance may explain the observed results you may skip this paragraph if you like your analyses air tight The most problematic stray capacitance apparently acts between the pot s wiper and ground like hooking a capacitor in parallel with the pot s bottom counter clockwise resistance This provides a low Z path to ground bypass for high frequencies With the high Z pad switched off the relative effect of this capacitor is greatest 48 when the resistances above including the sour
36. MS ALL FIXED RESISTORS ARE 174 W CARBON FILM EXCEPT CG MF 174 W METAL FILM RESISTOR 1 TOLERANCE BP BI POLAR ELECTROLYTIC CAPACITOR PANASONIC SU 35V ALL 1uF CAPACITORS ARE STACKED FILM TYPE S V ONLY ONE CHANNEL SHOWN EACH BOARD OF 2 HAS TWO CHANNELS RAIL BYPASS CAPS Ci C2 C1 AND C11 SERVE ONE WHOLE BOARD DC SUPPLIES CONTINUING TO ADJACENT CHANNEL ARE NOT SHOWN Q D JF C11 18uF 35V 15 Figure 12 Schematic diagram of CH1 CH4 s solid state line driver and clip alert indicator circuits channel fader and unbalanced output are also shown These circuits are identical for CH1 CH4 so only one is diagrammed here At the upper left of Figure 12 is pot R1 a channel fader in the vintage unit for either CH1 CH2 CH3 or CH4 the signal is taken directly from its wiper via a shielded cable leading to an output driver board Importantly the fader pots are grounded to the driver board s ground plane not to the vintage unit s ground to preclude an internal ground loop At the connection to the driver board a second shielded cable branches the signal to a 14 inch female high Z unbalanced output jack the other branch feeds the balanced output driver The THAT 1646 differential output driver chip U2 has a relatively low input impedance of 5 KO so it needs to be fed by an op amp stage This is the role of U1A which is one half of a Burr Brown now part of Texas Instruments OPA2604 dual op amp th
37. OUTPUT INPUT 5 o INPUT 1 INPUT 2 INPUT 3 30 r Lon i RT IM i q cic RIS RI bo i800 220k n 1 Ri2 1800 R8 l N E a v 250K INPUT 4 l R2I 3 330K ere RIG 220K cs T A isv 1 4 vi S 5 PHONO EQUALIZER LEGEND ASSEMBLY L OHMS 4722 MICROPHONE TRANSFORMER K 1000 N M 1 000 000 ALL CAPACITANCE VALUES IN MFD UNLESS OTHERWISE INDICATED R40 R4I R42 IOK 4700n 22K Iw ca c9 A ci9 8 ci9c 9 40 20 40 C17 B 9 cis K AS ve 15095 LINE R32 47K V4 6CG7 165v 25n va PLI L2 5 nas iy RS2 RS3 4 4 C20A c208 v3 000 1000 s 5 5v i5v ALTEC LANSING 1567 A AMPLIFIER 6 3VAC 3 ae FIRST MADE FOR IS67A AMPLIFIER TOUERANCES EXCEPT as NOTED peee e TRA LINE STRAP ison l 4 3 6 soon 3 4 OUTPUT CONNECTIONS mikey 15095 TRANSFORMER Pasun Oren WITHOUT TRANSFORMER 7V 60w 20 WATTS FRAGT iee bac om more wine o to in 201 OVEN i a com anoutan a sa S674 AMPLIFIER SCHEMATIC
38. Rebuilt and Modified Altec 1567A A Technical Report By Clark Huckaby http www clarkhuckaby com November 2012 Revision July 2013 2012 2013 Clark Huckaby All rights reserved 1 Introducti 2 The Origi 2 1 2 2 Table of Contents on nal Altec 1567A General Description of Original Design Pre Modification Condition of This Project s Particular Altec 1567A 3 Overview of Modified Unit 4 Schemati 4 10 5 Performa 5 14 5 15 5 16 5 17 5 18 5 19 5 20 5 21 General Architecture Channels 1 Through 4 Special High Z Circuitry in Channels 1 and 2 Channel 5 A Word About Solid State Outputs Clip Alert LEDs for CH1 Through CH4 Input Output Polarity Mechanical Layout and Power Supply cs and Circuit Descriptions of Modified Unit Note About Schematic Diagrams Power Supply Line AC and Power Transformer Primaries Power Supply Vintage Unit Grounding Rule for the Modified Vintage Unit Power Supply Auxiliary Panel Channels 1 4 Balanced Input Circuits Channels 1 4 Triode Stage Channels 1 4 Balanced Line Drivers Channels 1 4 Clip Alert Indicator Circuit Channel 5 nce and Applications of Modified Unit Standard Amplitude Units Channels 1 4 Impedance of Balanced Inputs Channels 1 4 Modeling Balanced Pads Channels 1 2 and 5 Impedance and Compatibility of Unbalanced Inputs Channels 1 4 Gain Channels 1 4 Cross Talk Channels
39. S IN OHMS D2 MF 174W METAL FILM RESISTOR i 1 TOLERANCE x STAR GROUND FOR VINTAGE UNIT Ax STAR GROUND FOR AUXILIARY PANEL 5 ETTI E 1N4004 4 WIRE RIBBON r a o TAx RS IMPED R6 37 750 124 ANCE Figure 8 Schematic diagram of balanced input circuits for CH1 CH4 of the modified Altec 1567A Except for R7 as noted the balanced signal paths for these channels are identical so only one is diagrammed here top Depending on channel relay control circuits use either 12 VDC polarity and each is shown middle and bottom Working left to right across the top of Figure 8 FII focus first on the signal path before describing the relay control circuits Female XLR jacks located on the auxiliary panel are wired to the relay boards using shielded twisted pair cable In its normal off state relay RY1 simply passes the balanced signal but its activation engages balanced U pad resistor network R1 R3 Nominal pad loss is 20 dB however this and how input Z changes depends on the channel as 18 explained in Sections 5 2 and 5 3 RY2 inverts the balanced signal when activated normally the balanced output jacks for CH1 CH4 have the same polarity as their input jacks With its contacts wired in parallel as SPDT RY3 normally connects the balanced signal to the input transformer s full primary coil pins 4 and 6 RY3 activation switches to the tap pin 5 to redu
40. X7As may be slightly different or my prediction may have some error because used classic graphical load line techniques tested my favorite vintage RCA 12AX7 specimen mid 1960 s black plates in a breadboard version of the circuit and obtained 35 4 dB gain It s a good bet that most high quality 12AX7s will provide such high gain Each channel s THAT1646 INPUT balanced line driver IC adds UNBALANCED BALANCED 5 5 dB gain when driving 600 CHE and CH2ony Q loads and nearly 6 0 dB into 10 KQ or more whether operating in differential or single ended output mode This is a result of the way these chips balance the output to emulate a transformer and is described further in Section 5 9 The total gain available in CH1 CH4 is summarized in the table at right It is based on QO _ Wo O S Z C 78 m e ze A H D O BALANCED Into 600 Q Load 35 the individual stage gains just discussed and confirmed by direct measurement for the most common configurations such as full coil balanced input to low Z balanced output 65 5 dB 0 75 dB on all channels 5 6 Channels 1 4 Cross Talk Cross talk between channels results from 1 coupling through their common power supply and 2 electromagnetic coupling through space As noted in Sections 4 3 and 4 4 star grounding and an individual B decoupling network for each channel improves the original Altec 1567A design helping to limit the first
41. a 12AX7 and is the first stage of the master channel It drives a pre tone control recorder output and via the tone control network treble and bass knobs the master volume control The master volume control is followed in turn by the remaining 12AX7 triode then a 6CG7 wired as a single triode serving as an output driver Negative feedback between the latter two triode stages reduces the master channel s gain but helps reduce distortion flattens frequency response and lowers output impedance The two main outputs are 1 an unbalanced output that does not depend on an output transformer being inserted and 2 a low Z balanced output when a model 15095 line output transformer is plugged in The original user manual see link in Appendix mentions that these two outputs can be used simultaneously The 15095 s two secondary windings can be hooked in series or parallel to drive 600 or 150 Q balanced lines respectively The VU meter is preceded by a 5 position rotary switch offering four sensitivities plus one off setting Overall the Altec 1567A uses four twin triode vacuum tubes V1 and V2 are the 12AX7s for input channels 1 4 V3 is the master channel s 12AX7 and V4 is the master channel s 6CG7 All triodes are configured as common cathode voltage amplifiers Operating as high gain low input signal preamplifiers the 12AX7s have shields and their heaters are powered with DC to minimize hum The 6CG7 is on a separate
42. a sine waveform When flux maxima are clipped due to core saturation the transformer s output voltage waveform is upset mostly in the zero crossing excursions not clipped at their maxima similar to cross over distortion in a poorly biased class B power amp Unlike triode saturation the distortion is symmetrical and adds odd order harmonics A rather extreme example is shown in Figure 21 bottom trace in left hand scope display Core saturation is covered further in the next section and Section 5 13 deals with performance of the necessary pre triode attenuators A significant aspect of Figure 21 is that two very different types of distortion are obtained using the same two channel cascade the only differences are the fader and pre triode attenuator settings Note CH2 s pre triode attenuator is set near full counter clockwise for transformer saturation Incidentally the two photos of the panel happen to be from slightly different angles so they form an ersatz stereo pair Try viewing them with crossed eyes for a 3 D like effect 5 12 Channels 1 and 2 Input Transformer Saturation Threshold A context for experimenting with transformer saturation to distort audio programs is to understand the susceptible frequency range and the amplitudes needed With their pre triode attenuators in the extreme counter clockwise range to compensate for excessive input levels fed CH1 or CH2 s transformers from the balanced output of CH4 The CH4
43. ably comes from the non ideal characteristics of these real transformers such as reactance from the transformer s self inductance Impedance drop off with increasing treble frequency is probably largely due to the triode stage s capacitive reactance Miller Effect see Section 5 4 reflected to the transformer s primary e Nan Cie 1925 0 t H a ONN Se ee oe a Bel JSE 100 EPAR a fea in Figure 15 Measured impedances magnitude only of CH1 CH4 s transformer balanced low Z inputs on the normal impedance setting using full primary coil versus frequency Data point symbols show the measurements and they are linked by smooth curves to represent each channel s characteristic CH1 and CH3 were indistinguishable By comparison switching to low Z handle up transformer primary tap used while keeping the pad off normal drops measured input impedances at each given frequency by about 4 fold curves not shown in Figure 15 to avoid clutter Measured impedance at 1 KHz is then 48 Q for CH1 and CH3 470 Q for CH2 and 98 Q for CH4 This 4 fold impedance drop is consistent with halving the number of turns in the primary winding confirming the tap is a true center tap i e with equal turns on either side In an ideal transformer the impedance in the secondary circuit 30 is reflected on the primary according to the square of t
44. ach to the bandwidth minimum at about 3 00 three o clock as the attenuator is turned counter clockwise Another way to show the data is Figure 24 where the bandwidth is directly plotted against pre triode attenuator setting Here bandwidth is expressed as the frequency at which response compared to that at 1 KHz which defines 0 dB crosses below 1 dB As the attenuator is turned clockwise left to right along horizontal axis the bandwidth difference due to high Z pad setting is small until 3 00 beyond which bandwidth continues to degrade if the pad is on but 47 starts to recover if the pad is off Fortunately for transformer saturation experiments see Section 5 12 sufficient attenuation requires settings in the counter clockwise end of the range although the high frequency emphasis there may be unwanted and require EQ For virtually all other applications and routine use these attenuators should be left fully clockwise and the high Z pads switched off gt mn E x wee EER 4 oe jee FREQUENCY AT 1048 MA ie MANES _ ae ies ce aS io _ Hn i 2 a4 OGARITHMIC 3 CYCLES X 60 DIVISIONS IL amp ESS
45. ack confers gain on one output leg when the other is grounded as discussed above By measuring output amplitudes across different load resistors while carefully observing the waveforms on an oscilloscope gathered the data shown in Figure 20 this clearly shows the amplitude difference between the differential and single ended clip thresholds black and red curves respectively at least for load resistances of 300 Q or more As shown in Figure 20 load resistance has a relatively small effect on clip threshold when it is greater than about 300 Q in single ended mode or 600 Q in differential mode In this load range the output voltage swing is limited by the driver s power supply voltage the slight threshold decrease as load resistance decreases is consistent with voltage drop across the driver s 57 O impedance As load resistance decreases beyond 300 Q single ended or 600 Q differential clip thresholds decrease more steeply In this range the slope suggests clipping is caused by an instantaneous current limit of about 57 mA approaching the chip s rated short circuit current of 70 mA The two highlighted data points in Figure 20 represent line driver clipping into the balanced inputs of CH1 and CH2 at 1 KHz rather than into dummy load resistors These loads are the channels respective input impedances at 1 KHz from Figure 15 Relevant when experimenting 39 with input transformer saturation one of CH1 CH4 s balanced output
46. ain see Section 5 9 Purist tube audio philosophers may reflexively consider this an anathema so let me digress briefly to defend hybrid architecture in this case No practical recording studio in the 21 century has an all tube or even all analog overall signal path Tube stages are used for let s face it effect pleasing non linearity At their best solid state stages are low noise and very linear in short they are neutral They are also inexpensive to implement Since vintage non linear stages need at some point to be interfaced with modern solid state digital gear anyway why not use sonically neutral solid state stages within modified vintage gear if this adds versatility and connectivity There are few non emotional reasons why not However attention is required to the abrupt limit of the linear range of solid state output buffers i e hard clipping a decidedly non vintage form of distortion This modified Altec 1567A s clip alert indicator LEDs address this as noted next 3 6 Clip Alert LEDs for CH1 Through CH4 When saturated the distortion mode of channel 1 4 s output drivers is hard clipping The clip alert feature is designed to help users tell if peak transients are exceeding clip thresholds without monitoring with an oscilloscope Trigger thresholds for the red alert LEDs are adjustable using 15 turn trim pots accessed through the auxiliary panel As presently adjusted the alert LEDs ligh
47. amp interaction involves a complex interplay of reactive elements when transformer coupled tube stages are used This adds up to tonal character which hopefully is often good Some of the sought after vintage tone may well depend on maximum power coupling and impedance matching By taking advantage of an unusual feature of the stock Altec 1567A s microphone input circuits and one irregular input transformer opted to give the modified version some diversity of balanced input impedances among channels 1 4 Users can then experiment and decide for themselves what sounds best in different cases Notice in the original schematic Appendix that each channel s input transformer primary was shunted with a 180 0 resistor An uncommon strategy for tube preamps this shunt accounts for most of the nominal 150 O impedance of the mic inputs with the normal input Z connection which uses the full primary winding omitted this resistor in CH2 and CH4 but included a 220 O resister in CH1 and CH3 R7 in Figure 8 Why use 220 Q rather than the very similar 180 Q A lack of will to go the distance perhaps As mentioned in Section 2 2 the model 4722 transformer originally in CH3 tested differently than the others While its DC resistance readings frequency response and voltage step up ratio matches the others its input impedance is inherently low In impedance tests it acts like it has an internal shunt resistor At any rate moved th
48. ase the generator s output amplitude until clipping is seen on the oscilloscope display then back it off just enough so that no clipping is seen The output buffer is now operating at its clip threshold Using a fine tipped slotted screwdriver turn that channel s clip alert trim pot clockwise if the nearby red LED is already on or counter clockwise if it is off continue turning until the LED changes state Fine adjust the trim pot so that the LED turns on when the output s clip threshold is just exceeded as judged using the scope 5 11 Channels 1 4 Types of Distortion There are at least five types of harmonic distortion distortion affecting a waveform s harmonic composition offered by CH1 CH4 While it may seem odd to discuss distortion as a performance feature remember that one goal of this modified Altec 1567A is to permit controlled amounts of distortion for musical effect All but the first of these could be useful 1 Hard clipping by the solid state output buffers or their input protection diode clamps discussed above 2 Routine even order harmonic distortion by the triode stages 3 Soft clipping by saturated triode stages 4 Routine low level distortion from 41 magnetic hysteresis in the input transformers 5 Distortion due to input transformer core saturation will discuss the latter four in turn Even order harmonic distortion accounts for much of the sought after warmth that good tube gea
49. ating loads 4 10 Channel 5 For CH5 the old Altec 1567A master channel was isolated by substituting an unbalanced high Z input network for the original mix bus among other modifications The modified channel is diagrammed in Figure 14 Comparison to the original schematic see Appendix may be helpful as the following description emphasizes the modifications 283 3034 z HIGH IMPEDANCE c12 22uF o a ae OUTPUT 400V TIP oD vx R1 RIB SLEEVE cia 3 pF 174 INCH FEMALE Sav ha a R21 18K VW R24 R23 Ree 18K 12K 7 5K y DC VOLTAGE MEASUREMENTS MADE ON DMM WITH 1 M OHM INPUT OD SPRAGUE ORANGE DROP POLYPROPYLENE FILM CAPACITOR Yk STAR GROUND N FOR VINTAGE UNIT Ax STAR GROUND FOR AUXILIARY PANEL M e N SHIELDED COMPARTMENT ON AUXILIARY PANEL PRE TRIODE R1 ATTENUATOR ALTEC 15895 LINE OUTPUT TRANSFORMER SW2 SHOWN IN NOMINAL 6 OHM POSITION OPPOSITE POSITION NOMINAL 150 OHM LOW 2 BALANCED HIGH IMPEDANCE UNBALANCED INPUT TIP OUTPUT o 47uF SWI SLEEVE 40 Y 2 dB PAD SHOWN IN 174 INCH 0 NORMAL POSITION FEMALE JACK 00Q00Q0 OUTPUT IMPEDANCE AR UE y y ALL RESISTOR VALUES IN OHMS FIXED RESISTORS 172W CABON COMPOSITION OR CARBON FILM TYPES UNLESS NOTED OTHERWISE MF METAL FILM RESISTOR Figure 14 Schematic diagram of CH5 in modified Altec 1567A As shown
50. ce impedance in series with the pot and below the wiper s position are equal This is the wiper position giving the attenuator network its greatest output impedance the parallel combination of resistances above and below the wiper The relative effect of high frequency bypass diminishes toward the extreme pot settings where impedance is lower This is a log taper 1 MQ pot explaining why maximum impedance and minimum bandwidth is offset clockwise to 3 00 Applied to the top of the pot the source impedance about 50 KQ in this case also helps shift the Z max clockwise When the high Z pad is switched on output impedance of the attenuator network continuously increases as the pot is turned clockwise relative high frequency bypass by the stray capacitance is maximum at full clockwise in that case A second smaller stray capacitance might explain the high frequency emphasis in the counter clockwise range this capacitance would be between the input and output of the entire attenuator network This would let high frequencies escape attenuation and its relative contribution would be inversely related to the amount of attenuation 5 14 Channels 1 4 Noise Subjectively these channels are quiet for tube gain blocks and compare favorably to similar equipment I ve listened to Objectively my noise measurements require a few assumptions which will make conservatively For each channel with fader set either full counter cloc
51. ce input Z Relay Boards Star Ground for eee CH1 and CH2 CH3 and CH4 Vintage Unit Figure 9 Photograph of interior of vintage unit its front panel swung open of the modified Altec 1567A Relay boards upper left place relays near transformer sockets Up to this point in the description the balanced input circuits of CH1 CH4 are identical Now comes the place where they differ the input transformers full primary winding is shunted by 220 0 resistor R7 in CH1 and CH3 but not in CH2 and CH4 Note that channels containing this shunt are comparable to the original Altec1567A design see Appendix which places a 180 Q primary shunt resistor across each input transformer Availability of un shunted inputs increases channel diversity as explained further in Section 5 2 Each relay is linked to a control switch mounted on the auxiliary panel For each of CH1 CH4 these are SW1 SW3 DPDT mini toggles wired SPDT respectively controlling RY1 RY3 Normal switch handle positions are down corresponding to inactive relay coils The relays operate at 12 VDC to balance the load on the auxiliary panel s bi polar power supply relay control for CH1 and CH2 uses 12 V middle region of Figure 8 while that of CH3 and CH4 uses 12 V bottom region of Figure 8 Normal settings of SW1 SW3 engage 750 O dummy load resistors R4 R6 respectively This resistance is equivalent to a relay coil so load on the 12 V supplies remains
52. ctor It compares the absolute value signal at its non inverting input to a reference voltage at its inverting input The reference voltage comes from the wiper of 15 turn trim pot R11 set up as a variable voltage divider series resistor R12 scales R11 to a maximum useful adjustment range Accessible through the front of the auxiliary panel next to the associated clip alert LED this trim pot sets the amplitude at which the LED activates Clockwise rotation increases the threshold 25 Comparators lack feedback so full open loop gain holds U4A s output at its maximum negative level until a super threshold peak is detected and then the output swings positive In series with U4A s output D5 passes only positive voltages on to LED driver stage U4B also wired as a comparator but with a fixed reference voltage determined by R15 and R16 The input network of U4B includes C16 which charges rapidly through U4A s low output impedance when a peak is detected Afterwards C16 discharges relatively slowly mostly through high value resistor R13 since U4B s input impedance is very high This extends the duration of transient peaks enough to make their resulting LED flashes visible Setting CH1 CH4 s clip alert thresholds requires monitoring the outputs with an oscilloscope will describe a calibration procedure in Section 5 10 Before shipping the modified Altec 1567A adjusted all channels to indicate clipping of balanced outputs into 600 Q flo
53. d impedances vary the latter depending on channel low Z switch setting and frequency Yet these channels have high gain see Section 5 5 and there will be cases when pads are needed such as using efficient mics on loud sources advise users to be aware of the imperfect pads and keep a keen ear on their sonic results so that the compromises necessary for pad design do not sneak into the artistic product If a high output mic has its own built in pad try that pad first 5 4 Channels 1 2 and 5 Impedance and Compatibility of Unbalanced Inputs In each channel equipped with a high Z unbalanced input jack the signal is applied to the top clockwise most terminal of a 1 MQ pot via a coupling capacitor in the case of CH5 The wiper of this pre triode attenuator pot hooks directly to the grid of a 12AX7 triode see Figures 10 and 14 The resistive component of the unbalanced input s impedance remains near 1 MQ regardless of the attenuator and high Z pad settings However inter electrode capacitances in the triode including the gain dependent Miller Effect reduce input impedance in a frequency dependent manner total capacitance is about 104 pF at the gain used When the pre triode attenuator and pad are in their normal positions full clockwise and off respectively see Section 5 13 this drops the magnitude of the unbalanced input impedance from near 1 MQ at 20 Hz to 840 KQ at 1 KHz to about 80 KQ at 20 KHz according to the
54. dB at 102 Hz 1 dB at 227 Hz a gradual barely significant peak reaching 0 25 dB centered at 10 6 KHz 1 dB at 20 8 KHz With the balanced output still under load results for the unbalanced output were the same except the gradual peak reached 4 dB at 11 7 KHz and high frequency response was extended crossing 1 dB at 24 7 KHz Note that the low frequency response measured poorer than expected did not have time to trace the stage s responsible for this before shipping the modified unit None of the new coupling capacitors should be to blame Nor did formally investigate the tone controls effect on frequency response the scope test with a square wave is a quick and dirty method Probably a gentle clockwise twist of the Bass knob will help compensate for the measured low frequency loss perhaps use its 12 o clock position instead of the marked 11 ish position As always one s ears must be the final arbiter of what sounds best 5 19 Channel 5 Noise and Distortion Compared to CH1 CH4 CH5 is subjectively and objectively noisier Since all five channels have a very similar triode gain stage at the front end the main difference is due to the output driver stages which are tube based in CH5 V3A and V4 and also contribute more voltage gain than the solid state drivers of CH1 CH4 Additionally any signal loss in the tone control network which did not attempt to measure effectively degrades the channel s signal no
55. ded by 250 KQ This math yields the following table Fader Attenuation dB Unbalanced Output Z Q O full CW 50K 4 1 78K maximum Zour 8 68K 14 42K 20 23K 28 9 7K 34 4 9K 40 2 5K did not determine how accurately the faders are calibrated on the vintage panel but 40dB bottom row of table is marked close to full counter clockwise note that Altec omitted the or minus signs in labeling the faders In any case most fader settings require that the unbalanced outputs feed fairly high impedance inputs to avoid significant loss For example a standard unbalanced 10 KOQ line level input would cause an additional 6 dB loss at about the 28 fader setting and a 15 6 dB loss at full clockwise On the other hand patching these outputs to the modified unit s own high Z inputs provided on CH1 CH2 and CH5 will cause little loss as will patches to instrument level inputs on guitar amps pedals et cetera Strictly speaking CH1 CH4 can t be considered all tube and transformer signal paths when using just the high Z unbalanced outputs the diode clamps D1 and D2 in Figure 12 protecting the solid state output stages see Section 4 8 also affect the high Z outputs at high amplitudes Hard clipping limits the output swing to 18V that s 12 7 Vams or 22 1 dBV also see Figure 26 This level is so high that practical situations where clipping occurs should be most infrequent If necessa
56. der load measured CH5 s balanced output when the meter read 0 VU at each range setting and the results are in the following table observing the vintage tradition see Section 5 1 output levels are expressed in dBm 60 VU Range Multiplier Setting Output Into 600 Q at 0 VU Indication dBm 0 1 0 4 3 0 8 7 2 12 11 2 1 Test signal constant non dynamic 1 0 KHz sine waveform Balanced output Z setting nominal 600 Q secondary windings connected in series This data show that the range multiplier switch s 4 dB steps are accurate to within 0 2 dB or about 2 3 in terms of Vams which is better than expected And meter linearity was fairly good as 6 VU indications corresponded with outputs averaging 6 37 dB of those giving 0 VU readings data not shown The absolute output amplitudes may suggest that a 0 VU meter reading was originally calibrated to indicate a 0 dBm balanced output on the range 0 setting 4 dBm on the 4 setting et cetera If the modified unit s type 15095 transformer is compromised see Section 5 20 perhaps that accounts for 0 8 to 1 0 dB lower outputs than expected if the expectation is 0 VU 0 dBm for the balanced output terminated at 600 Q The meter bridges the transformer s primary not its secondary as discussed next Regarding how the VU meter monitors the output of the 6GC7 circuit instead of the line transformer s secondary the original Altec 1567A manual see App
57. e alert to the likelinood of feedback Cross talk can close a positive feedback loop by coupling a portion of the cascade s output back to the input causing oscillation when phases correlate properly Thus when adding up enough gain to cause severe distortion always start with your monitors turned down very low until you are sure that the set up is stable Ear splitting high frequency oscillation cannot be avoided if you are going to explore all of the distortion possibilities this experimental equipment offers so monitor very softly any time you attempt to increase distortion in cascaded channels You can easily damage your ears and those of others nearby and or monitors with this gear so cannot stress enough the caution you must exercise 5 7 Channels 1 4 Frequency Response The modified Altec 1567A has many control settings and I O options some of which affect frequency response These include frequency dependent impedance and the imperfect pads of transformer coupled inputs as mentioned in Sections 5 2 and 5 3 The high Z pads and pre triode attenuators of CH1 and CH2 have profound effects that will be described in Section 5 13 Even the channel fader settings subtly affect frequency response as will summarize shortly The response curves in Figure 19 were measured under conditions that give nearly the best flattest and widest response possible in CH1 and CH2 using the balanced inputs CH3 and CH4 data are similar to that of CH1 and are
58. e other half is used in the board s other channel The OPA2604 is an excellent audio performer with very low noise and distortion for a high impedance FET input device here it is configured as a non inverting unity gain buffer voltage follower In series with its input R2 is meant to assure stability of this stage fast acting diode clamps D1 and D2 protect U1A from damage when peak signal 23 voltages exceed 18 V or drop below 18V Note that under such conditions clamping is not isolated from the high Z unbalanced output and distortion will show up there as well as at the balanced output This matter is discussed in Sections 5 8 and 5 14 CHI _ S Hi Z Output Lo Z Output Alert Threshold 15 Turn Trim pot C P Alert LED CH2 Clip Alert Circuits Balanced Line Drivers S NX THAT1646 Ground Plane we Sg i for CH1 r a Bottom Flange of Auxiliary Panel Driver Board Support Plate And Shield Figure 13 Labeled photos of CH1 CH4 s solid state output driver boards Looking up toward the underside of the modified unit the bottom photo shows how both boards mount to the assembly The top photo is a close up of the CH1 CH2 board 24 U1A s output branches to the clip alert circuit described below via R4 and to the balanced output driver IC U2 via R3 The THAT1646 is implemented exactly as set forth in Figure 5 of That Corporation s spec sheet and applications guide for this IC
59. e output impedance In the case of CH5 the full clockwise feedback setting is minimum feedback maximum gain not zero feedback not open loop did not attempt to test or model the gain reduction of this setting versus open loop Thus can t calibrate map to knob position precisely how gain reduction feedback increase as the control is turned counter clockwise reduces the V4 stage s predicted 3 22 KO open loop output impedance However the observed 2 0 KO impedance for the 1 30 setting just clockwise of the design feedback level is at least approximately consistent with the 3 22 KQ open loop prediction Also one can predict that the 11 dB feedback knob range see Section 5 17 should cause CH5 s unbalanced output impedance to range by something like three fold from near 1 KQ at maximum feedback to near 3 KO at minimum The corresponding impedance range for the balanced output is closer to two fold due to loss in the transformer primary s DC resistance which acts in series with the V4 stage source impedance 5 21 Channel 5 VU Meter One obvious and retro cool looking feature unique to CH5 is the illuminated VU meter With two knobs on the vintage panel associated with it illumination control and range switch it s somewhat surprising that the VU meter itself was an optional accessory for the Altec 1567A which the user can install in minutes without soldering as the original manual says see Appendix for link Un
60. e reverse the amount of output stage distortion is affected by both the feedback control see Section 5 17 and the fader setting In general for any given output level CH5 delivers more distortion than CH1 CH4 At the design feedback level and with balanced output set for the nominal 600 Q and terminated at 600 Q distortion at 1 KHz measured 0 45 for a 8 dBV output using a 42 6 dBV input and arbitrary fader setting it was 2 15 for a 18 8 dBV output using a 35 5 dBV input and full fader setting 5 20 Channel 5 Output Impedances and Transformer Characteristics The low Z balanced XLR male and high Z unbalanced 1 4 inch outputs of CH5 may be used separately or simultaneously A dummy load resistor need not be hooked to the balanced output when using only the unbalanced output The two outputs are in parallel with the line transformer interceding for the balanced output The output impedance of the final triode stage V4 the 6CG7 thus determines the impedance of each output reflected through the transformer in the case of the balanced output evaluated CH5 s output impedances using three approaches 1 The direct measurement method as given in Section 5 9 second paragraph for the balanced output at each setting of the output Z switch 2 Inferring V4 stage output impedance from measured voltages across each side of the output transformer while driving a load at each output Z setting and 3 calculating output impeda
61. eedback knob affects gain downstream of the fader in the channel s two stage output driver This is despite the counter intuitive placement of the feedback knob to the left of the fader everyone knows signals flow from left to right Even though one can easily turn the output fader clockwise to saturate the first triode in the output driver V3B in Figure 14 for distortion effects there is no downstream attenuator pot to compensate for increased level Generally the feedback control does not affect a sufficient range of gain to be used for that purpose and works by a different principle Instead the feedback control has its own interesting effects on distortion and tone and interacts with the fader in complex ways to allow a range of voicing best explored by trial and error Feedback control performance is described more specifically in the following section 5 17 Channel 5 Variable Feedback Feature The vintage Altec 1567A used a fixed negative feedback loop in the master channel but as described in Section 4 10 the modified version recruits the un used passive input channel s fader to serve as CH5 s variable feedback control This adds significant sonic variety to CH5 because changing the feedback level not only changes the output driver s gain but affects its linearity increasing feedback linearizes 54 performance thus reducing the routine even order harmonic distortion that triode stages add Section 5 11 introduces ho
62. endix for link stated The VU multiplier is connected directly to the amplifier output rather than to the line side of the output transformer so that the VU meter may be used even though the 15095 transformer is not used Very little compromise is made in the resistive termination of the meter even though the range multiplier is of a simple type In the most sensitive position VU the meter termination is 3450 ohms 11 low and in the least sensitive position 4150 ohms 6 4 high maintaining suitable ballistic characteristics A little algebra confirms that these high and low departure percentages point to 3900 Q as the target source impedance termination for the meter The classic VU meters are designed to bridge a 600 Q line when hooked in series with a 3600 Q resistor the instrument s source impedance is thus 3900 Q the 3600 0 resistor in series with a driver load network impedance of 300 Q This matches the meter s own internal impedance of 3900 Q for maximum power coupling to the meter The Altec 1567A designers aimed for this source impedance because it affects the meter s transient response i e its ballistics which is a critical aspect of a VU meter s dynamic accuracy for the classic VU meter characteristics see Chapter 26 VU Meters and Devices by Glen Ballou in Handbook for Sound Engineers Third Edition Glen M Ballou Editor 2002 Focal Press part of this chapter i
63. es to the output transformer primary and via shielded cables to the 4 inch unbalanced high Z output jack on the auxiliary panel and the VU meter network on the vintage panel The original meter circuit was not re built however its ground lead was segregated from that of the tone network which it had originally shared Accessible on the back of the vintage chassis next to the output transformer socket DPDT mini toggle switch SW2 was added to easily set CH5 s nominal balanced output Z for either 600 Q secondary windings hooked in series or 150 Q parallel hookup Via a shielded twisted pair cable the auxiliary panel mounted XLR male balanced output jack is wired such that its pin 2 has the same polarity as the channel s unbalanced input But note the unbalanced output has the opposite polarity since this channel uses an odd number three of common cathode thus inverting triode stages 5 Performance and Applications of Modified Unit 5 1 Standard Amplitude Units Most of this report expresses signal amplitudes in dBV or decibels referred to 1 Vams i e 0 dBV 1 Vams for ease of calculation When gear to be compared or interfaced specifies amplitudes in dBu or dBm conversion may be desired Since dBu is referred to 0 7746 Vams 0 dBV 2 22 dBu just subtract 2 22 from any dBu value to get ABV or add 2 22 to a dBV value for dBu Units of dBm were most frequently used for vintage gear where 0 dBm 1 mW milliwatt conve
64. esistances yield primary to secondary turns ratios of 4 90 1 and 9 94 1 for the series and parallel cases respectively This is quite close to 5 1 and 10 1 ratios predicted from the nominal impedances printed on the Altec type 15095 transformer and given in the original schematic see Appendix which are primary 15 KQ secondary 600 Q and 150 Q for the series and parallel cases respectively If a 600 Q load is supposed to reflect 15 KQ on the primary as this suggests the reason obtained 18 2 KO could be due either to a slightly defective transformer as suggested by the DC resistance readings discussed above and or errors in my measurements which may be intensified by the exponential square when relating turns to impedances 59 In any case by calling the final triode stage s output impedance 2 0 KQ this second approach predicts balanced output impedances within 15 of the ones directly observed in the first approach Under the test conditions feedback knob at 1 30 balanced output terminated with 600 Q CH5 s unbalanced output Z at 1 KHz is 2 0 KQ in parallel with 18 2 KQ assuming series connection or 1 8 KO It is simply 2 0 KQ when the balanced output is unloaded Importantly however all of these output impedances are affected by the feedback knob setting as discussed next in connection with my third approach to evaluating CH5 s output impedances The third approach involves modeling the performance of the 6CG7 V
65. etween the primary and secondary windings that is hooked to ground noted that it is not terminated with a separate wire as in many isolation transformers It s probably hooked internally to the transformer s housing mounting base 4 3 Power Supply Vintage Unit The schematic for the stock Altec 1567A Appendix shows the original power supply integrated with the overall unit For the modified unit have drawn the corresponding power supply separately as Figure 5 including the actual loads in the case of tube heaters and pilot lamps which keeps the other schematics signal path oriented for less clutter The vintage power transformer has three secondary windings one for the high voltage plate circuits B one for the 12AX7 heaters V1 V3 and one for the 6CG7 heater V4 plus the two incandescent lamps for VU meter illumination While it s logical to describe them separately note that the standby switch SW1 simultaneously affects both the heater circuit and the B supplies for V1 V3 Also note that the original rectifier diodes are replaced with modern units D1 D4 for better performance and reliability The high voltage supply uses voltage doubling rectification since the bridge driven by the high voltage secondary includes C1 and C2 Compared with the original these capacitors have larger values 100 uF versus 60 UF this additional capacity makes the bridge s output 364 V versus the original s 340 V all voltage
66. f of SW1 that is wired across R10 is 14 open This resistor is in series with the heaters for V1 V3 so it limits the current through them by about one half including the current surge on cold start up Therefore powering up the modified preamp in standby mode should make 12AX7 heater failures less likely than in the original design Via its current limiting resistor R11 flashing red LED device D5 is powered by the voltage drop across R10 to advise users the 12AX7s are on standby About 30 seconds after power up in standby mode SW1 can be switched to normal operating mode standby toggle up shorting R10 to extinguish D5 and complete power up of the 12AX7s Powering the unit down should use the reverse procedure i e going to standby mode before main power off so the standby switch is set to the recommended position before the next use The final circuit of the modified vintage unit s power supply is for the AC powered heater of V4 and the two meter illumination lamps It is unchanged from the original design A mid range setting of illumination pot R12 is recommended for maximum lamp lifespan 4 4 Grounding Rule for the Modified Vintage Unit As was routine in the Altec 1567A s era the chassis served as the ground distribution network for most of the common nominal zero volt nodes in the power supply and in the audio circuits The vintage chassis mounted multi section electrolytic filter and cathode bypass capacitors were eq
67. g than hard clipping Sufficient over drive of CH1 CH4 also soft clips negative input peaks as triode current gradually transitions toward cut off for instantaneous grid voltages less than about 3 5 V However at such high amplitudes clipping on the positive side is so severe that the results may be quite harsh even soft clipping has its practical limits Soft clipping in CH1 CH4 is always asymmetrical favoring even order harmonics One Patch Two Types of Distortion FUNCTION GENERATOR CH1 INPUT 100 Hz SINE CH1 INPUT 100 Hz SINE mm t CH2 OUT XFORMER SATURATION INPUT PAD OFF NORM Z CH1 OUTPUT FADERS LOW Z BALANCED OUTPUT o LOW Z BALANCED PAD OFF NORM Z CH2 PRE TRIODE ATTENUATORS LOW Z BALANCED OUTPUT OSCILLO SCOPE BOT TRACE Figure 21 Two examples of waveform distortion available by cascading channels of the modified Altec 1567A In this patch block diagram at far left CH1 s balanced output feeds CH2 s balanced input CH1 s input signal is a 100 Hz sine wave top oscilloscope traces Images of the front panel control settings used to demonstrate input transformer saturation left photo or triode saturation right photo are below the corresponding oscilloscopic results bottom traces are waveforms at CH2 s output All real audio transformers routinely distort signals due to a magnetic memory effect of the iron alloy core ca
68. gure 10 Shielded cables to and from this network were made as short as possible Figure 11 shows the inside of the auxiliary panel s shielded area its cover removed with close up images of the high Z build outs see Section 4 10 for a description of CH5 s The shield cover is designed to be detached and removed with care without disturbing the cables running into or behind the shielded area in case service or modification of the enclosed circuitry is needed Figure 11 Photos of the high impedance circuitry inside the shielded compartment of the auxiliary panel The left image shows CH5 s build out while those of CH2 and CH1 are in the right image The two interior shields separating the channels are visible in the right image Each pot s mounting bushing and anti rotation tab pass through a rectangular aluminum piece the bottom of which engages the shield cover when installed A Neutrik unit with gold plated contacts the 14 inch female jack for the high Z unbalanced input of CH1 and CH2 has normally closed contacts linking the input transformer s signal to the pad attenuator network Inserting a plug opens that connection and substitutes the plug s signal 21 SW1 is a DPDT mini toggle switch wired as SPDT for enhanced reliability its normal setting handle down applies the signal directly to the top CW or clockwise most terminal of pre triode attenuator pot R5 whose wiper is hooked to the triode s grid Th
69. h for easy toggling between series and parallel secondary connections The unit came with three excellent Telefunken 12AX7 tubes and one Raytheon 6CG7 tube All of these tested good and on a breadboard mimicking the vintage mic preamp circuit each 12AX7 triode gave 35 dB gain as loaded by a fader pot s 250 KO resistance at 1 0 KHz and low noise To coax maximum lifespan out of these particular 12AX7 gems decided it was worthwhile to design into the modification a soft startup feature for them this standby switch scheme is detailed in the vintage power supply circuit description Section 4 3 Before its tear down performed a smoke test on the as is unit powering it up gradually through a Variac variable line AC auto transformer The unit s power transformer and other power supply components worked and nothing seemed to overheat Most DC voltages were within 15 percent of those published on the schematic Appendix the plates of V4 were 20 percent low considered this a decent DC result for service neglected gear of this age However audio results were not studio grade by anyone s standard There was hum in the output probably due to old and weak electrolytic filter capacitors Of course my re build plan included replacing all of them even upgrading to somewhat higher capacitances Cosmetically the modified unit won t look as vintage because use axial electrolytics within the chassis not the now
70. he high Z input s pre triode attenuator knob unless killing some high end is specifically desired see Section 5 13 One final remark about the unbalanced inputs note in the schematics Figures 10 and 14 that these inputs are direct coupled in CH1 and CH2 but AC coupled via capacitor C1 in CH5 While direct connection in CH1 and CH2 keeps input coupling like that of a stock Altec 1567A be aware that DC offset at those unbalanced inputs will cause distortion proportional to the amount of offset The CH5 input is immune to DC offset Significant offsets are not present in the modified unit s own outputs and they should be rare in external sources 5 5 Channels 1 4 Gain In each of these channels voltage gain occurs at three stages 1 input transformer 2 triode circuit and 3 output driver For individual stages and whole 34 channels pads off attenuators and faders full clockwise gain measurements were made at 1 0 KHz by comparing input and output voltages read on the Hewlett Packard 331A Distortion Analyzer s RMS voltmeter Input signals from the B amp K Precision 3011B generator were low amplitude to insure a low distortion output Measuring actual amplitudes across inputs makes gain figures independent of source and input impedances Output readings used normal or defined load conditions for example an isolated triode circuit was loaded by the voltmeter s 1 MQ input Z shunted by a 330 KO resistor for a load equivale
71. he turns ratio so if primary turns decreases by 1 2 input impedance becomes 12 1 4 of the full coil value Engaging the pad switches for the CH1 CH4 balanced inputs causes their impedance to range between 1130 1260 Q regardless of the channel frequency and low Z switch setting Series resistors R1 and R3 in the pad networks see Figure 8 dominate input impedance in this case as explained in the following section see Figure 16 Model B Equation 2 MODEL A MODEL B PREAMP PREAMP BALANCED CHANNEL BALANCED CHANNEL SIGNAL SOURCE PAD OFF SIGNAL SOURCE PAD ON amma es zj Poa carr ee m gt r l l l l l Zin ZLoaD Zs 2 mi l l l l l l l l Luain S ee eer eRe ce Zs ZsouRCE Zs ZsouRCE EQ 1 INPUT IMPEDANCE EQ 2 INPUT IMPEDANCE 1 Zin ZLoap Zin RI R3 bog ek Zloap R2 EQ 3 LOSS dB DUE TO Zgource EQ 4 LOSS dB DUE TO Zgqygce AND PAD 1 1 1 eo Zi oan Zioap R2 LOSS 20 log LOSS 20 log Zsource ZLoaD 1 ZsoURCE RI R3 1 1 Zloap R2 EQ 5 DECREASE IN AMPLITUDE WHEN PAD IS SWITCHED ON dB PERCEIVED LOSS _ LOSS DUE TO Z ource AND PAD dB LOSS dB DUE TO Zsource DUE TO PAD dB MODEL B EQ 4 MODEL A EQ 3 Figure 16 Modeling impedances and signal losses for CH1 CH4 s low Z balanced input pads Model A left column applies when pad is off and Model B right column is with pad engaged both models are recruited for Equation 5 at
72. hen experimenting with transformer saturation this knob must be nearly fully counter clockwise High Z Pad The normal position of this toggle switch is down The up position engages a 20 dB pad that expands the useful counter clockwise range of the attenuator knob It may be useful when applying an exceptionally hot signal to the balanced input 3 4 Channel 5 This channel has a high Z input with a pad switch and attenuator pot built into the shielded high Z enclosure of the auxiliary panel Otherwise it is similar to the old master channel retaining its all tube signal path and the VU meter except negative feedback is made variable using the original MIX 5 fader pot for the old passive input channel eliminated upon 10 modification Decreasing feedback clockwise yields more gain and harmonic distortion CH5 s transformer coupled low Z balanced output can be set for nominal 600 or 150 ohms series or parallel secondary connection respectively using a toggle switch added to the rear of the vintage unit 3 5 A Word About Solid State Outputs At least with respect to their balanced outputs CH1 CH4 are hybrid meaning they have both tube and solid state circuitry The outputs of these channels each use a high quality op amp followed by a state of the art THAT 1646 balanced line driver chip As a byproduct of the way the latter chip creates a transformer like low Z differential output they contribute nearly 6 dB voltage g
73. high Z input peaks falling below the triode saturation threshold are about 1 7 dB lower for CH5 than for CH1 CH4 in Figures 14 and 10 note that cathode bias is 0 82 V for CH5 s first triode V3A versus about 1 0 V for CH1 CH4 5 16 Channel 5 Keeping Track of Knobs Control of amplitude and tone in CH5 s signal path is quite versatile but involves five knobs The one on the auxiliary panel pre triode attenuator should usually be fully clockwise as already discussed so will focus on the other four which are on the vintage panel The bass and treble knobs control a tone network or stack as it was sometimes called driven by input stage triode V3A see Figure 14 Neutral settings of these controls are near their 12 o clock positions but not exactly marked the positions that gave the best 1 KHz square waveform on oscilloscope for the overall channel and hence represent an approximately flat frequency response this test was done with the feedback control set at the vintage design level marked near 12 o clock Marked knob positions can be seen in Figure 1 Directly following the tone network along CH5 s signal path is the knob call the channel fader or the channel s output fader formerly the original design s master volume control Normally one thinks of an output fader as the final amplitude control point of a channel after any gain control But this does not apply to CH5 as modified the variable f
74. ht decrease in SNR at counter clockwise settings is predicted from the thermal Johnson noise of the stage s output impedance see table in Section 5 8 for output impedance at the fader s wiper Balanced Output at O dB Attenuation dBV 110 o o Signal to Noise Ratio SNR dB So o I oO fez 50 51 10 15 Harmonic Distortion Percent 20 scene CURVES Predicted using put driver p formarea N optimal ou SOLID CURVES Based on conservative asurement St driver aiis 40 30 20 10 0 Fader Attenuation dB Figure 25 Left chart Signal to noise ratio SNR at CH1 CH4 s balanced outputs terminated at 600 Q as a function of fader attenuation Labeled output values for each pair of curves solid and dashed is output amplitude with full clockwise fader 0 dB attenuation add the fader attenuation to the labeled output value for actual output level at any fader setting Solid curves use measured triode stage and output driver noise the latter being conservative dashed curves use measured triode stage noise and the published THAT1646 IC s noise figure see text Red shaded area indicates clipping conditions for the line driver in differential mode dark shading only or single ended mode both light and dark shading Right Chart Percent harmonic distortion due to the triode stage as a function of tha
75. icrophone mixer There are five input channels a summing amplifier mixer and a master channel with three outputs ALTEC 1567A BEFORE MODIFICATION Block Diagram of Audio Signal Path C l INPUT 1 Channel RECORDER Input Transformer Fader aga OUTPUT Altec 4722 VU Output G3 Q Driver OUTPUT INPUT 2 DUET f Channel iz 8 nput Transformer Fader op Master Altec 4722 Volume Line Output Summing Bass Transformer Amplifier Altec 15095 Negative Feedback Channel Fader a INPUT 3 Input Transformer Altec 4722 KEY TO SYMBOLS USED IN BLOCK DIAGRAMS Plug In Balanced Low Impedance Input or Output Audio YYY Transformer Ground Referenced High e 2 Impedance Input or Output INPUT 4 Channel Input Transformer Fader Altec 4722 WD INPUT 5 Channel Fader Solid State Stage Normalled Jack for High Impedance Input Vacuum Triode Xj Switch or Relay Stage Potentiometer Figure 2 Block diagram of audio signal path in original stock Altec 1567A The symbol key inset at lower right also applies to Figure 3 s block diagram One input channel Number 5 is passive and offers a high impedance high Z unbalanced input only The others 1 4 each have a single stage voltage amplifier using one triode of a 12AX7 vacuum tube Altec Lansing s way of making these active channels versatile was to provide octal sockets into which the following could be inserted 1 a simple lin
76. in for this to possibly work Note that CH1 CH4 s polarity switches affect only their balanced inputs and that the balanced output connection already compensates for inversion by the triode stages as described in Section 4 8 So with a balanced patch between channels the second channel s normal polarity setting would tend to compound harmonic distortion while it s invert setting should subtract it This is but one of many experiments with sonic character worth a try on the modified Altec 1567A At the triode operating points for CH1 CH4 triode saturation over driving occurs when instantaneous positive grid voltage approaches the cathode bias voltage near 1 V Current through the triode is then maximal further positive voltage swing at the grid cannot cause current to increase Voltage drop across the plate load resistor is maximum so the output signal flattens out soft clips on the negative bottom side of the waveform An inverted example is in the bottom oscilloscope trace on the right hand side of Figure 21 inverted because the polarity compensated buffered output fed the scope Clipping is soft because thresholds are imposed relatively gradually with vacuum triodes Indeed the point at which routine harmonic distortion gives way to soft clipping is not distinct but it occurs as distortion reaches 3 to 5 percent see Figures 25 and 26 in Section 5 14 Such limiting is less harsh soundin
77. including all new ceramic tube and transformer sockets and all new coupling and electrolytic capacitors Exceptions original wiring and parts were left in the tone control and meter range networks The internal grounding scheme is upgraded to a hierarchical star grounding rule rather than multiple chassis tie points To house most of the additional circuitry required by the modifications including the solid state output drivers their power supply input and output jacks and additional knobs and switches the vintage unit is permanently married to a four rack space black aluminum panel called the auxiliary panel Figure 1 shows front and rear views of the assembly with some key parts labeled It is held together with thick aluminum side panels and stiffened with two horizontal steel braces handles in the rear The entire 25 pound seven rack space assembly is rack mountable Each channel s I O jacks and controls are aligned vertically between the two panels for an intuitive layout except that most of channel 5 s controls and its VU meter occupy the top and right side of the vintage panel Labels for controls and jacks have channel specific colors for ease of use Locating I O jacks on the front of the auxiliary panel rather than less accessibly on a rear panel facilitates experimenting with patching the various channels together in different ways Bass Treble Feedback _ VU Display CH5 CHS CHS CHS Range Meter Illumination
78. indings supplied with actual DC not by direct ohmmeter readings on my digital multi meter the latter uses pulses giving inaccurate results for inductive devices The DC resistances measured 1610 Q for the primary 35 0 Q for the secondary between pins 1 and 3 and 47 9 Q for the secondary between pins 4 and 6 Thus an expected equal resistance for the two secondary windings was not observed possible explanations are insulation failure a short between turns within the pins 1 3 secondary or a high resistance connection linking pins 4 6 to their winding did not have a pristine type 15095 unit for comparison In the second impedance determining approach fed a 1 KHz sine wave to CH5 s input and with the feedback control at about the 1 30 position flattest frequency response slightly clockwise from the design level see Section 5 17 hooked a 600 0 load to the balanced output and measured the voltage at each output when the Z out switch was in each position series nominal 600 Q versus parallel nominal 150 QO secondary With channel input amplitude and fader position constant but arbitrary respective RMS voltages at the unbalanced and balanced outputs i e primary and secondary sides of transformer were 24 25 and 4 00 for the series connection and 26 0 and 2 48 for the parallel connection Calculating impedances and turns ratios from this data requires four reasonable assumptions 1 Switching from the series to parallel sett
79. ing doubles the effective primary to secondary turns ratio The nominal and observed four fold difference in impedance between these settings discussed above supports this assumption because reflected impedance is a function of the square of the turns ratio in an ideal transformer 2 The output impedance of the final triode stage V4 the 6CG7 and its source voltage were constant during the test 3 The unbalanced output measurements represent the combined voltage drop across the 1610 0 DC resistance of the primary see above paragraph in series with the balanced output s load impedance reflected through the transformer to the primary with the load held constant the latter is four fold greater for the parallel versus the series case consistent with the first assumption 4 The effective load on the secondary is the load resistor 600 for this test in series with the DC resistance of the secondary which is 82 9 Q or 20 2 Q for the series or parallel hookups respectively as suggested by the data in the preceding paragraph The data and assumptions given above are sufficient to simultaneously calculate the following 1 A source V4 stage output impedance of 2 0 KQ and 2 reflected load impedances of 18 2 KQ and 72 8 KQ for the series and parallel secondary connections respectively when driving a constant 600 Q load Voltage drops across these impedances compared with those in the secondary after accounting for voltage drops in the DC r
80. inimum length patch cables to the CH5 input and the CH1 CH4 outputs to be mixed Since CH5 lacks a low Z balanced input that could be used with a microphone a common application may be as an input channel for instruments Thus it can act as an active direct interface DI box complete with gain and tone controls Of course one or both of CH5 s outputs can drive CH1 CH4 for extra gain and triode saturation distortion effects But unlike 53 CH1 CH4 it may not cleanly deliver sufficient output amplitudes for input transformer saturation experiments Also since CH5 is noisier than CH1 CH4 see Sections 5 14 and 5 19 expect quieter results when CH5 is the second rather than the first channel of a two channel cascade The input impedance of CH5 is similar to that of CH1 and CH2 when the latters unbalanced inputs are used see Figure 18 and Section 5 4 This is not surprising given their similar high Z input circuitry identical except CH5 is AC coupled see sections 4 7 and 4 10 As with CH1 and CH2 CH5 s pre triode attenuator should be kept fully clockwise and the high Z pad switched off for most applications except for rare cases when extreme attenuation or a high frequency roll off is desired see Section 5 13 for characteristics of the high Z attenuator pad switch If first stage triode saturation is not wanted and input amplitude is too high level should be turned down at the source not at the pre triode attenuator Maximum
81. input was a sine waveform of various frequencies with amplitude too low for triode saturation its output was thus clean up to the solid state driver clip threshold Amplitude at the balanced input of CH1 or CH2 was measured using the H P 331A s RMS voltmeter while an oscilloscope compared CH1 or CH2 s output waveform to that of CH4 s input At distortion thresholds it s easy to distinguish driver clipping from transformer saturation because they manifest at maximum excursions versus zero crossings respectively Theoretically maintaining a given magnetic flux amplitude including the saturation threshold as frequency doubles requires doubling the input voltage This is because flux amplitude is proportional to current in the windings which due to their inductive reactance require double the voltage to maintain constant current as frequency doubles Thus the saturation threshold should rise at 6 dB per octave Shown in Figure 22 my actual results approximate this prediction but the observed slope is closer to about 7 dB per octave It is unclear whether the difference is due to 1 the transformer is not ideal 2 the test signal is not a pure sine 44 waveform having been amplified in CH4 s triode stage 3 measurement error or 4 a combination of these Figure 2
82. is is a robust mil spec 1 MQ log taper pot from Precision Electronic Components Ltd its operating voltage limit is 500 Vrms which should be sufficient to withstand distortion experiments using transformer saturation see Sections 5 11 5 12 and 5 13 In SW1 s handle up setting pad engaged R4 is placed in series with pot R5 and R6 shunts the R4 R5 series The result is a 20 dB loss while keeping the resistive load presented by the network near 1 MQ However capacitive reactance in the triode lowers the network s input impedance to about 840 KQ at 1 KHz for the normal settings attenuator full clockwise and pad off see Section 5 4 for more details If you are concerned about the high Z build out network of CH1 and CH2 causing high frequency loss due to stray capacitance your concern is justified Fortunately as will show in Section 5 13 the most useful settings of the pre triode attenuator pot fully clockwise or the region near fully counter clockwise are not adversely affected But users of this experimental gear need to be aware of this 4 8 Channels 1 4 Balanced Line Drivers The solid state differential output driver and clip alert indicator circuits used by CH1 CH4 are diagrammed in Figure 12 Only one of the four identical driver indicator channels is shown the driver stages using ICs U1 and U2 are in the top half and the indicator circuit using U3 and U4 the bottom half of this schematic This circuitry is b
83. is transformer to CH4 in the modified preamp Users should be alert to CH4 s irregular and possibly defective input transformer while taking advantage of its different impedance To evaluate the impedance of each balanced input used the millivolt meter in a Hewlett Packard 331 Distortion Analyzer which has a 1 MQ input to measure RMS voltage drops across a 1 0 KQ resistor in series with the channel input Low amplitude test signals were sine waveforms of various frequencies from a function generator B amp K Precision 3011B whose verified 50 Q output Z was factored into the calculations Note that this technique gives only the magnitude of the impedance vector not its angle i e the proportion of resistive and reactive components is not determined 29 In the case of impedance and pad switches set for normal handle down the results for CH1 CH4 are plotted in Figure 15 For each channel input impedance peaks in the audio mid band this is most dramatic in CH2 which lacks the primary shunt resistor and impedance reaches nearly 2 KQ at 1 KHz CH4 s irregular transformer also lacks the shunt but this channel has a broader impedance peak which reaches only 390 Q By virtue of their shunt resistors CH1 and CH3 probably most closely reconstitute stock Altec 1567A mic input channels in the modified unit they display a fairly broad impedance peak reaching 185 Q Sagging input impedance with decreasing bass frequency presum
84. ise ratio SNR 56 measured noise under the conditions that yield 54 5 dB overall gain between CH5 s input and balanced output as reported in Section 5 18 i e feedback control set at the design level balanced output set for nominal 600 Q and terminated at 600 Q except the input was grounded instead of receiving a signal Noise at the balanced output as read on the Hewlett Packard 331A s average responding voltmeter was corrected to represent true RMS see Section 5 14 subtracting meter self noise was not necessary given this channel s relatively high noise output Noise was 2 03 MVpams 53 8 dBV with fader full clockwise and 1 01 mMVems 59 9 dBV with fader full counter clockwise An assumption that all measured noise is in the audio band seems reasonable given the channel s frequency response described in Section 5 18 Assuming that the pre and post fader gain stages generate uncorrelated noise the first triode stage s contribution as amplified by the subsequent stages is the square root of the difference between the two squared RMS readings given above or 1 76 MVams That noise decreases with fader attenuation turning fader counter clockwise while the relative effect of the post fader stages constant noise 1 01 MVays increases This makes the equivalent input noise EIN increase as the fader is turned down as given in the following table Fader Attenuation dB EIN dBV SNR dB O full
85. k between socket pins 5 and 7 for a high Z unbalanced input 2 a phono equalizer assembly for old style record players channels three and four only or 3 a model 4722 microphone transformer for a low impedance low Z balanced input Only the transformers are shown in Figure 2 With its center tapped primary winding the 4722 offers either 150 or 38 Q nominal input impedance Each input channel has a big knob labeled MIX N where N channel number which is a channel fader in today s jargon These rotary fader pots feed a monaural mix bus an assignment that cannot be changed without modification Worth noting is an important difference between this typical 1960 s mixer head and today s consoles The vintage units lack gain controls analogous to the ones usually found at the top of modern channel strips The gain of the vintage channels is always set at maximum so useful fader settings are often in the lower counter clockwise range of the rotary faders Using modern consoles engineers normally set channel gains so that faders linger near 0 dB for the best signal headroom noise compromise In an Altec 1567A balanced input channel the transformer and triode yield a combined voltage gain of about 60 dB This is too much gain for some hot sources and no fader setting can give a clean signal suggesting input pad options as a useful modification Fed by the mix bus the summing amplifier uses one triode of
86. kwise or full clockwise measured total noise using the Hewlett Packard 331A s voltmeter on XLR pin 2 of the buffered outputs in single ended mode loaded at 600 Q The balanced channel inputs were open not terminated but recall that CH1 and CH3 each have a built in 220 O shunt resistor all relay control switches were in normal down positions and no pre triode attenuation or high Z pad was applied on CH1 and CH2 Readings differed by less than 1 5 dB between channels will present results from CH1 as typical With full counter clockwise fader channel noise comes only from the solid state output stage the raw measurement was 283 UVams The voltmeter s own noise its input shorted to ground was 68 UVaws The H P 331A s voltmeter is average responding so all noise readings have been multiplied by 1 13 to get these true RMS figures Channel noise is unrelated to voltmeter noise so subtracting the latter from the channel s reading requires taking the square root of the difference between the two squared readings this gives 275 UVaus as the corrected output stage noise voltage But only the part that is in the audio band 20 Hz to 20 KHz is relevant The published small signal bandwidth spec for the THAT1646 output driver chip is 10 MHz A fair assumption is that its noise power is equal per frequency increment i e white noise equal power per Hz extending out to 10 MHz At this point need to digress about the f
87. lled hysteresis Essentially hysteresis makes the device s transfer characteristic have two separate but close parallel curves one for each direction of current Normally in the absence of a DC offset or core magnetization waveform distortion is symmetrical and odd order harmonics are favored The relative amount of distortion can be large for very low amplitude signals quite loosely analogous to digital audio s quantization 42 43 distortion and it affects low frequencies the most Perhaps transformer distortion accounts for some of the vintage character audio engineers seek in their tone An excellent resource for transformer topics is Chapter 11 Audio Transformers by Bill Whitlock in Handbook for Sound Engineers Third Edition Glen M Ballou Editor 2002 Focal Press a PDF file of that chapter is available at http www jensen transformers com an Audio 20Transformers 20Chapter padf Only CH1 and CH2 are set up for practical experiments to hear input transformer saturation A sufficiently high amplitude signal on the primary causes magnetic flux density in the core to reach a maximum saturate With practicable input amplitudes only the lower end of the audio band can be targeted as explained in the next section Alternating magnetic flux density in the core matches the phase of the current in the windings which lags the voltage by 90 degrees So flux maxima occur when instantaneous voltage is crossing zero in
88. n different ways and in any order In this digital age most audio engineers agree that well maintained vintage outboard vacuum tube gear lends body warmth and character to recorded tracks However opinions differ on the extent that the venerable old equipment should be modified At one extreme purists want to stay true to the vintage design On the other hand pragmatists welcome modifications that add versatility This modified Altec 1567A is on the pragmatic end of the spectrum obviously it is a hot rod not a simple restoration It is an experimental transformer coupled tube based gain engine tailored to the adventurous audio engineer Solid state line drivers buffer the outputs of the four transformer coupled mic preamp channels Relay controlled pad polarity and impedance switches are added to their input circuits Two of these channels are given variable attenuators and an unbalanced input option between their input transformer and triode stages this allows experimenting with transformer saturation as a distortion effect As a fifth independent channel the old master channel is provided an unbalanced input and a few other modifications it retains an all tube signal path tone controls transformer balanced output and the VU meter It can be an input channel or Dl if desired The original Altec 1567A vintage unit was stripped down to the bare chassis before rebuilding with upgraded parts in critical cases
89. n series with their input which drops input potential by about 2 1 volts and diverts some unnecessary heat dissipation away from the regulators Physically close to the input pins C9 and C10 are bypass capacitors recommended when using three terminal voltage regulators At their outputs D11 and D12 help protect the regulators in the unlikely event of back EMFs produced by the loads 16 and C11 and C12 help filter out noise in the regulator outputs Of course the power supply s 18 V outputs come directly from the regulators yellow LEDs D21 and D22 mounted on the auxiliary panel to the left of the main power switch indicate power on status To get the 15 V outputs each regulator s output potential is dropped through a series of four diodes D13 D20 then filtered by C13 and C14 Vintage Unit Shielded Hi Z Area Star Ground for Auxiliary Panel Bottom Flange of Auxiliary Panel Figure 7 Labeled photo of installed power supply board for auxiliary panel 17 4 6 Channels 1 4 Balanced Input Circuits Linking the female XLR jacks to the input transformer primaries CH1 CH4 s balanced input circuits are almost identical so the schematic in Figure 8 includes just one channel s signal path to minimize clutter For each channel three Omron G5A 234P DPDT relays RY1 RY3 execute pad polarity and impedance options along with associated components six relays occupy each of two perf boards mounted adjacent to input transf
90. nce from published 6CG7 tube characteristics The direct Zour Measurements for the balanced output were taken at 1 KHz with the feedback knob at the design level With the output Z switch at the nominal 600 O setting transformer secondary windings connected in series Zour measured 169 O Set for the nominal 150 Q output the parallel connection Zout was 42 Q While these results are reassuringly close to an expected four fold impedance difference between the two settings the measured impedances are each nearly 3 6 times lower than their nominal values This somewhat contradicts my assertion in Section 5 2 that vintage era engineers aimed for maximum power transfer by matching source and load impedances apparently this was not exactly the case at the Altec 1567A s transformer coupled output There is certainly nothing wrong with driving a load from a 3 6 fold lower impedance As will explain shortly the second approach to determining output impedances also yields the transformer s voltage step down ratio which equals the turns ratio the number of turns in the primary winding per each effective turn in the secondary Let me first report my DC resistance measurements on the transformer because this affects the math It also suggests that this 58 particular Altec type 15095 line output transformer may be defective or damaged slightly even though it still works measured DC resistances by looking at voltage drops across w
91. nt to a 250 KQ fader pot Such a measurement equals the voltage output of the channel s high Z unbalanced output working into an open circuit note that load presented by the output driver circuit s very high impedance is negligible Channel gains were equal within 0 75 dB and the average result is given here With the normal input transformer impedance setting in which the input couples across the full primary winding the transformers voltage gain is 25 dB On the low Z setting which uses the center tap transformer gain is 30 3 dB This increase of 5 3 dB is less than the expected 6 0 dB when halving the primary to secondary turns ratio of an ideal transformer Being real the Altec model 4722 input transformers are expected to fall short of ideal performance probably due mostly to loss in the DC resistance of the secondary winding However should note that my tests on the low Z setting were not as extensive replicated as on the normal setting some measurement error is also possible The triode circuits deliver 35 dB gain when loaded by 250 KQ the fader pot resistance in these channels Note that gain will decrease according to the load on a channel s high Z unbalanced output see Section 5 8 Based on a 12AX7A characteristic chart published by RCA in 1960 a predicted gain for this circuit is 30 3 dB The better than predicted gain could be a function of the excellent vintage Telefunken 12AX7s used or 12A
92. ompact as possible It also eliminates the need for the original vintage unit s power switch the modification places a standby switch for the 12AX7s at that physical location A single fuse protecting the entire system is located on the auxiliary panel the position of the original fuse is occupied by a flashing standby red LED indicator on the modified vintage panel 4 Schematics and Circuit Descriptions of Modified Unit 4 1 Note About Schematic Diagrams In this section the design of the modified Altec 1567A is presented as a set of seven schematic diagrams three for the power supplies three for CH1 CH4 and one for CH5 You may need to zoom your PDF viewer to read the detail in these drawings When comparing them to the schematic of the stock vintage unit Appendix please note that numbered components e g SW2 R3 etc are not meant to correspond between the stock and modified units Also sequential numbering is reset for each different schematic of the modified unit Therefore text references to component numbers apply specifically just to the schematic being described The only exception is the vacuum tube designations V1 V4 which refer to the same parts in all schematics both within this section and in the Appendix ORIGINAL POWER TRANSFORMER OF ALTEC 1567A SECONDARY NOT SHOWN METAL OXIDE VARISTOR HEAVY BRAID CWITH BLU INSULATION a AVEL LINDBERG MODEL Y2365a2 16 VA TOROIDAL TRANSFORMER
93. ormer sockets in the vintage unit Figure 9 is a view inside the vintage chassis showing these boards ALTEC 4722 INPUT TRANSFORMER LOW IMPEDANCE 562 MF RY1 RY2 RY3 9 BALANCED INPUT GSA 234P GSA 234P GSA 234P XLR FEMALE 8 A ne r e 10 10 D aa ay Tt z Be I I K gt ot O pia Sj z1 I gt gt gt F I 3 3 D be R2 6 TOT ten ga to I Ez Tux isa mr a E 83 I W mn ws OG R3 562 MF l Vs yir 2 12 4 1 12 1 12 Vk CHANNELS 1 AND 2 RELAY g7 g z7x S preity CONTROL USES 12V 1 1 i CONE CHANNEL SHOWN HERE z 5 z EXCEPT FOR R7 IN CH1 AND CH3 ONLY A aN BALANCED AUDIO CIRCUIT FROM INPUT JACK TO TRANSFORMER IS IDENTICAL Er J 4 WIRE RIBBON FOR CHANNELS 1 THROUGH 4 12V fn ONLY ONE CHANNEL IS SHOWN HERE a sw2 SW3 o4 oy INPUT JACKS AND CONTROL SWITCHES ARE ON AUXILIARY PANEL RELAYS ARE i 7 Ta LOCATED IN VINTAGE UNIT ON BOARDS l l l NEAR INPUT TRANSFORMER SOCKETS lo 1o 1o oe l TAx CONTROL SWITCHES AND RELAY CONTACTS SHOWN IN NORMAL POSITION IN WHICH R4 POLAR RS IMPED R6 RELAY COILS HAVE ZERO CURRENT THE 750 ITY 75 ANCE 758 OPPOSITE STATES FOR THE CONTROL 172W 172W 172W SWITCHES ARE LABELED AS FOLLOWS j S SW1 PAD ON RY1 RY3 SW2 POLARITY INVERT SW3 IMPEDANCE LOW CHANNELS 3 AND 4 RELAY 3 CONTROL USES 12 CONE CHANNEL SHOWN HERE Z RY2 T 12 4 12 ALL RESISTOR VALUE
94. ossible With high output microphones on loud sources the low Z not high Z pads should be used and if the mic has its own pad switch it should be tried first The reason pre triode attenuation should be avoided is that it can cause significant bandwidth loss depending on the setting The problem is stray capacitance associated with the attenuator networks and their I O lines Rather than attempting a formal model will give my observations first and then remark on how stray capacitances may explain them Using CH1 and assuming results for CH2 and CH5 would be similar at different pre triode attenuator and high Z pad settings measured response relative to that of 1 KHz for frequencies of 1 KHz and greater With source Z 50 Q the sine wave generator fed CH1 s balanced input which was set for full primary coil nominal 150 Q and low Z pad off With fader full clockwise amplitude was measured at the low Z output which was used in single ended mode terminated with 600 Q Frequency response with the high Z pad turned off is portrayed on the right hand side of Figure 23 the pre triode attenuator settings that were examined are diagrammed below the curves To help sort out the curves those in red represent clockwise range settings until the setting with maximum high frequency roll off is reached at 3 00 three o clock and the black curves are more counter clockwise settings While high frequency loss is present for most settings
95. r offers It is always present hence routine using CH1 CH4 because each channel s triode gain stage lacks feedback Such a stage has a non linear transfer characteristic incremental changes along the input waveform s voltage axis do not map to exactly proportional changes in the output waveform For example the output voltage swing caused by a change of instantaneous input voltage from say 100 to 150 mV is greater than that caused by a 100 to 150 mV input change This asymmetry adds even order harmonics accounting for much of the musicality of triodes With faders full clockwise and input amplitude just sufficient to give the maximum un clipped buffered output of 20 5 dBV single ended mode into 600 Q 1 KHz my Hewlett Packard 331A instrument measured total harmonic distortion THD for CH1 CH4 at about 0 5 percent Greater THD is available by increasing the input amplitude while compensating with the fader and less obtains by decreasing the input amplitude In Section 5 14 the right hand graphs in Figures 25 and 26 show how distortion depends on output amplitude for a triode circuit like those used in CH1 CH4 Theoretically since common cathode stages invert the signal the second of two identical cascaded channels could either remove or increase routine harmonic distortion produced by the first depending on the second channel s polarity switch setting the first channel s fader attenuation should equal its ga
96. ration in CH2 can severely distort bass frequencies at amplitudes well below the driver s clip threshold If required CH1 CH4 s clip alerts are easily adjusted if a function generator oscilloscope with 10X probe s and a small screw driver are available A spare female XLR connector with shell removed clip leads and various resistors can be used to test different loads on the outputs connect such dummy load resistors between pins 2 and 3 Or when observing clipping into the input of another channel or device the shell of the patch cable s female XLR end should be loosened and slid back for access to the terminals while plugged in In either case for differential balanced output mode oscilloscope probe s can hang on either XLR pin 2 or 3 or both if multiple scope channels are available and their grounds clipped to pin 1 Do not ground a probe to an output leg pin 2 or 3 unless single ended mode is specifically desired in which case it is best to tie pin 1 directly to either pin 2 or 3 anyway Set your function generator for a low amplitude and patch its output into an input of the channel to be adjusted a pad may be required if the generator s minimum output exceeds 40 dBV or 10 mVams A 1 KHz sine waveform makes a good test signal but sometimes triangular waveforms are better for observing nascent clipping on the scope With the channel fader and pre triode attenuator for CH1 and CH2 at full clockwise slowly incre
97. re does not strictly adhere to an ideal star grounding rule even though use the term star ground as an approximation of this network It is imperfect for two practical reasons First the star ground terminal is not a point but a short loop albeit heavy copper wire as noted in Section 4 2 Second in contrast to what my schematics might imply in an effort to avoid clutter some chain grounding is used within channels or tube stages over short distances in cases where relatively low current is expected These sub stars originate at the negative terminals of decoupling or filter capacitors serving that stage or channel You may visualize this as a hierarchical star grounding rule in the modified vintage unit 15 4 5 Power Supply Auxiliary Panel As shown in Figure 6 the auxiliary panel s power supply has three sets of bi polar outputs 18 V for the balanced line drivers in CH1 CH4 15 V for the associated clip alert circuits and 12 V to control relays in the balanced input signal path of CH1 CH4 This power supply was built on a perf board using point to point wiring with attention to heavy and redundant ground conductors leading to the star ground terminal at the bottom of the board A photo of the installed board is shown in Figure 7 AX Star Ground for Auxiliary Panel DC voltages referred to Ax 1502 ew 12V and 12V outputs are unregulated so they require constant proper loads AVEL
98. requency response of the voltmeter On the 1 mV range that used the H P 331A voltmeter s published bandwidth is 5 Hz to 3 MHz This spec refers to the instrument s flat within 0 45 dB frequency response signals beyond 3 MHz also deflect the meter As H P Application Note 206 1 dated October 1977 available at http www hpmemory org an pdf an_206 1 pdf puts it it may be desired to measure the noise in an audio amplifier In this case we are interested in the noise only from 20 Hz to 20 KHz as this is the maximum range the ear can hear This is easily done with the 3045A program the spectrum analyzer discussed in Application Note but without external filters the 331A will measure the noise out to several megahertz As there could be significant noise beyond 20 KHz the 331A might 49 read considerably higher than the desired result One must not depend on the roll off of the audio amplifier to limit the noise as it often will not Indeed there s no reason to think the THAT 1646 output driver s own noise is bandwidth limited until 10 MHz Unfortunately do not know the noise bandwidth of the H P 331A more accurately than out to several megahertz So will make an assumption According to Walter G Jung IC Op Amp Cookbook 3 Ed 1986 ISBN 0 672 22453 4 p 44 if asystem s bandwidth is limited by a single pole low pass filter 6 dB per octave roll off defined by 3 dB response at fo its noise bandwid
99. resistor R32 in Appendix to set feedback between the output driver s two triode stages V4 and V3B Instead used an interpretation of an adjustable feedback modification on an Altec 1567A mentioned by Eddie Ciletti in his December 2010 Mix magazine Tech s Files column pages 60 62 My version wires the old mix 5 250 KQ pot the original passive input channel fader R18 in Figure 14 as a rheostat to replace the fixed 47 KQ feedback resistor At 22 KO R17 limits the maximum negative feedback allowed when R16 is full counter clockwise On the vintage panel marked the feedback knob position equivalent to the original fixed resistor the design feedback level more clockwise settings decrease feedback thus increasing gain and harmonic distortion compared to the original design and vice versa for more counter clockwise settings see Section 5 17 Aside from the shielded cables to the front panel necessary to patch in R17 and R18 other components completing the feedback loop are like the original design C12 blocks DC C10 may be for stability and or to tweak high frequency response and R15 couples feedback to V3B s cathode As in the original units master channel V4 s output is AC coupled using a 1 0 uF capacitor C14 in Figure 14 This is the only non SBE new coupling capacitor in the modified vintage unit it s a generic metalized film tubular unit sold by Antique Electronics Supply From there the signal branch
100. rsion to dBV or dBu requires knowing the load 28 impedance or resistance Conversion is trivial for 600 Q loads because dBm dBu in that particular case and it is the most common case The original Altec 1567A manual see Appendix for link uses dBm Section 5 21 of this report expresses CH5 s output in dBm for easy cross reference to that manual Volume Units VU which include aspects of signal dynamics are not necessarily converted easily into decibel units see Section 5 21 5 2 Channels 1 4 Impedance of Balanced Inputs In the classical era of audio engineering when the Altec 1567A was designed optimum signal transmission from a source Such as a mic to an input as in a preamp channel was usually assumed to need maximum power coupling This calls for equal source and input impedances where voltage at the input terminals equals that dropped across the source s internal impedance a 6 dB voltage loss In our modern era audio gain models use voltage terms at least in preamp stages so impedance matched transmission lines seem inefficient to us For example today s mic inputs commonly have impedances some 10 fold higher than the mics they host this is called a bridging connection for a much smaller voltage loss across the mic s internal impedance However don t necessarily favor throwing out yesterday s audio traditions for the sake of a few dB of efficiency such decisions can have trade offs Source pre
101. rves the 12AX7s V1 V3 via one half of DPDT standby switch SW1 In standby mode toggle handle in down position high voltage to the six 12AX7 triode circuits is switched off and R3 serves as a dummy load equivalent to these circuits Standby mode is intended as a soft start feature for the 12AX7s while these tubes are warming up at reduced heater current see below keeping their high voltage turned off may help minimize cathode stripping and extend their lifespans In normal operating mode standby toggle handle up the DC characteristics of the B supply network for V1 V3 is equivalent to the original design however the four triode circuits of V1 and V2 i e channels 1 4 are decoupled through individual R C filters In other words R40 and C19A in the original schematic see Appendix was the single R C stage common to all four triodes of V1 and V2 the modification Figure 5 replaces this with series stage R4 C5 followed by parallel stages R5 C6 R6 C7 R7 C8 and R8 C9 These parallel stages respectively decouple V2B V2A V1B and V1A It is a more robust and stable approach and extends cross talk suppression among CH1 CH4 to higher frequencies than the original design provided The DC power supply for the 12AX7 heaters is like the original except for higher value filter capacitors C10 and C11 2200 uF versus the original 1000 uF for cleaner DC and the addition of the standby switch network In standby mode the hal
102. ry the clip alert indicators can be set to indicate diode clipping by adapting the calibration procedure described in Section 5 10 5 9 Channels 1 4 Balanced Output Characteristics The THAT 1646 balanced output driver ICs used in CH1 CH4 act like output transformers except in at least five ways 1 Across the audio band output impedance is essentially independent of frequency 2 A broad range of input impedances can be driven directly a load resistor shunt is not required when feeding higher Z inputs e g 10 KQ line level inputs 3 When saturated the distortion mode is hard clipping 4 Ground referencing one output leg for single ended operation nearly halves the 38 clipping threshold unless current limited see below 5 As set up in CH1 CH4 the driver chips may be damaged if hooked to mic preamp inputs with phantom power active so this condition needs to be carefully avoided As a direct approach to determine balanced output impedances Zour measured output voltage with ELoap and without Eno loan typically used 1 Vams various known load resistors Rioap across the output between XLR pins 2 and 3 with pin 3 grounded then solved the equation Zout Rioap Eno LOAD Eroan Eroab Note that the 1 MQ impedance of the H P 331A s voltmeter that used negligibly affects results for low Z outputs in this method Output impedance of the THAT1646s as deployed in CH1 CH4 tested 57 Q at 1 KHz within THAT
103. ry similar to those of CH1 CH4 its gain is likely near 33 dB after accounting for an estimated 2 dB loss due to loading by the tone control network with its knobs in the neutral positions and 500 KOQ fader pot This suggests that post fader gain the V3B V4 output driver stages is 37 1 dB under the conditions where 70 1 dB overall pre transformer channel gain was measured above paragraph The two post fader triode stages are ina variable feedback loop and the minimum feedback setting increases gain by 7 dB compared to the design feedback level see Section 5 17 guessing again if this minimum feedback setting causes 2 dB gain reduction compared with open loop gain then the estimated open loop gain of the V3B V4 stages would be 37 1 7 2 46 1 dB While modeling the output impedance of the V4 stage Section 5 20 also calculated its open loop voltage gain as 9 5 dB Therefore the estimated open loop gain of the V3B stage is 46 1 9 5 36 6 which is within the reasonable range for a 12AX7 triode stage absent feedback note also that this value is a focal point for accumulated errors in this estimate rich analysis As noted in the previous section the flattest frequency response is obtained when the feedback knob is between 1 and 2 o clock i e 1 30 just clockwise of the design feedback level mark At that setting relative to 0 dB at 1 KHz frequency response measured at the balanced output terminated by 600 Q was 3
104. s can deliver 8 dB higher amplitude to CH2 s balanced input than it can to CH1 before clipping at 1 KHz This is discussed further in Section 5 12 In general the dependence of CH1 CH4 s buffered output clip threshold on load impedance requires careful consideration when adjusting the clip alert indicators as addressed in the next section CORSET mga a eS Oo O oar i SSeS e HEHH E He Hee EE E HE Je E ae EE ie E HE maar J T00 i ses Ok nae RESISTANCE u J Figure 20 Threshold for clipping of 1 KHz sine waveform at CH4 s balanced output operating in differential black or single ended red mode CH1 CH3 performance should be identical Thresholds judged using oscilloscope and RMS outputs measured on H P 331A s voltmeter Small data points represent measurements using resistors as loads and the two large black data points represent the transformer balanced input of CH1 or CH2 as load and are plotted using their impedance at 1 KHz see Figure 15 5 10 Channels 1 4 Understanding and Adjusting Clip Alerts As described in Section 4 9 the clip alert circuits do not work by directly detecting clipping at the balanced line driver outputs Instead they are threshold detectors monitoring the driver chips inputs Since the output clipping threshold depends on load and whether the output mode is differential or single ended see previous section meaningful use of the clip alerts require
105. s included in a book preview available at http books google com At first was excited to read the manual s VU meter termination values because these could be used to deduce Altec s evaluation of the final triode stage s impedance in parallel with the load reflected on the line transformer if present This might help confirm my own figures given in Section 5 20 However for resistances used in the VU meter network R35 R39 in original schematic see Appendix called R21 R25 in Figure 14 there is no single 6CG7 output network impedance value that satisfies both 3450 ohms termination for the 0 VU setting and 61 4150 ohms at 12 VU the former predicts 7 45 KQ and the latter a less likely 71 5 KQ Possibly the original Altec 1567A manual made an error in one or both reported meter termination impedances If the 6CG7 stage s output impedance is 2 0 KQ see Section 5 20 and the load impedance reflected to the transformer s primary is the nominal 15 KQ the output network impedance is 1 76 KQ in that case VU meter termination is 3150 Q 19 2 under the target 3900 Q for the 0 VU setting and 3950 Q 1 3 over target for 12 VU 6 Appendix Original Altec 1567A Schematic The original Altec 1567A manual and schematic is available online from AnalogRules com http www analogrules com manuals altec1 html For your convenience the original schematic is reproduced on the next page page 62 62 RECOR
106. s measured under normal operating load To compensate the first filter voltage divider network resistor R1 is 4 7 K versus the original 2 2 K so that B for V4 and all B voltages down the line matches the original design In addition to R1 s greater resistance the modification also uses somewhat higher value filter capacitors 13 throughout the network and indeed an additional R C stage see below than does the original power supply This results in less ripple hum in the B lines compared to the stock unit i gt 303V be B for Y4 Z fa i gt 283V a tw B for Y3 ce R4 R5 5 C3 47uF 2714 zZ 450V 2584 T 4 7K 22K B for V2B gt Fed a 1724 0 5 172W Se ce 4 uF 22uF w 450V 3504 A at Ux Re 7Y stk 2u amp tN gt 8584 i UK 22K B for V2A STANDBY SWITCH O 1veuw c7 SHOWN IN STANDBY T 22uF POSITION i ie 3504 Vx 258V 22K B for V1B gt 2584 22K B for YIA T 22uF 350V Vx T yx ORIGINAL POWER TRANSFORMER OF ALTEC 1567A CONTROL POT CCW Vx Star Ground for Vintage Unit DC Voltages referred to VX with standby switch R12 25 in normal operating position and all loads normal All RESISTOR Values in ohms Figure 5 Schematic diagram of power supply in the modified Altec 1567A s vintage unit See Figure 4 for the transformer s primary circuit After R1 the rest of the high voltage filter voltage divider network se
107. s their adjustment for the specific output conditions used Prior to shipping the modified Altec 1567A set CH1 CH4 s clip alert thresholds to indicate clipping into floating balanced or differential mode 600 Q load resistors Referring to Figure 20 one can see how this setting gives reasonably accurate about 1 dB performance for balanced loads 470 Q and greater i e for clipping mainly due to the balanced driver s voltage limit 40 However using single ended mode with the same loads would let severe clipping go undetected by the alert circuit Staying in balanced mode but dropping the load impedance to about 200 Q would have the same result Conversely adjusting the clip LEDs to trigger on such lower thresholds would give a false indication of clipping for high impedance balanced loads Given the high amplitude capability of these drivers clipping is not likely to be a concern except when over driving a following stage for distortion effects such as with input transformer saturation experiments CH1 and CH2 are equipped for such experiments since they have pre triode attenuators in their input transformers secondary circuits As shown in Figure 15 CH2 s balanced input impedance exceeds 470 Q between about 33 Hz and 7 5 KHz Within this band clipping due to driver saturation is accurately indicated 1 dB by the associated clip alert LED as presently adjusted for 600 Q loads As explained in Section 5 12 transformer satu
108. some high frequency emphasis and bandwidth extension appears by 9 00 as the pot is turned counter clockwise flattest response in this counter clockwise region must be between 9 00 and 10 00 With the high Z pad switched on left hand side of Figure 23 bandwidth is severely limited when the pre triode attenuator is fully clockwise but improves for all other settings Flattest response is somewhere between 9 00 and 10 30 Compared to the results with the 46 high Z pad turned off the high frequency emphasis at 9 00 is greater 3 4 dB emphasis at 27 5 KHz with pad on versus 1 2 dB emphasis at 26 1 KHz with pad off 4 F Ss X amp oN x e S pA tree Tet Do pea jt penay Ale _ i i kad ladda bd _ __ _ _ 4 4 4 4 i f T t CETTINGS EXAMINED EXCEPT FULL CCW Figure 23 Effect of CH1 s pre triode attenuator and high Z pad setting on measured frequency response relative to response at 1 KHz 0 dB Same conditions and method as used in Figure 19 except frequencies lt 1 KHz were not examined and various pre triode attenuator settings plus both high Z pad settings were evaluated Response curves are at top diagrams at bottom show the pre triode attenuator settings examined when high Z pad was on left or off right In the latter case red settings correspond to red response curves which highlight the appro
109. t for only about 0 2 uV of the 17 9 uV figure The observed driver output noise is thus 8 dB larger than expected from THAT Corporation s specs Much of this difference is likely due to a conservative estimate of voltmeter noise bandwidth so actual driver noise is probably between 103 and 95 dBV i e worse than in laboratory conditions but better than the figure am reporting With channel fader full clockwise 0 dB attenuation the output noise reading was 565 UVawms corrected to 561 UVams due to voltmeter self noise l Il assume the triode gain circuits noise lies entirely within the audio band which is reasonable given Figure 19 As amplified by the output driver triode circuit noise is then simply the square root of the difference between 561 Vrms squared and the broad band driver noise 275 Vrms Squared or 489 UVeams 66 2 dBV Accounting for 5 5 dB output driver gain when working into 600 Q triode circuit noise at the top of the fader is 260 UVams 71 7 dBV Compared to output driver noise triode noise is so large that the driver noise is not significant unless fader attenuation is between infinity and about 30 dB as explained next Subtracting channel gain see Section 5 5 from the 66 2 dBV output noise figure gives the equivalent input noise EIN referred to the balanced inputs gain is 65 5 dB so EIN 131 7 dBV for the unbalanced inputs of CH1 and CH2 gain excludes the 25 dB input transformer step up making EIN
110. t rises above the 1 KHz reference response by no more than 4 dB This high frequency emphasis flattens out as fader settings increase beyond 18 dB to 12 dB This frequency response data uses CH1 CH4 s solid state buffered outputs the output driver circuits impart virtually no load on the fader pot wipers Results will vary when the high Z unbalanced outputs are used depending on the impedance of the device driven and the length of the shielded patch cable used The next section discusses the impedance of the unbalanced outputs and how it depends on fader setting 37 5 8 Channels1 4 Unbalanced Output Impedance and Applications At the triode operating point used in CH1 CH4 the dynamic plate resistance Rp is about 87 5 KQ for these common cathode circuits with cathode bypass source impedance at the plate is Rp in parallel with the 220 KQ plate load resistor or 62 6 KQ With a 250 KO fader pot set full clockwise 0 dB attenuation output impedance for the unbalanced line is 50 KQ 62 6 KQ in parallel with 250 KQ As the fader is rotated counter clockwise visualize the wiper as dividing the 250 KQ pot resistor into top clockwise most and bottom portions Output impedance becomes the bottom resistance in parallel with the series combination of the top resistance and 62 6 KQO Attenuation relative to the full clockwise setting in dB is 20 times the logarithm of the fraction bottom resistance divi
111. t stage s output amplitude This chart is aligned vertically with the SNR chart so that SNR curves at full clockwise fader align with the triode circuit output amplitude as emphasized by the arrows the 5 5 dB vertical offset equals the output driver s gain see Sections 5 5 and 5 9 Distortion was measured on an isolated breadboard version of CH1 CH4 s triode circuit at 1 KHz using the H P 331A distortion analyzer The B amp K Precision 3011B function generator that was used for test signals throughout this project delivered 0 6 distortion with its 1 KHz sine waveform output therefore 0 6 was subtracted from all raw distortion readings before stating distortion percentages in this report including this chart Example Using 30 dB fader attenuation say the RMS signal amplitude across a 600 Q load is 0 5 dBV this means the output would be 30 5 dBV at full clockwise fader if the output driver stage could achieve that without clipping which it can t Nevertheless the SNR curves labeled 30 5 dBV apply in this case Thus the SNR is between about 93 dB and 95 5 dB depending on the driver noise evaluation used the triode stage output is 25 dBV and harmonic distortion is about 2 5 percent 52 High Z Unbalanced Output at 0 dB Attenuation dBV 110 o o 90 80 N Signal to Noise Ratio SNR dB 10 15 Harmonic Distortion Percent 60 20 50
112. t up exactly at the clip threshold only when driving floating 600 ohm loads This occurs at a very high RMS output level of 25 5 dBV 53 volts peak to peak In normal operation it should only be a concern when driving another channel to transformer saturation However dependency of accurate clip indication on the load is an issue users must bear in mind see Sections 4 9 5 9 and 5 10 3 7 Input Output Polarity For balanced inputs and outputs all XLR jacks are wired according to the modern standard of pin 1 ground pin 2 or hot and pin 3 or cold Available on CH1 CH2 and CH5 the unbalanced inputs 1 4 inch jacks have the same polarity as XLR pins 2 hot However all five unbalanced outputs 1 4 inch jacks match the polarity of XLR pins 3 i e inverted with respect to the other inputs and outputs The rationale is given in Sections 4 8 and 4 10 3 8 Mechanical Layout and Power Supply To wrap up this overview will mention some infrastructural features that are not directly in the signal paths and hence are not block diagrammed in Figure 3 The rebuilt vintage unit and the auxiliary panel of the modified Altec 1567A see Figure 1 are permanently married both electronically and mechanically It was logical to locate the auxiliary panel s power supply on the far right and to put a single main power switch for the entire assembly there This keeps the line AC circuitry as physically 11 c
113. th is 1 57f will assume fo 3 MHz for the H P 331A voltmeter which is conservative because error is rated only 5 percent within 0 45 dB for 5 Hz to 3 MHz the actual fe must be higher However its effective filter characteristic could be above first order with greater than 6 dB per octave roll off l II assume it is first order without evidence other than to say it would have been harder for the instrument s designers to achieve flat response given a higher order filter characteristic nor should it be necessary to build in such response If my function generator could go a couple octaves beyond its 2 MHz limit would test rather than assume Thus conservatively call the voltmeter s noise bandwidth 4 7 MHz A 4 7 MHz bandwidth for the voltmeter allows expressing the output driver s 275 uV noise figure as 127 nV per root Hz noise density so the driver circuit s observed noise voltage is 17 9 Vrms 95 dBV in the 5 Hz to 20KHz band The THAT1646 s published output noise spec balanced 600 Q load 0 Q source 18 V supply 22 Hz to 20 KHz bandwidth is 101 dBu which is 6 9 UVams Or 103 dBV The 5 Hz versus 22 Hz lower bandwidth boundary is insignificant Also the difference in source is effectively small the modified Altec 1567A s unity gain OPA2604 op amp driving the THAT 1646 is rated very low noise at 10 nV per root Hz translating to 1 5 UVams for the audio band uncorrelated with the driver s noise this would accoun
114. ude reaches a limit when additional triode distortion becomes intolerable and or driver stage clipping occurs On the left hand side of Figure 25 SNR is plotted against fader setting for five maximum i e fader fully clockwise output amplitudes in 10 dB increments beginning with 0 5 dBV Actual output amplitude is less than the curves labeled values by the amount of fader attenuation Output driver clipping conditions for the single ended and differential output modes are indicated by red shading Solid curves embody my conservative driver noise evaluation while the dashed curves are based on THAT Corporation s published figure actual SNR is probably between these curves Harmonic distortion due to the triode stage is shown in the right hand graph which is lined up with the SNR chart to compensate for the driver s gain 5 5 dB Figure 25 is designed to show the trade off between SNR and triode distortion For example 90 dB SNR requires at least one percent harmonic distortion 100 dB SNR requires at least four percent and so on Similarly for CH1 CH4 s unbalanced outputs SNR and its relationship to harmonic distortion is given in Figure 26 This chart assumes a very high load impedance connected to these outputs Clipping due to the diode clamps protecting the solid state output drivers see Sections 4 8 and 5 8 is indicated in red For any given triode stage output amplitude the fader setting has little effect on SNR The very slig
115. uilt on two perf boards one for CH1 CH2 one for CH3 CH4 which are mounted to an aluminum support plate adjacent to the output jacks on the auxiliary panel The support plate doubles as a shield to help isolate these outputs from the adjacent low Z input jacks A photo of the installed boards and a close up of the CH1 CH2 board is shown in Figure 13 After noting some general aspects of these boards will discuss the driver circuit later in the present section and then the clip alert indicator in Section 4 9 The output driver circuit uses 18 V power supply rails and the clip alert circuit 15 V rails Electrolytic 10 uF capacitors C1 and C2 bypass the 18V rails at power supply connections to each board as do C10 C11 at the 15V connections The DC supply pins of each IC package are bypassed with 0 1 uF capacitors e g C5 and C6 for U2 For minimum inductance at these bypasses stacked film capacitors are used with absolute minimum lead lengths the capacitor and IC pins share the same perf board hole at each IC s power connections For quietest and most stable performance the balanced output driver circuits occupy areas of the perf boards that were prepared with solid ground planes Since this meticulous construction technique may be unfamiliar will summarize it First the component layout is carefully planned for minimum point to point interconnection distance beneath the board Second using 0 1 inch grid graph paper an actual si
116. uipped for this grounding method twist lock tabs securing them to their metal mounting flanges doubled as their common negative terminals and provided convenient lugs for nearby circuits other ground connections Note in original schematic how of the two 60 uF capacitors in the high voltage supply s bridge network only C17B was in a multi section unit because it s negative terminal was grounded its partner C21 had to be an discrete axial lead device since it floated above ground by about 170 V While this grounding strategy obviously works and simplifies manufacture theoretically it is not ideal and practically mechanical connections to the chassis can corrode or loosen causing long term reliability concerns In contrast an ideal star ground is a single point on the chassis to which all ground connections are made It is the one node whose potential is precisely zero volts by definition and all ground connections are referred to it hence the grounding network is shaped like a star Granted there is resistance in each wire to this node causing local ground potential errors but these can t interact with each other or accumulate as they might in a mesh or chain shaped or even a chassis based grounding network An ideal star ground scheme makes internal ground loops which can couple hum or crosstalk into the audio path impossible Having said all that hasten to state that the modified vintage unit described he
117. w non linearity causes distortion Lowering feedback by turning the control clockwise increases gain while letting the channel s second and third triodes V3B and V4 in Figure 14 add more distortion to the signal Reducing feedback also increases output impedance discussed further in Section 5 20 and reduces bandwidth In short changing the negative feedback level affects all major aspects of amplifier performance Near its 12 0 clock position marked the feedback knob setting resulting in the same feedback loop resistance as that of the vintage Altec 1567A called the design feedback level Using moderate fader settings and a low amplitude 1 KHz sine waveform input measured the changes in channel gain and distortion caused by different feedback settings The total gain range available between the feedback knob s extreme positions is about 11 dB Relative to gain at the design feedback level full counter clockwise subtracts about 4 dB and full clockwise adds about 7 dB Importantly the full clockwise setting is minimum available feedback not zero feedback or open loop The following average increases in total harmonic distortion were measured as feedback was decreased from full counter clockwise to the design feedback level 0 14 from the design feedback level to full clockwise 0 40 These figures should be taken as approximate since it would require more care than used to isolate CH5 s interacting distortion sources in
118. ze template for the ground plane is designed to cover a maximum continuous area it contains holes to give non grounded component pins free access to their intended perf board holes Third 0 005 inch thick copper foil is cut to match the template sanded lightly on the bottom surface and applied to the top of the perf board using heat tolerant epoxy cement Finally during assembly grounded component pins are splayed out and then soldered directly to the ground plane the non grounded pins which pass through the plane are 22 soldered point to point beneath the board Careful attention to layout and construction yields performance rivaling a well designed double sided PCB T TIP HIGH Z V UNBALANCED 184 OUTPUT SLEEVE 8 1 4 INCH FEMALE m l C1 c7 1QuF 1 uF BP z 1 1035y re LOW Z BALANCED OUTPUT R1 CHANNEL FADER LOCATED IN VINTAGE UNIT f GP AX Star Ground for Auxiliary Panel GP Ground Plane for Output Drivers R15 R16 15V 22K 33K lt t WW WWW cig tor THRESHOLD ADI Tee A 15 TURN R4 L R6 TRIM POT R12 470K S cG 10K R11 10K 5 6K R17 R6 R9 MATCHED TO 1 0K WITHIN 0 5 poy CCW 4B 172 TLave c9 Opa Ur SZ Ren Len CLIP ALERT B Bik RS u3A 478K 172 TLa7e u4A H 1uF 172 TLeve cc A Star Ground for Auxiliary Panel CG Local Ground for Clip Alert Circuits ALL RESISTOR VALUES IN OH

Download Pdf Manuals

Related Search

Related Contents

PayPal Name-Value Pair API - 2009 Developer's Guide  Guide d`installation CE6000  De Dietrich DHD585X cooker hood  Black Box Voting - Copyright ©2003  Char-Broil 463247412 Charcoal Grill User Manual  RX Series Real-Time Operating System Start for CubeSuite UM  カタログPDF （7.64MB）  CLUB DES OISEAUX  Philips CD640 User's Manual  RM806 USER`S MANUAL  

Copyright © All rights reserved. Failed to retrieve file