PAS - AudioRegenesis
Contents
1. Swann LAS by George Ronnenkamp Part 2 The PAS Line Stage Pacific Audio Regenesis The line stage serves as the control center for a stereo system providing the means to adjust the volume balance and tone It also provides the gain necessary for any of the selected input sources to adequately drive the power amplifier although with modern input sources it often ends up serving as an active attenuator While the circuitry employed in the amplifier portion of the PAS line stage is not at all unusual the implementation of the tone controls certainly is So unusual in fact that knowledge of how the tone control circuit functions particularly the bass control is key to understanding many of the myths and criticisms surrounding the PAS as a whole There are some important differences between the PAS 2 3 and the PAS 3X in this regard as well Thus before getting into performance testing we ll first have an in depth look at the line stage circuitry especially the tone controls Those less technically inclined may wish to skip ahead to the summary portion of this article Gain Related Circuitry PAS 2 3 3X For simplicity and ease of understanding Fig 1 displays the basic line stage amplifier topology with the feedback filter and tone control circuitry removed The circuit consists simply of two cascaded 12AX7 common cathode capacitively coupled stages Nothing unusual here except perhaps the open loop gain which is rather high2. 0 42dB 20HzZ 20KHz The plot with the pots set as per the assembly manual however has some obvious deviations from the intended flat response measuring 0 65 1 75dB 20HZ 20KHz Contributing factors include that fact that the bass and treble pots in the test amplifier measured 680K and 445K rather than the ideal 750K and 400K values plus there was some deviation from the intended 7 00 o clock knob positioning described in the manual The 1 0 a a a PELER ea AR LE A EEEE AE EE Le a AS EAE a EE EEEE a E a ET ee a EEE ES CEEA ee oe EAA ORS D ere ne eae el 0 5 PEPE Rca EEE A EA MN MM OETI R E R E EIEII Nf EEES EE EE AER ncaa m onr a eee a ee eee Se cee e 95 LOptinized Se ie ae see E 4 0dirr aeaaaee nes ee Cement ene tear Geen a ee aeee aE ee Scenes EE E E E E S E E ie eee eee 100 1K 10K Frequency Fig 28 PAS 2 Line Stage Measured Frequency Response copyright 2015 10 Pacific Audio Regenesis potentiometer shafts are splined which means knob positioning is restricted to the increments allowed by the number of splines In the case of the test amplifier an exact position of 7 00 o clock would not have been possible without slightly rotating the entire potentiometer in the chassis While that could have been done the intent was to test the amplifier as it most likely would have been assembled following the directions in the manual In relative comparison the response depicted by the optimized plot was obtained with t
3. 0 06 E HEE a E L eaS N Ea cont aa MEA A AA ROEA A ARAE CEA PATNA a ELA AE Pa EAS BASA aata drive any power amp of the period The 0 1 distortion level was poteke a e e ee reached at 6 0VRMS out Note that the test results reflect the PAS 2 0 04 seesseenseetsfeseensnensegpensnensnedpecessenaporsseredheesessserssecnanensnenssenssenssensaghivnsenesenenssedparsssrsssenssenssensesnsssnsnsepessnsnsnsnsescsnsneredessesesesesesnsngeenssansnsnsadpesssnscersepessesssesgeeceseenapenseensdhessenes with the tone control knobs installed as per the instructions in the eer eee assembly manual The IMD test was also repeated with the tone 0 03 S a b Lt controls adjusted for optimum flatness using the square method and 111 YF PF ep the results were virtually identical ETE PE sO a Ave eee TEE E eee EE A PAS 2 3 Signal to Noise os Bat BES pT 1 0V 2 0V 5 0V 10V The SN of the test subject PAS 2 was measured and found to be 82 6dB relative to 2VRMS output and 76 9dB relative to 1VRMS Output Voltage RMS Switching in a 400HZz high pass filter to eliminate the hum component Fig 31 PAS 2 Measured IM Distortion improved these figures by about 3 5dB Disclosure Note The SN measurement was performed using an HP 8903 Audio Analyzer which has an input impedance of 100K The 62K load resistor was remplaced with 100K and the 510K load resistor was removed to maintain proper loading at the output of the PAS PAS 2 3 10KHz Square Wave
4. 1K Fig 4 displays the actual measured output impedance vs frequency of the test subject PAS 2 It is interesting to note that Dynaco specifies the OP impedance as 1K While that may be true of the ete Fe ee Pe a amplifier stage it is obviously not true of the line stage 0 400 aK 0K as whole and thus is somewhat misleading The terminating impedance specification 500K or higher is more telling Fig 4 PAS 2 Measured Line Stage Output Impedance VS Frequency Output Impedance Frequency Considering the line stage thus far it would appear that flat low end response would only be achieved with a power amplifier impedance of about 470K That seems like a pretty significant restriction but was it Remember the era in which the PAS was designed when the input impedance of power amplifiers was typically in that range With the 62K resistor predominantly determining the load even if the power amplifier input impedance was 250K the deviation from a flat low end response would less than 1 0dB at 20Hz At this point Dynaco recommends removing the 510K resistor which will restore flat response So in actual fact the PAS could easily drive most power amplifiers of the period It was the introduction of solid state power amplifiers with their much lower input impedance that began posing problems for the PAS Then criticism about poor low frequency response and high output impedance began to flourish and perpetuate eventually being applie
5. EE SEEE Ea EEEa r eea EE eE ES ES EE SEn E EAE ASEET I Frequency Fig 44 LT Spice PAS 2 3 1 2Hz Subsonic Resonant Peak at Output Another modeled response plot was created this time looking at the output of the first amplifier stage Fig 45 displays the response at the plate and after the 02uF coupling capacitor where the peak is seen in more dramatic fashion The peak gain at the plate is about 35dB which exceeds the open loop gain at that point by 5 dB After the coupling capacitor the peak gain is actually about10dB greater than would be expected What s going on here First off that step in the response heading below 100Hz but before the peak is actually expected due to the implementation of the bass control as discussed way back on page one The rise in gain heading down towards the peak exists because the negative feedback can only be sustained as long as the coupling capacitors can pass the frequency of interest Below about 10Hz the negative feedback is losing its grip as the reactance of the coupling capacitors quickly increases and approaching 1Hz is all but gone As well some wild phase shifts occur as the frequency approaches and continues below 1Hz promoting a peak in gain before falling off again as the capacitors lose the ability to pass any signal at all Note that the response at the plate continuing below1Hz flattens since the first stage input is DC coupled After 02UF 77 0 1Hz
6. Response Fig 32 displays the measured 10KHz square wave response of the PAS 2 test amplifier compared to the response shown in the assembly manual The results are very similar The tone control knobs were installed as per the instructions in the manual then set to center for the test A470K load was connected to the output Note that readjusting the tone controls for optimum flat response did not alter shape of the 10KHz waveform only slightly affected the amplitude Dynaco did not specify the amplitude of the square wave output displayed in the manual The PAS 2 measured result was taken at 3V P Poutput PAS 2 Measured Fig 32 PAS 2 10KHz Square Wave Response copyright 2015 12 Pacific Audio Regenesis PAS 3X Computer Simulated Response LT Spice was used to simulate the gain and frequency response of the PAS 3X The model schematic presented in Fig 33 reflects the circuitry with the tone controls set to center which in the case of the 3X means they are bypassed R7 R12 cS 470k 100k Rser 1 By ioe C1 R9 02 18k ug P G vE 004p es R1 Bal Vol T 1500p AC1 47k i i an nh12ax 2meg SINE 0 0 6 ig k 200k nht2ax7 L ac oct 100 20 20k a 3 3meg 33 ae o R13 R14 p R4 R6 47k 50k 1K 4 7k a R RS 4 7k i Fig 33 PAS 3X Line Stage LT Spice Model A frequency response simulation was run covering 20Hz 20KHz with the resulting gain vs frequency plot is displayed in Fig 34 The gain is vir
7. at about 60 dB The OL loop output impedance of about 62K as well as the gain will both be reduced with the application of significant negative feedback Feedback amp Bass Control PAS 2 3 not 3X Fig 2 displays the simplified line stage with the feedback and bass control circuitry added The components in the low frequency feedback path are highlighted in red Note that the 33pF capacitor that normally appears across the 47K FB resistor has been deleted for simplicity since it is not relavant to the tone control analysis The 0075uF capacitors can be considered out of circuit at very low frequencies The 750K bass pot is represented by two fixed resistors that simulate the control knob positioned at center Although it is a linear potentiometer the resistors are not equal in value because the pot is not exactly at physical center for flat response as described in the assembly manual Notice that one side of the potentiometer is in the feedback path while the other side is actually in series with the output of the amplifier Looking at the feedback path we see the 390K portion of the bass pot in series with 47K which together provide 22dB of NFB resulting an amplifier gain of 38dB As frequency increases at about 100Hz the 0075uF capacitors come into play and begin bypassing the potentiometer Above 1KHz effectively only the 47K resistor remains dominant in the feedback path reducing the gain to 20dB So there is a step in gain vs
8. even with a 20K load the response is within 1dB at 20Hz The response above about 120Hz is essentially unaffected by the selected test loads Note that the drop in low end response is primarily due to the reactance of the 1uF output coupling capacitor not the line stage amplifier itself Response dB 1K Frequency Fig 36 PAS 3X Line Stage Measured Loading Effects PAS 3X Harmonic Distortion Dynaco did not publish a THD specification only stating that it was below the measurement capabilities of commercial test equipment of the period The total harmonic distortion of the test subject PAS 3X was measured over the range of 20Hz 20KHz with readings taken at the rated output level of 2VRMS as well as 1VRMS Fig 37 displays the test results Disclosure Note An HP 8903 audio analyzer was employed to measure the harmonic distortion The 8903 has an input impedance of 100K which is within the terminating impedance range specified for the 3X Thus the 62K and 510K internal load resistors were left in place 0 04 sss EESE NEE BETERE ERENER Y PARLE BREEN EARE R Secon EEEE AAEE A AR AE EEE A AERE BESEER ANEA t MERRE EAL 3 CERERE AAE ARE EREE NEEE ARAARA AAAA EENAA AAEE E AANE INE BARETA EENAA ES cose coos EI EA EEE E EE N 0 03 ERE PE ALE ESAE POELE VERE EEEN ORE AS DEA abhi sascha TEE Seda a ssa Sa seb CEREREA CAETERAS PEA ses OEE AEAEE EEEE kin EATA avn taba EEA ERE LEE AEE E CA A 2V RMS pe Q se 0 02 PEPEE ASEE eer Meroe LEA
9. frequency at the unloaded line stage output If one were to consider the simplified line stage concept only as shown if Fig 2 it would seem to be utterly impractical Of what use is a preamplifier with 360K output impedance and non linear response However the output circuit is not quite complete and what happens nextis perhaps the single most important thing to understand about the PAS2 3 Fig 1 Basic Line Stage Topology 470K 0075 0075 Ta T 390K 360K Output Gain Bass Pot Unloaded 12AX7 47K 20Hz 100Hz 1KHz 20KHz Frequency Bass Response Output Unloaded 4 7K Fig 2 Feedback amp Bass Control PAS 2 3 copyright 2015 Pacific Audio Regenesis Unlike most other preamplifiers in order for the PAS 2 3 to deliver flat low end response it must be properly loaded Fig 3 displays the complete LF signal path to the power amplifier The low frequency signal path as well as the load components are highlighted in red Although the treble control is not shown its implementation covered later does not impact the signal path Note the additional 62K and 510K resistors and 470K power amp load typical of Dynaco tube PA s With the bass control set to the flat position the 360K series component of the bass pot and the parallel combination of the 62k 510K and 470K PA load form a voltage divider resulting an an18dB low frequency signal attenuation Recall that previously the unloaded LF gain was about 38dB Now the l
10. occur at about 0 8HZz while the very first release of the PAS 2 with its 02uF output coupling capacitors will present a peak near 5Hz Again with the tone controls centered Advancing the bass control increases the frequency of the peak while reducing the bass lowers the peak frequency The worst case presented by the original PAS 2 results in a peak near 20Hz with the bass pot fully advanced and around 8Hz at half bass boost One might speculate this is the reason Dynaco increased the value of second stage coupling capacitor from 02uF to 22uF fairly early on in PAS 2 production although we may never know for sure While possible effects can be modeled and measured how likely is it that conducive conditions would ever be encountered in actual use In the PAS 2 3 test case where the line stage was already driven undistorted to the point where most power amplifiers would be severely be overloaded it took an additional subsonic peak input Fig 47 LT Spice PAS 2 3 500Hz Overload Distortion level nearly twice as great to induce distortion Furthermore the conditions required to induce distortion occurred with the volume control set to maximum something that would rarely occur with any regularity in the real world At more realistic volume settings subsonic input levels would be attenuated significantly further reducing the likelihood of overload induced distortion In reality the subsonic anomaly is likely more of a curiosity rath
11. the bass control set to the mid position What the load actually consists of however is not immediately apparent Because of the unusual implementation of the bass control which places a portion of the bass pot within the FB loop and a portion in series with the output of the line stage the output must be terminated into a load of about 50K to obtain a flat response The complete load is actually the parallel combination of two internal load resistors 62K on the PCB and 510K on the output jack and the load presented by the power amplifier For example if the power amplifier has an input impedance of 500K then the total load would be the parallel combination of 62K 510K and 500K which turns out to be 49 8K It is apparent that the internal 62K resistor is the dominant factor in determining the overall load provided the PA input impedance is sufficiently high The response was considered to be acceptably flat with power amplifiers having an input impedance greater than 250K As the input impedance of the PA approaches 250K Dynaco recommended snipping out the 510K resistor on the output jack to bring the overall load closer to 50K again So the PAS would have no problems driving most tube power amplifiers of the period Concerns only began to emerge when solid state amplifiers with much lower input impedances began to appear When considering the suitability of the PAS 2 3 for use with any particular power amplifier the first step is to dete
12. the center position the treble pot is effectively out of circuit and the gain remains at 20dB copyright 2015 002 Treble Pot I 02 Fig 14 PAS 2 3 Full Treble Boost 4 7K 002 Treble Pot ip 02 Fig 15 PAS 2 3 Full Treble Cut 4 7K 100K 470K 0 22 0075 i 0075 Vol 12AX7 002 47K 364K Fig 16 Feedback amp Treble Control PAS 3X Treble Pot 6 Pacific Audio Regenesis Fig 17 shows the PAS 3X treble control at full and partial boost with the feedback path highlighted in red At full boost the potentiometer element is still out of circuit due to the center break However the wiper is now connected to the 02uF capacitor placing the 1K resistor in parallel with the 4 7K resistor resulting in 824 ohms Thus the feedback components effectively become 47K and 824 ohms resulting in an amplifier gain of 35dB for a treble boost of 15dB In the partial boost condition some portion of the 36K resistive track of the potentiometer also appears in series with the 1K resistor The gain depends upon the value of that total effective resistance in parallel with the 4 7K resistor Fig 18 displays the treble control under full and partial cut conditions At full cut the wiper is connected to the 002uF capacitor placing the 1K resistor in parallel with the 47K resistor which results in 980 ohms The pot element remains out of circuit due to the center break Thus the effective feedba
13. would have appeared at the output had the additional 1uF output capacitor not been included With DC input levels in this range Dynaco s own ST 70 and Mk IV power amplifiers in which the inputs were DC coupled could be subject to LF instability due to the way in which the input pentode was biased Those who have incorporated the popular tone control bypass modification should take note Instructions on performing this modification appear in various places online and often do not include the additional capacitor Refer to the PAS 3X manual for instructions on installing the additional 1uF output coupling capacitor PAS Line Stage Common Concerns amp Quick Summary This is primarily for those who couldn t or didn t want to slug through all of the previous pages of technical stuff For the most part concerns and criticisms regarding the PAS line stage revolve around output impedance issues This summary will focus mainly on those issues Does the PAS Have High Output Impedance PAS 2 3 is unique in that the output impedance cannot be considered in the normal fashion If one were to just measure the output impedance over the audio range it would be found to be quite high about 50K at the low end of the band but then drops to less than 1K at the high end This characteristic is displayed due to the unusual way the tone controls are implemented particularly the bass control The bass control potentiometer is split within the circuit such tha
14. 1Hz 10Hz 100Hz 1KHz Frequency Fig 45 LT Spice PAS 2 3 1 2Hz Resonant Peak at Output of 1st Stage The concern here is the potential for the second stage in particular to be over driven should subsonic excursions of sufficient level especially in the region of the peak be present at the input Being DC coupled with essentially no NFB the first stage not only exhibits its full OL gain but also produces an additional 5dB at the resonant peak Even after passing through the coupling capacitor the peak level is still 10dB greater than that which would have been presented to the second stage had NFB still been in full effect With the second stage driven into compression by the elevated subsonic signal audible higher frequencies superimposed will also become distorted To demonstrate this effect modified my PAS 2 3 computer model to include a second subsonic signal generator and another 47K resistor to form a simple mixer at the input The model schematic is shown in Fig 46 The subsonic generator was set to 1 2Hz that being the maximum peak frequency while the audible generator was set to 500HZ Initially the level of the 1 2Hz signal was set to 0 and the 500HZz signal to a level 0 38VRMS which produced 2VRMS at the output Enough to fully drive almost any power amplifier but not so high as to overload the preamp Then the 1 2Hz level was increased until the 500Hz signal began to appear distorted This occurred as the 1 2Hz input leve
15. Bal 68K PAS 3X 270K PAS 2 3 33K PAS 3X Fig 20 displays the relevant blend circuitry which includes 47K source isolation resistors and a selection of blend resistors Note that the selector switch actually has two Put Right more positions which are not shown since they relate to mono select functions rather than blend There are four blend settings ranging from full stereo to full mono The switch is shown in the first position where the blend resistors are essentially out of circuit This is the normal full stereo position When set to the second position the series combination of the resistors bridges the two inputs resulting in a portion of Blend SW a a the signal from one channel appearing at the other In the third position only one resistor bridges the inputs and the blend effect becomes PAS 2 3 stronger In the fourth position both channels are tied directly together resulting in a full mono condition Note that the blend resistors differ in value between the PAS 2 3 and 3X and significantly so resulting in the PAS 3X having more aggressive blend between the full stereo and mono positions Fig 21 shows the effective blend from one channel to the other in the various switch positions The original intent of the blend feature as described in the owners manual was to reduce the hole in the middle effect which characterizes some types of stereophonic source material Later however Dynaco also had something else in mi
16. EESO AAEN reel POR CEA AATA A ASE A AAE EDREAL AN EI ALLIE EA EETA ALAIA E E omens neal lean LEASE AAAA Riera PELAA AAAA Serre tera EAEE ARE EAEE eer imam pees Meth BESE AELE AEEA 1K Frequency Fig 37 PAS 3X Harmonic Distortion Notice that the distortion at the low end of the band does not rise as it does on the PAS 2 This is primarily due to the increased negative feedback that occurs at low frequencies as a result of the bass pot being bypassed Bypassing the treble pot however does not affect the feedback at higher frequencies and the rise in THD atthe top end of the band remains Overall the PAS 3X exhibits extremely low harmonic distortion for a vintage tube preamplifier basically confirming Dynaco s claim that it was below the measurement capabilities of test equipment of the period copyright 2015 14 Pacific Audio Regenesis PAS 3X Intermodulation Distortion 60Hz 7KHz 0 10 Dynaco specified the IM distortion of the PAS 3X as being less than 0 05 at 2 volts sufficient output to drive any amplifier The IM 0 08 distortion of the PAS 3X test subject was measured at numerous output levels starting from 0 5VRMS and increased to the point where the IMD reached 0 1 The results are displayed in Fig 38 The IMD at 2VRMS was 040 into a 470K load This would certainly be more than enough to drive any power amp of the period IMD The 0 1 distortion level was reached at6 5RMS out 0 04 PAS 3X Signal to Noise The S
17. N of the test PAS 2 was measured and found to be 81 2dB relative to 2VRMS output and 77 3dB relative to 1VRMS Switching in a 400Hz high pass filter to eliminate the hum component improved these figures by about 3 5dB Disclosure Note The HP 8903 audio analyzer used to measure the SN has a 100K input impedance Since this is within the minimum recommended drive limits for the PAS 3X the internal load resistors were left in place PAS 3X 10KHz Square Wave Response Fig 39 displays the measured 10KHz square wave response of the PAS 3X test amplifier compared to the response shown in the assembly manual The results are very similar The tone control knobs were installed as per the instructions in the manual and set to center for the test A 470K load was connected to the output Unlike the PAS 2 no attempt was made to adjust the tone controls for optimum flat response since they are effectively bypassed at the the center position Dynaco did not specify the amplitude of the square wave output displayed in the manual The PAS 3X test subject was measured at 3V P Poutput PAS 2 3 3X HF Stability with Capacitive Loading Square wave testing thus far was performed with the PAS terminated into a 470K resistive load directly at the output connector However in actual use the PAS will be connected to a power amplifier through a length of coaxial cable which along with the PA itself will incur some capacitive loading A 470pF capacitor
18. PAS 2 3 3X Subsonic Resonance This is a subject am somewhat apprehensive to bring up as it may invoke unnecessary alarm for a scenario that is unlikely to manifest in typical use Nevertheless it s a phenomenon that technically exists and thus should be disclosed in any serious analysis of the PAS line stage circuitry am referring to a resonant response peak in the 0 5Hz 5Hz range that could theoretically result in audible distortion It is difficult to characterize the response at such low frequencies using traditional test equipment and procedures so it is not surprising that this particular PAS phenomenon remains relatively unknown It was through computer modeling that first became aware of it While the relatively modest peak as it appears at the output seems rather innocuous there are potentially some overload and distortion implications associated with it We ll examine this phenomenon in more detail using the PAS 2 3 circuit as a test case Fig 44 displays the frequency response from 0 1Hz 1KHz as generated by my LT Spice model The model schematic is the same at that indicated in Fig 26 Aresponse peak of about 4dB relative to 1KHz occurs at approximately 1 2Hz As far as an actual measured result on the test PAS 2 was able to confirm a similar response peak in that general region using a function generator and oscilloscope did not attempt to produce a precise plot since my most accurate test equipment for that purp
19. ain correct loading at the output copyright 2015 11 Pacific Audio Regenesis eovecsvencececssecssescccesesesefoscecesesesovescscesee ocnccctiBpascsosel ovccseccsesesefeosecssescadpeoseccsssfocseoseedpoceseee seconsessceeossesesessseccsousesessseccsoesosesssenssoeGpscsesessecesossscsccsesssosessefonsccsosesseocsescseendpescsscaseessosedpecseecssonesel envecsseccedposvenssongpeessvensfoccevendbesssesceressvessseeseussovensseavesesovescueaseecueven dpesseccsoseccuecsecssoseccseoseeGpesescsessecosesscossoe ecessccsoossocoselecceccssesesedhsenccoeseceljosoeccoesefocoseosedpsesesee ocerevesssesceseoesessssessesesesssecosecsecccsecese THD 1K Frequency Fig 30 PAS 2 Measured Harmonic Distortion With most power amplifiers requiring much less than 2VRMS drive for full output it can be seen that the PAS 2 3 would be well under 0 03 THD in any typical application This is excellent performance for any vintage tube preamp PAS 2 3 Intermodulation Distortion 60Hz 7KHz Dynaco specifies the IM distortion of the PAS2 3 as being less than 0 05 at sufficient level to drive any amplifier The IM distortion of the PAS 2 test subject was measured at numerous output levels starting oe r Egi from 0 5VRMS and increasing to the point where the IMD reached 0 07 Sn an aa Ls 5 m ENE sn ia Ed 2i an EOR 0 1 The results are displayed in Fig 31 The IMD at 2VRMS was Ze ef o FP 046 into 3 470K load This would certainly be more than enough to S
20. always understood however is how it comes about The bold red lines in Fig 50 highlight the signal path and components related to the DC offset The origin is the 1 45VDC bias voltage seen at the cathode of the first amplifier stage From that point the DC path can be followed through the 47K feedback resistor the 750K bass pot and the parallel combination of the 62K and 510K output load resistors The series combination of the FB resistor and bass pot and parallel output load resistors forms a voltage divider at the output jack thus reducing the DC voltage to 94mV This of course represents the idealized case where the bias voltage and all related component values are exactly as indicated in the schematic Note that any changes made to the values of the output load resistors as may be required to maintain proper loading for various non Dynaco power amplifiers will also change the level of the DC seen at the output 7 94mVDC Fig 50 Path for DC at Output The fact that DC was allowed to appear at the output at all is rather curious The earlier Dynaco PAM 1 mono preamp was very similar in design but it did include an additional capacitor at the output to block DC Yet up until the introduction of the PAS 3X where blocking became an absolute necessity DC was allowed to appear at the output In the case of the 3X with the bass pot copyright 2015 19 Pacific Audio Regenesis effectively bypassed at center over 800mVDC
21. by ignoring them copyright 2015 5 Max PAS 2 3X Bass Boost Cut dB 1 2 PAS 3X 1 2 PAS 2 3 D x AA OU gt Led o x lt Bona W AAHH Frequency Fig 11 PAS 2 3X Measured Bass Control Response 47K i Fig 12 PAS 2 3 3X High Frequency Feedback Path Treble Pot 36K T Fig 13 Feedback amp Treble Control PAS 2 3 Pacific Audio Regenesis Fig 14 displays the treble control in the full boost position The full 400K potentiometer resistance is now effectively in parallel with the 47K resistor resulting in about 42K We won t include the 1K resistor since its influence is small relative to 400K The 1K resistor is in parallel with the 4 7K resistor resulting in 824 ohms Thus the feedback network is now effectively 42K and 824 ohms resulting in an amplifier gain of 34dB Relative to the center position gain of 20dB there is now a treble boost of 14dB Note that had a direct connection been used instead of the 1K resistor the 02uF capacitor would have appeared across the 4 7K resistor As frequency increases a point will come where the reactance of the capacitor approaches zero and all feedback is effectively removed as the cathode of the first stage becomes fully bypassed Fig 15 shows the treble control in the full cut position The 1K resistor is effectively in parallel with the 47K resistor resulting in 980 ohms The full 400K pote
22. circuitry highlighted in red may appear somewhat confusing at first Two fixed resistors represent the potentiometer set to the center position and note that the values are far from being equal That s because unlike the bass pot the treble pot is not linear rather it has a reverse log characteristic Notice that the potentiometer is essentially in parallel with the 47K and 4 7K resistors Furthermore it represents virtually the same ratio as the 47K and 4 7K resistors Thus the effective feedback ratio remains the same and the gain is still 20dB The 1K resistor limits the maximum boost cut with the control set to the extremes Without it at high frequencies the amplifier would be running with either no feedback or 100 feedback at the min max limits That will become apparent shortly Note Given the location of the 1K resistor the most precise examination of the treble control circuitry would involve numerous wye delta transformations That s a whole lot of ugly calculations Fortunately we can still achieve a very good understanding of how the treble control operates with the more simple approach employed in the following analysis During the following analysis under boost and cut conditions we will continue to consider the capacitors as short circuit Even though they will exhibit some influence as the control extremes are approached it will not be great and the analysis will be significantly easier to understand
23. ck components are now 980 ohms and 4 7K producing an amplifier gain of about 1 6dB Relative to the 20dB flat position there is now atreble cut of 18 4dB In partial cut conditions some portion of the 364K potentiometer track appears in series with the 1K resistor Thus the gain will be related to the effective resistance in parallel with the 47K resistor Under partial boost or cut conditions the PAS 3X treble pot is in parallel with only one of the 47K or 4 7K feedback resistors while the PAS 2 3 pot impacts both Under full boost or cut conditions unlike the PAS 2 3 the 3X pot element is an open circuit However the difference at the min max positions is minimal since the 400K pot resistance in the case of the PAS 2 3 is large enough to have only small influence While the response at full boost cut conditions is very similar between the PAS 2 3 and 3X there are some significant differences in between Fig 19 illustrates the actual measured response of the test subject PAS 2 and PAS 3X with the treble control set at the 1 2 and maximum boost cut positions The maximum boost cut results have been averaged since they were nearly identical At the 1 2 boost position the PAS 2 response has increased significantly while the 3X not very much At 1 2 cut the PAS 2 and 3X are quite similar but less effective than one might expect And so under boost conditions the PAS 2 is most responsive within the first half of the cont
24. d universally even when the PAS was driving the very tube amplifiers for which it was designed Understanding the load requirement is the key here With simple resistor changes the PAS 2 3 can be matched to power amplifiers with input impedances as low as 50K Bypassing the tone controls a popular modification allows the PAS to drive virtually any power amplifier We ll be looking at this and other modifications in detail in a separate article copyright 2015 2 Pacific Audio Regenesis Since we re already looking at the bass control we ll continue on with how the boost cut operates Although the PAS tone controls are considered to be of the active type albeit unusual look upon the bass control as more of a hybrid One side the active part operates within the feedback loop while the other side serves as a passive voltage divider Fig 5 displays the bass control in the full boost position The full potentiometer value 750K is now in 750K the feedback loop greatly reducing the NFB and results in epee an amplifier gain of 40dB Also all of the series resistance has been removed and the load is now connected directly to the amplifier stage via the 0 22uF capacitor Thus thereisa 20dB bass boost relative to the flat position 470K Vol Fig 6 displays the bass control set to the full cut position 4 7K The 47K resistor is predominant in the feedback path Fig 5 PAS 2 3 Bass Full Boost increasing the NFB to maximum resultin
25. e the series combination of the capacitors around the 18K resistor have a negligible effect on the response given the impedances involved Frequency Response Evaluation With a basic understanding of the circuitry involved we ll now delve into the actual performance of the PAS 2 3 and 3X line stage including both computer modeled and laboratory measured results PAS 2 3 Computer Simulated Response LT Spice was used to simulate the gain and frequency response of the PAS 2 3 The model schematic is presented in Fig 26 The triode employed in the model was the NH12AX7 downloaded from 470K 12AX7 1500pF Filter Switch Fig 24 Scratch Filter PAS 2 3 3X Filter Response dB Frequency Fig 25 Measured PAS 2 Filter Response Duncan s Amp Pages www duncanamps com Potentiometers are represented by fixed resistors with Volume set to maximum and all others set to reflect mid position Loudness compensation is disabled Note that since neither of the TC potentiometers is at actual physical center as described in the user manual and the treble potentiometer is a log type the representative resistors in the model reflect my calculation of what would result in a flat response This also assumes the ideal case where the potentiometers are exactly to spec that being 750K for the bass and 400K for the treble pot In reality there is a tolerance of 20 on the actual potentiometers the affect of which will also be covered fu
26. er than a concern In any case should there be a problem what could be done Although the intent of this particular article is PAS performance not modification will pass along a suggestion that comes by way of Dave Gillespie with whom was fortunate to be able to consult during the writing of this series Dave is well known for his outstanding technical articles and projects posted on popular forums especially Audiokarma org This modification can largely mitigate potential problems and consists of simply reducing the value of the first stage coupling capacitor from 02uF to 0033uF The effect of the modification which modeled after the first stage coupling capacitor Fig 48 PAS 2 Measured 500Hz Overload Distortion copyright 2015 18 Pacific Audio Regenesis can be easily seen in the stock modified plots displayed in Fig 49 The peak is actually still there but has been shifted up to about 3Hz and reduced to the point of being difficult to discern in the PAS 2 3 scenario It is more easily seen in the PAS 3X since the low end response step is not present But in either case effectively about 10dB of additional headroom has been added at the input to the second stage significantly reducing the possibility of overload In repeating the mixed signal lab test on my modified PAS 2 test subject now at 3Hz the new peak and 500HZ it was found that the level of the 3Hz signal had to be increased to 2 1VRMS from 0 7VRMS orig
27. evident In contrast Fig 42 shows the response with no capacitive loading In keeping with the extended more linear high frequency response without the 33pF FB cap a very impressive square wave is presented However it is not sustainable in the real world where at least some capacitive loading will exist The 33pF feedback capacitor value would seem to provide the best compromise between high end response and stability over the range of reasonably expected load capacitance Y gee PAS 2 with 470pF capacitive load at output PAS 2 with no capacitive load at output and 33pF feedback capacitor removed and 33pF feedback capacitor removed Fig 41 Fig 42 As well again with no 33pF FB capacitor extended HF response sweeps were performed on the PAS 2 test subject This revealed an expected response peak the frequency and amplitude being dependant upon the amount of capacitive load In the case of a 470pF a peak of about 5dB occurred near 120KHz followed by a rapid drop off in response The modeled result displayed in Fig 43 shows the response from 1KHz 500KHz for both no capacitance and 470pF load capacitance conditions The PAS 3X also exhibited a peak of about the same amplitude but slightly lower in frequency The actual measured results were not precisely plotted but did correlate well with the model No Capacitance 10KHz 100KHz Frequency Fig 43 LT Spice Model PAS 2 3 Response 33pF FB Capacitor Removed
28. g in an amplifier Vv gain of 20dB But now the total resistance of the bass pot is in series with the load forming a maximum loss voltage divider Thus the output at the load is reduced about 23dB from the flat response setting if the load is 50K For any boost cut setting in between the min and max limits the gain 470K will be determined by the combination of the portion of the pot that appears in the feedback path vs the portion of that appears in series with the load Vol 750K Feedback amp Bass Control PAS 3X l Bass Pot Now lets examine how the 3X differs in its implementation of the bass control Fig 7 shows the simplified line stage and bass control with the feedback path highlighted in red The unusual potentiometer is obvious but also note the additional 1uF capacitor The capacitor has nothing to do ai with the bass control function rather it prevents DC Fig 6 PAS 2 3 Bass Full Cut Originating at the cathode of the first stage from appearing at the output More on that later The special potentiometer consists of a 750K resistive track plus additional conductive tracks on either side and dual wipers When set to the center position as shown the wipers straddle the resistive element shorting it out effectively removing the bass pot from the circuit In the feedback path the 47K resistor is predominant resulting in an amplifier gain of 20dB Unlike the PAS 2 3 no portion of the pot remains i
29. he knob center positions being determined by instrumentation This resulted in a knob position of just past 6 00 o clock when set fully counter clockwise Note that while this position was optimum for this particular amplifier it may not be for others When restoring a PAS 2 or 3 typically use the square wave method to determine the best center position for the tone controls for that particular amplifier however that will not be an option for most Also note that the amount of pot rotation involved between the optimized and per manual positions is quite small so relatively small movements can quite easily result in 1db or more deviation from a flat response If you regularly use the tone controls this may be of no consequence However if you desire the response to be absolutely as flat as possible it may be an issue since without a signal generator and oscilloscope there is no way to really be sure The popular tone control bypass modification is of course one solution PAS 2 3 Loading Effects We have already discussed the requirement of proper loading if flat low end response is to be maintained For interest the response of the test subject PAS 2 was measured with some less than optimum PA loads The results shown in Fig 29 display the response at 31 points from 20Hz 20KHZ relative to 1KHz with the tone controls optimized The optimum 470K load result is also displayed for reference Note that 250K is the point at which Dynaco
30. inally before similar distortion appeared That s a 9 5dB improvement basically confirming the modeled results Modified 0033uF 0 1Hz 1Hz 10Hz 100Hz 1KHz Frequency Fig 49 LT Spice PAS 2 3 1st Stage Response Stock and Modified In concluding this section will emphasize that it is NOT my intent to raise alarm or to suggest that all PAS preamps must be modified to prevent possible distortion due to this anomaly Only to inform about its existence and possible side effects that might occur if only in theory The simple modification described here is provided as a possible solution should you have reason to believe it may be a problem in some particular case or you wish to improve your PAS if only in theory However this is more likely a case of the old a solution in search of a problem scenario Also be aware that the source itself may have issues if it is presenting significant subsonic output levels The only actual recommendation will make is directed at owners of the original release of the PAS 2 If upon checking the value of the second stage coupling capacitor you find it to be O2uF would recommend increasing it to 22uF as is found in later PAS 2 3 and 3X versions Be sure not to increase the value of the first stage coupling capacitor which has always been 02uF PAS 2 3 not 3X DC at Output It is quite well known that a small DC offset of about 100mV exists at the output jacks of the PAS 2 3 What is not
31. l approached 0 7VRMS Fig 47 displays the modeled output where the 500Hz signal can be seen superimposed upon the notably higher 1 2Hz level Note the section where the 500Hz signal appears to be compressed During this time it is actually being clipped on the top side copyright 2015 17 Pacific Audio Regenesis v1 R7 R12 330 aii 100k Rser 1 c4 c1 R9 u2 22 c5 c6 n 18k i 0075p 0075p c2 R1 ME RR 004p TA R15 R16 345k E 41k R11 nh12ax a 415k pat 500H ii R3 nh12ax7 1 2meg a z 750k 200k a oF R20 R21 470k nae lt 7 3 3meg c3 R19 62k 510k R4 RE R14 33p V ae R23 4 7k 50k 41k R18 375k a F v RS v3 41k T A Se A Tk 1 2Hz a 7 NZ Fig 46 LT Spice PAS 2 3 Two Signal Subsonic Overload Distortion Model Next the test subject PAS 2 was set up using the same test frequencies and input levels employed in the model The signals were applied to the Spare L R inputs and mixed via the A B blend function with the output measured at the left channel The input to the right channel was disconnected at the blend switch so that the model circuit was properly duplicated i e not loaded by the RC volume and balance pots The output scope trace shown in Fig 48 looks much like the modeled result It should noted that this scenario represents the PAS 2 3 condition with the tone controls centered Other versions of the PAS will present somewhat differently as will altering the position of the tone controls For the PAS 3X the peak will
32. n series with the output and thus there is no low frequency gain step The output impedance of the amplifier is also much lower due to the significant negative feedback and now the 1uF capacitor outside the FB loop along with the total effective load impedance determines the low frequency cutoff With a PA load of 20K for example the response would be down about 1dB at 20Hz So the PAS 3X could drive power amplifiers of much lower impedance Dynaco specified a terminating impedance of 100K or higher which is perhaps a bit conservative l l l l l 750K 1 0 i Bass Pot l l l l l l l l 62K 510K 100K or higher 47K Power Amplifier 4 7K Fig 7 Feedback amp Bass Control PAS 3X copyright 2015 3 Pacific Audio Regenesis Fig 8 displays the actual measured output impedance of the test subject PAS 3X For reference the dashed line shows the impedance of the PAS 2 3 At 20Hz the output impedance is about 5K almost 10 times less than the PAS 2 3 and drops quickly to less than 1K as frequency increases The rise in low frequency impedance is predominantly due to the reactance of the 1uF capacitor relative to the load impedance Output Impedance Note The 62K and 510K load resistors are not really needed with the bass pot centered and effectively PE EEL PoP eg mee erect out of circuit However they do play a role in the off 0 400 aK OK center response particularly in bass cut conditions Fre
33. nd That something was the introduction of their Derived Center Channel System which we will not get into in this article However the blend resistor values were altered in the PAS 3X to be more compatible with that system 750K d _ Fig 20 Blend Function PAS 2 3 3X Bal 250K PAS 3X 0 20 40 100 Fig 21 Blend vs Switch Position PAS 2 3 3X Loudness Compensation The human ear does not respond the entire frequency range at the same loudness level As volume changes so does the ears response to frequency At decreased z Volume volume levels the listening experience may be lt Pot characterized by a significant loss of bass response koadiess accompanied by a slight loss of treble A loudness circuit Switch provides a measure of compensation for this effect by typically boosting the low end of the audio band as an 4 7K inverse function of the volume control setting Since the loss of treble response is much less it is also less common to see high frequency compensation employed within loudness circuits Interestingly Dynaco did include it in their PAM 1 mono preamplifier but only low end compensation appeared in the PAS 1 2 Vol The PAS employs what is perhaps the most commonly used method which involves a tapped volume pot tap typically 10 resistance The circuit configuration is displayed in Fig 22 An RC network in parallel with the tap on the volume pot creates a low frequenc
34. nd Well as those begin to come online various pole interactions occur resulting inthat dip around 117Hz However the small simulated aberration is quite minor compared to the response deviation that could actually occur due to the way the tone control center positions are determined as per the manual 20 0 17 0 100Hz 1KHz 10KHz Frequency Fig 27 PAS 2 3 Line Stage LT Spice Model Gain vs Frequency PAS 2 3 not 3X Measured Frequency Response The test subject PAS 2 was measured at 31 points over a frequency range of 20Hz 20KHZ into a 470K load resistor The output was set to the rated 2VRMS level The test consisted of two response runs the first with the tone control Knobs positioned as described in the assembly manual That being setting the pots to the fully counter clockwise then installing the knobs such that they point to the 7 00 o clock position The controls were then set to the front panel indicated center position for the test run For the second test run the controls were optimized for best response using an oscilloscope and 50Hz 1KHz square waves The response plots relative to 1KHz are displayed in Fig 28 For reference the modeled result is also plotted on the response graph The measured 1KHz gain was 18 9dB nearly identical to the modeled result It can be seen that the modeled and measured response with the optimized tone control settings were in close agreement The measured optimum response was 0 06
35. ng the PAS to a SS power amplifier without regard to impedance typically results in some loss of bass response The severity of the loss depends directly on the input impedance of the particular power amplifier The PAS line stage amplifier itself actually does have the potential to drive virtually any power amplifier by virtue of the large amount of NFB employed and resulting low output impedance It is the unusual implementation of the bass control that negates that potential low output impedance from being seen at the output jacks If the tone controls could be dispensed with the PAS immediately becomes compatible with many more SS power amplifiers The popular tone control bypass modification in its most basic implementation improves the drive capability considerably but can be taken further by simply removing the 62K and 510K internal load resistors They are no longer required with the bass pot bypassed and only serve to unnecessarily burden the line stage The PAS 3X automatically bypasses the tone controls at the center position however the 62K and 510K load resistors remain in place for proper operation off center The resistors could also be removed but the bass controls may not respond as intended should you decide to use them Before condemning the PAS for its inability to drive low impedance loads in stock form it should be remembered that it was never intended to do so It is a vintage piece that worked well with tube power amplifier
36. ntiometer resistance is in parallel with the 4 7K resistor resulting in about 4 6K The effective feedback circuit is now 980 ohms and 4 6K resulting in an amplifier gain of 1 7dB Relative to the center position gain of 20dB there is nowa 18 3dB treble cut Again had the 1K resistor not been present the 002uF capacitor would appear across the 47K resistor As frequency increased at some point the 47K resistor would be bypassed completely and the feedback would approach 100 For any position in between center and the full boost cut limits the gain will be determined by the portion of the potentiometer resistance that appears across the 4 K and 4 7K feedback resistors Feedback amp Treble Control PAS 3X Now lets examine how the 3X differs in its implementation of the treble control Fig 16 illustrates the simplified line stage with the treble control added The feedback path is highlighted in red Again we see a rather special potentiometer The resistive track is essentially cut into two separate non connected halves When set to the center position as shown the wiper does not contact either of them Highly unusual to say the least The two resistor values representing the two halves still maintain the same greatly offset values displayed by the PAS 2 3 treble pot at center Since the wiper is floating no portion of the treble pot appears in parallel with the 47K and 4 7K feedback resistors Thus when set to
37. oaded gain is 38db 18dB 20dB That low end step has been removed and the response is flat over the whole audio range at the specified 20dB gain And so the proper load for a PAS 2 3 is the parallel equivalent of 62K 510K and 470K which works out to be about 50K Remember that number 50K But what about the output impedance Isn t it still too high Well in fact the output impedance of the PAS 2 3 is frequency dependant At low frequencies the output impedance is the parallel combination of the 62K and 510K load resistors in parallel again with the impedance seen looking back into the line 470K 0075 0075 Vol 390K 360K rs l l Bass Pot l l 62K 510K I 470K 47K i l l l i Power Amplifier DETE E E l 4 7K Fig 3 PAS Line Stage Loaded zid NZ Gain stage Looking back we see that the 360K portion of 0dB the bass pot is predominant the output impedance of itii 1KHz 20KHz the amplifier itself being quite low due to the significant Frequency Response Loaded negative feedback Thus at the low end of the audio band the output impedance becomes 62K 510K and 360K in parallel or about 48K However as the frequency increases the 00 75uF capacitors come into play and begin bypassing the bass pot The output impedance of the amplifier becomes more dominant rather than the bass pot and internal load resistors At 20KHz the output impedance of the amplifier is almost totally dominant and with high NFB becomes less than
38. ose does not extend below 10Hz In and of itself one might tend to dismiss this as just a curious design anomaly however looking a little closer at what s going on within the feedback loop it becomes more apparent that it has at least the potential to be not quite so innocuous copyright 2015 16 Pacific Audio Regenesis errr rrrrrrrr errr rir ee Ce Le LLL CL CL ee ee ce ey a si Sa iS de sin Loos in csi sin do et in its oa ci gt A ee Be ee ee eee ee ae ee ee ee E a eee ee ee Se ee ae seed eee ee ee a ae as Se aes ee Pe eee Spar lar Wey PRE ae Le Pte a ee RUS ey ae ere See ge res T See ot ie A dB 5 1 1 ot 1 i i 1 1 i ot p p 1 oot i g g oot 0 0 Aea aea bere Mace rial a r a a e ae ee ee a i ance hare en oe a fo imine bane aara ea raas eeka oa ee Neal a ed ei or ee rir ee eae ee i bri ania ern Pa ea fede Peas ono e 7 We ie rat cat a net oa ad lee a S ae ee at i eas Seams ee rad 5 re a a a a a a a a a a a a a a a lt ee a oaaae aea a e a D a RO Gee ee ee ee et et ee et ee i i ee ee et ee et ee et Se eee eee See ee Se ae wn Fw a in tm mm em em em ee te ete ete ee 20 0 hee cfhn 222 ne ten ene cnn acc en cen de en de een deena noses ecenecesedeweseneenen ESE e aS
39. quency Fig 8 PAS 3X Measured Line Stage Output Impedance VS Frequency Fig 9 illustrates the PAS 3X bass control at full and partial boost with the feedback path highlighted in red At full boost both wipers reside on the conductive track of the potentiometer and the full 750K resistance is added to the 47K feedback resistor The resulting gain is about 40dB providing a 20dB boost over the 20dB flat position gain Like the PAS 2 3 at full boost there is no series resistance so the output impedance remains low 0075 0075 Bass In the partial boost condition one wiper remains on the aa Ld 470K conductive track while the other shorts out a portion of the 750K ue resistive track The reduced portion of the potentiometer ae remaining in the feedback path increases the NFB reducing 47K Load the gain thus lowering the boost Unlike the PAS 2 3 there is still no series resistance so the output impedance remains Fi low oe ee Partial Boost Fig 10 shows the bass control under cut conditions In full 4 7K cut the 47K resistor is predominant in the feedback path resulting in an amplifier gain of 20dB However all of the 750K resistive element is now in series with the load creating a voltage divider and thus maximum bass cut The magnitude of the cut is like the PAS 2 3 load dependant and the internal load resistors 62K 510K are back in play Given the same PA load 470K the PAS 3x will exhibit the same bass c
40. recommends removing the 510K resistor located at the output jack Although not shown on this graph doing so will return the response to a state very similar to the 470K plot Also note that the PAS 2 3 is capable of properly driving 100K loads by suitably altering the values of the internal 62K and 510K load resistors No Load Relative Gain dB 1K Frequency Fig 29 PAS 2 Line Stage Measured Loading Effects PAS 2 3 Harmonic Distortion The PAS was often advertised as the distortionless preamplifier and indeed it pushed the limits of commercial measurement equipment of the period Thus Dynaco did not publish a THD specification Given the significant amount of negative feedback it employed that s not too surprising The total harmonic distortion of the test subject PAS 2 was measured over the range of 20Hz 20KHz Readings were taken at the rated output level of 2VRMS as well as 1VRMS Fig 30 displays the test results Note that the results depict the PAS 2 with the tone controls knobs installed as described in the assembly manual THD testing was also performed with the tone controls set for optimum flat response using the square wave method and the results were nearly identical within the bounds of test error Disclosure Note The HP 8903 audio analyzer employed to measure the harmonic distortion has an input impedance of 100K Thus the 62K internal load resistor in the PAS was changed to 100K and the 510K removed to maint
41. rmine the input impedance of the power amplifier If it is above 250K the PAS is suitable as is If less than 250K you can alter the values of the 62K and or 510K internal resistors such that the parallel combination of the internal resistor load and the PA load is close to 50K Theoretically the lowest power amplifier load impedance you could drive would be about 50K with both internal resistors removed from the circuit With PA input impedances less than that loss of bass response begins to occur The PAS 3X is somewhat different With its special tone controls effectively out of circuit when set to center it does not have the same 50K overall load requirement as the PAS 2 3 With increased NFB and no bass pot in series with the output the 3X exhibits a much lower output impedance at low frequencies than the PAS 2 3 Although Dynaco specified a minimum PA impedance of 100K the 3X can drive loads of 50K with little loss of bass response Still in stock form it could not properly drive all power amplifiers that appeared as SS took hold of the market copyright 2015 20 Pacific Audio Regenesis It has been claimed that the PAS is not well suited for use with SS power amplifiers Is this true If you have read the previous paragraphs concerning output impedance and load matching you will understand how solid state power amplifiers with input impedances much lower than typical tube amplifiers could present a challenge to the PAS Connecti
42. roblems It should be noted however that implementing the tone control bypass modification which bypasses the bass pot may result in over 800mVDC appearing at the output This level of DC applied to the inputs of the ST 70 and MK IV in particular may result in LF instability Although not always mentioned in online instructions for performing the tone control bypass an additional 1uF output coupling capacitor should also be installed as was applied in the PAS 3X to prevent such a high level of DC from appearing at the output This concludes the line stage analysis In the final installment of this series we ll have a close look at the power supply section of the PAS preamplifier As with the phono and line stages over the years there has been no shortage of opinion and controversy over its implementation and operation copyright 2015 21 Pacific Audio Regenesis
43. rol rotation while the 3X is most responsive in the last half Under cut conditions both are similar with the last half of the rotation being the most responsive Treble Boost Cut dB It s interesting to note some anomalies in Dynaco s published tone control response specifications Their stated specifications for the PAS 2 3 and 3X are 20dB at 20Hz and 14dB at 20KHz However that does not quite match up with the response plots provided in the PAS 2 3 manual which look more like 20 23dB at 20Hz and 14 20dB at 20KHz j 02 Partial Boost 36K Fig 17 PAS 3X Treble Boost aa ve 1K Full Cut 4 7K me Partial Cut 02 w l Fig 18 PAS 3X Treble Cut SS ee eee oe Max PAS 2 3X OK 20K Frequency Fig 19 PAS 2 3X Measured Treble Control Response Besides the tone controls the PAS has other sound altering features including blend loudness compensation and filtering We ll take a look at how each of these operate copyright 2015 T Pacific Audio Regenesis Blend 47K i i l t Left The PAS incorporates some unique features in regards to P stereo and mono configuration Asix position Stereo Mono rotary switch provides the means to progressively blend mix a portion of one channel to the other plus several ways to select mono sources We will only examine the blend feature here since the mono options are just basic Blend Switch switching functions 750K 4 560K PAS 2 3
44. rther on v1 R7 330 w 470k Rser 1 S C1 R9 02 18k U1 E c2 c3 R1 Bal Vol ie 1500p aiii V2 AC 1 47k R11 1 2me SINE 0 0 6 200k nh12ax7 g 750k er ac oct 100 20 20k 3 3meg S S R13 R6 day 50k 1K 4 7K ee EE Pa 4 7K N R12 100k C4 U2 224 c5 00754 R15 nh12ax7 C7 415k 002 R17 1k c8 R14 33p 47k R18 375k Fig 26 PAS 2 3 Line Stage LT Spice Model copyright 2015 9 C6 0075p R16 cout gt 335k R20 R21 R22 510k R19 62k 470k Load 25K V VA V c9 Pacific Audio Regenesis A frequency response simulation was run covering 20Hz 20KHz with the resulting gain vs frequency plot is displayed in Fig 27 The line stage gain is a maximum of 18 9dB at 20Hz and a minimum of 18 2 dB at 117Hz The gain at 1KHz is 18 7dB Relative to 1KHz the response could be stated as 0 2 0 5dB 20Hz 2OKHz Note that the specified gain of the line stage is 20dB so where did the extra dB go Well in fact the gain of the line stage amplifier is just slightly over 20dB relative to the input level at the grid of the first stage Taken as whole however the 47K source isolation resistors located on the blend selector switch along with the balance and volume pots form a voltage divider which slightly reduces the actual overall gain Of note in this plot is the 0 5dB dip in response at about 117Hz Remember all those capacitors we ignored to make the analysis of the tone control circuits easier to understa
45. s of the period Does the PAS require extremely short interconnects to the power amplifier No This silly assertion appears on online forums from time to time have even seen claims warning that interconnects should be no longer than two feet The pretext is that because of the high output impedance high frequencies will be attenuated due the capacitance of the cables The fact is for the PAS 2 3 the output impedance is only high at the low end of the audio band At higher frequencies the output impedance drops dramatically to the point where interconnect cable capacitance is not a significant issue especially in the lengths typically employed have performed response tests using 8 ft cables of mediocre quality that showed negligible attenuation at 20KHz The PAS 3X presents even less of a concern since it exhibits a lower output impedance over an even broader frequency range DC is present at the output jacks Is this a concern A small DC offset does appear at the output jacks of the PAS 2 3 not 3X It originates at the cathode of the first amplifier stage and appears at the output by way of the 47K feedback resistor and bass pot This path in combination with the 62K and 510K internal load resistors forms a voltage divider resulting in about 100mVDC appearing at the output jacks This is somewhat curious since most of Dynaco s own tube power amps of the period were also DC coupled at the inputs however did not seem to present any p
46. t one portion resides within the feedback loop reducing the NFB while the other appears in series with the output of the line stage This results in a rather high output impedance at the low end of the band However the bass pot is also bypassed with capacitors which is the route taken by higher frequencies and thus the output impedance drops as frequency increases Although such a significantly variable output impedance over frequency would seem to be a problem it actually isn t as long as the PAS 2 3 is driving a relatively high impedance load within an expected range Such loads were presented by Dynaco as well as most other tube power amplifiers of the period The PAS 3X however is different due to its custom tone control potentiometers which when centered are effectively out of the circuit In particular the bass pot is bypassed and thus there is no series resistance to the load plus increased NFB at the low end of the band And so the 3X exhibits much lower overall output impedance that is also much more consistent over frequency Typically about 5K at 20Hz and less than 1K above 100Hz meaning the PAS 3X can drive much lower load impedances than the PAS 2 3 Is it true that the PAS requires a specific output load For the PAS 2 3 in particular this is true in the technical sense but is also widely misunderstood A fairly specific load is in fact required in order for the response below 1KHz to be as flat as possible with
47. the 1 2 and maximum boost cut positions with a 470K PA load Note that the MAX curves are the averaged response for the PAS 2 and 3X which were nearly identical The 1 2 response curves represent the bass control set half way between center and min max rotation Note the significant difference in the amount of boost and cut at the 1 2 rotation setting While the PAS 2 approximates half of the maximum available boost or cut the PAS 3X is already near the limits Virtually all of the response happens within the first half of the rotational travel As well the special 3X bass potentiometer has less overall travel to begin with than 20 the conventional PAS 2 potentiometer Thus the PAS 3X bass control will be much more sensitive to adjust It 20 is also interesting to note that the PAS 2 3 manuals included tone control response plots which are conspicuously absent in the 3X manual 20 10 PAS Treble Control Now we ll look at the treble controls Fig 12 displays the simplified line stage with both treble and bass controls absent The 33pF FB capacitor has been omitted for simplicity The high frequency feedback path highlighted in red is common to the PAS 2 3 and 3X The capacitors can be considered short circuit and the 47K and 4 7K resistors determine the amplifier gain of 20dB Feedback amp Treble Control PAS 2 3 not 3X Fig 13 illustrates the simplified line stage with the treble control included The feedback
48. tually flat at 18 26dB over most of the audio range dropping slightly at the extremes Relative to 1KHz the response could be stated as 0 0 23dB 20Hz 20KHz As with the PAS 2 3 the gain is slightly less than the specified 20dB due to the 47K source isolation resistors associated with the blend switch The gain relative to the grid of the first stage is 20 49dB Gain dB 100Hz 1KHz 10KHz Frequency Fig 34 PAS 3X Line Stage LT Spice Model Gain vs Frequency PAS 3X Measured Frequency Response The measured frequency response was 0 0 36dB over a frequency range of 20Hz 20KHz at 2VRMS out into a 470K load Fig 35 displays the results plotted at 31 points over the test frequency range For reference the modeled result is also plotted on the response graph The measured vs modeled results track very closely with some divergence beginning at about 5KHz and reaching a maximum of 0 12 dB at 2O0KHz Still extremely close overall The gain at 1KHz was 17 96dB relative to the input jack and 20 56dB relative to the grid of the first stage 1 0 m c O Q Q a4 100 1K 10K Frequency Fig 35 PAS 3X Line Stage Measured Response copyright 2015 13 Pacific Audio Regenesis PAS 3X Loading Effects Loads ranging from 100K to 20K were connected to the output and the response measured from 20Hz 20KHz The results are displayed in Fig 36 Compared to the PAS 2 3 the loading effects are much less severe and
49. ut as the PAS 2 3 or about 23dB 0075 0075 Bass When partial cut is employed however unlike the PAS 2 3 wer ey ae Fig 9 PAS 3X Bass Boost the amount of NFB does not change One wiper continues to 750K 1 0 reside on the conductive track keeping the 47K resistor as Full Cut the determining gain factor The other wiper shorts out part ITK Load of the resistive track resulting in a lesser voltage division and thus less bass cut a And so in the PAS 3X for incremental boost conditions only O i Partial Cut the NFB is altered while for cut conditions only the amount 4 7K of resistance in series with the load is altered In the PAS 2 3 itis a combination of both At the point of maximum boost cut the PAS 2 3 and 3X are essentially equivalent however there are significant differences in the response in between For the purpose of comparison to the test subject PAS 2 I enlisted my daily driver PAS 3X In terms of disclosure the 3X has been restored not modified and retains the original circuit configuration and component values Basically a new PAS 3X The line stage components in the stock test subject PAS 2 are original and measured within specifications The Telefunken tubes from the PAS 2 were installed in the PAS 3X for this and all subsequent comparative performance testing Fig 10 PAS 3X Bass Cut copyright 2015 4 Pacific Audio Regenesis Fig 11 illustrates the measured bass response at
50. was added to the output of the PAS 2 test subject in parallel with the 470K load This is representative of a particular 3 ft length of low quality cable had at hand which exhibits rather high capacitance The 10KHz square wave response displayed in Fig 40 still looks reasonable with just a bit of well damped overshoot A better quality interconnect of similar length with a capacitance of around 100pF will show little difference from the unloaded response Various capacitive loads up to 01uF were applied with no tendency for instability or oscillation at any time To further explore the margin of stability the 33pF feedback capacitor was removed from the PAS 2 test amplifier and various capacitive loads applied There was no tendency to outright oscillation at any time however square wave response was affected as you would expect Fig 41 displays the 10KHz response with the 4 70pF load capacitance connected More copyright 2015 15 SE eS SE enn Sn ee On i Se See Se ee Se So one en Cet Ste Se et oe Se oe ee ee E O Se Se ee one Seen eS Set See ee ee ee Se Ce ST Sn a Se oe orn SE Re See Se ee ee On ee Se ee Se a or i one E S T O S E O E E E E E E E 0 5V 4 0V 2 0V 5 0v 10V Output Voltage RMS Fig 38 PAS 3X Measured IM Distortion PAS 3X Measured Fig 39 PAS 3X 10KHz Square Wave Response Fig 40 PAS 2 with 470pFadditional load at output Pacific Audio Regenesis prominent overshoot and some ringing are now
51. y boost in response when the switch is set to the open position As the wiper is moved farther away from the tap point louder the parallel circuit has less and less effect resulting in 3 4 Vol volume sensitive compensation Fig 23 shows the measured response of the PAS 2 test subject with the volume set to the 1 2 and 3 4 positions The PAS 3X was not measured but employs exactly the same circuit Loudness Comp dB Frequency Fig 23 Measured Loudness Compensation PAS 2 copyright 2015 8 Pacific Audio Regenesis Scratch Filter The PAS also includes a switchable low pass scratch filter to reduce ticks pops hiss and other high frequency noise The circuit displayed in Fig 24 is located between the amplifier stages The related components are highlighted in red When the switch is closed the capacitors have a path to ground and a 2 pole low pass filter is formed The first pole consists of the output impedance of the first amplifier stage and the 1500pF capacitor and the second consists of the 18K resistor and 0O4uF capacitor Fig 25 illustrates the filter response as measured on the test subject PAS 2 The 3X was not measured but employs the exact same circuit The response is down 3dB at about 6 5KHz beyond which it continues rolling off at about 16dB per octave When the filter switch is open the 3 3M resistor isolates the1500pF and 004uF capacitors from the ground path and the filter is disabled In this cas
Download Pdf Manuals
Related Search