The Quartz Crystal Anomaly Detector
Contents
1. 5 00 L second measure 4 50 L ments 4 00 3 50 520 5 560 5 600 5 640 5 680 5 720 5 760 5 800 5 840 5 880 5 920 Time Days SEQAVG Freq Jump Detection Window Length 3000 Window Offset 0 Jump Threshold 1 918866e 09 3 000 sigma Jumps 1 Figure 25 Region of Largest Jump Date 08 31 09 Time 14 45 42 Dato Points 384000 thru 398000 of 2237172 Tau 9 9614700e 01 File qcad 66 001 Figure 26 shows the FREQUENCY DATA region around day Region of Popcorn Noise 3 50 where popcorn noise type frequ ency 3 00 excursions were observed Other 250b tip NM ne oo o crystals have shown this kind of behavior more clearly which seems to be assoc iated with excessive crystal drive level Frequency pp10 s b gt in 0 00 O SONO E E O O E N E G EE bundles 0 20 495 4 439 4 452 4 466 4 480 4 494 4 507 4 521 4 535 4 548 4 562 4 576 4 58 Time Days Figure 26 Region of Popcorn Noise 10 F T 10 Allan Deviation Date 08 31 09 Time 12 37 16 ESE 9 96e 01 1 99e4 O00 3 98e 00 7 97e 00 1 59e4 01 5 19e4 01 6 38e 01 1 28e4 02 2 9064 D2 5 10e4 02 1 0264 03 2 04e 03 4 08e 03 8 16e4 03 1 65e404 3 26e 04 5 5 5e 4 04 1 351e4 05 2 bles 2 617e 11 ipeni r DOM dS O OTOT ET T 4d Y WIDE ee E EET 2 72e 11 bn Mee e es pencio Debbie Melslleaiemecel Rhe nals Ses lne Trenes iy eid2. 2 Versions Hamilton Technical Services QCAD RF Amplifier Revio O i 121509 1 Figure A1 2 QCAD RF Amplifier Schematic Figure A1 3 Input Return Loss Figure Al 4 Output Return Loss S11 2 37 dB VSWR 1 03 at 13 4 MHz S22 48dB VSWR 1 01 at 13 4 MHz Figure Al 5 Forward Gain Figure Al 6 Reverse Isolation S21 12 2 dB at 13 4 MHz S12 70 dB at 13 4 MHz File The Quartz Crystal Anomaly Detector doc W J Riley Hamilton Technical Services Revision D December 23 2009
3. Block Diagram The single channel QCAD system is implemented on two 3 8 x 2 5 circuit boards as shown in the block diagram of Figure 7 The detector board comprises the crystal and crystal network its oven oven temperature controller and set point DAC and the AM detector preamplifier and ADC The processor board contains the DDS RF source the signal processor and the RS 232 PC interface Detector Section Crystal Under Test Heater Monitor Signal Monitor 13 4 MHz 1 dBm M i ii Scope Signal Sync Monitor Monitor LEDs ISP Control Pushbuttons 12 V Power 3 3V Power Figure 7 QCAD System Block Diagram Minimum Impedance Frequency The QCAD system tunes the DDS to the minimum impedance frequency of the crystal under test That frequency fm is slightly below the series resonant frequency f as shown in the impedance circle of Figure 10 and impedance plot of Figure 9 3 6 For the 13 4 MHz crystals described in Table 1 the minimum impedance frequency is lower than the series resonant frequency by about 2 1x10 The QCAD system is not particularly concerned with the absolute crystal frequency but rather its stability Quartz Crystal Resonator Impedance Circle Crystal Impedance f Impedance Reactance Resistance Inductive Lo Ri RERO CC f 21 2zNL C where C load capacitance Co f X 0 f amp Ccy2C 0 f t c 2Cg Impedan
4. TC Plot amp Axis 30 00 to 65 00 deg C 4 38 deg Cdiv TC Plat r Axis 1 93040e 07 to 1 67456e 0 3 735e 07 div Figure 33 20 C Hour Scan Setup Crystal Resistance The detector voltage at the center of the resonance provides information about the crystal resistance as shown in Figure 35 for crystal resistance values of 10 20 and 30 ohms It is therefore be possible to estimate the minimum impedance resistance of the resonator by detector voltage measurements made at the resonance center with normal RF level and off resonance at a reduced RF level that produces the same detector voltage The crystal resistance estimate is given by ko e 9 W1 W2 PerkinElmer QCAD System TC Data for Crystal S N Bliley 040 2 0 ECT Ces VN ER ECHO NR ee oe lini eee M DEN VUA DEM UMS PE NOR OO ee FOR e evo go EDEN ES 6 Relative Frequency pp10 200 000m 160 000m 120 000m 80 000m 40 000m 0 000m 7345M v 4 V Figure 35 Temperaure C Figure 34 Scan Results at 20 C Hour Micro Cap 8 Evaluation Version qcaddiodedetector cir R4 10 30 13 452M F Hz Crystal Resonance vs Resistance Characteristic where 72 ohms is the RF source impedance seen by the shunt crystal W1 is the normal RF amplitude word and W2 is the off resonance reduced RF amplitude word This resistance estimate is independent of the actual detector and preamplifier transfer functions de
5. A 7 7627e 4 B 2 2953e 4 C 7 4431e 8 lu Sense 1 1 1 1S a3 Oven Temp C Bridge VDC DAC VDC DAC Code 1u M3F122 Heater Demand Current R55 75 mA Each Heater 47 5k 4 5 Matts Total 25V Current R33 R34 R35 C84 R56 Sense 1 10 1 lu 2 49k Mets Q4 X Oven 6lu mJFi22 Heater HTRRET Current R40 R41 R42 Sense 1 10 1 Hamilton Technical Services QCAD Oven Section Soca 52672009 Figure 14 QCAD Oven Temperature Controller Schematic Processor Hardware The QCAD processor board uses a Microchip Technology PIC16F648A single chip micro controller to process the discriminator signal from the detector board to control an Analog Devices AD9852 48 bit direct digital synthesizer DDS and to implement an RS 232 user interface via a MAX3235E level converter The DDS and PIC are both externally clocked at 20 MHz from a 10 MHz reference via a discrete tuned circuit push push frequency doubler and MCIOOLVELI6 ECL and LMV7219 5V comparators The processor operates at 5 volts the DDS at 3 3 volts and a 74LCV244A logic level converter is used between them The microprocessor flash program memory can be reprogrammed via an in circuit programming connector The DDS uses an internal x6 clock multiplier and is followed by a 7 pole 15 MHz elliptic LPF The DDS spectrum is quite clean having no visible spurious components except for some 72 dBc noise around the carrier as shown in Figure 15 The QCAD processor al
6. T Read E Hs Close Figure 22 QCAD Configuration Screen Read or change QCAD configuration then press Close Performance The performance of the QCAD system has met its requirements and exceeded its expectations In particular the noise level and corresponding short and medium term stability is much better than needed to detect significant frequency jumps for the GPS RAFS application The temperature and drive level sensitivity of the crystal are even larger and are even more easily resolved As expected the most critical aspect is crystal temperature stability which requires good oven temperature control especially if the measurements are made at a fixed temperature rather than close to the crystal turnover point Little variation has been seen in the level of short term instability between different crystals but the system could certainly detect a noisy device Although the DC level of the AM detector provides information about the crystal resistance no attempt has been made to quantify that parameter with this system At a minimum doing so would require compensating for the RF drive level and diode detector temperature dependence Network analyzer and bridge methods are better suited for that purpose The objective here was to detect and quantify frequency jumps as a function of time temperature and drive level If as is likely there is a corresponding resistance change it doesn t really
7. a Windows user PerkinElmer QO4D Communications Program A interface that can configure and COM1 Opened trol th t d t th HCAD System SN 1 contro e system and capture e Oven Setpoint Temperature 25 0 C 1 HF Attenuation 0 0 dB NA data as shown in Figure Data Stream Interval 0 996147 seconds Figure 21 Main Screen of Windows User Interface Table 2 QCAD Processor Commands Cmd Type Bytes Description Cmd Type Bytes Description 0 Command 1 Reset PIC processor I Write 2 Enter temporary DAC setting 1 Read 1 Get PIC S W revision K Write 2 Enter oven set point temperature 2 Read 4 Get HF DDS sub word L Write 4 Enter LF DDS sub word 3 Read 4 Get LF DDS sub word M Write 2 Enter modulation rate 4 Read 2 Get modulation rate N Command 1 Tune down or set DAC to 0 5 Read 1 Get DAC MS bit O Command 1 Select open loop operation 6 Read 3 Get DAC word P Read 1 Get servo state 2 Read 1 Get PC data stream rate R Write 1 Enter PC data stream rate 8 Read 1 Get servo gain S Command 1 Save all settings in EEPROM 9 Read 2 Get average detector voltage T Read 2 Get DDS amplitude A Read 2 Get peak detector voltage U Command 1 Tune up or set DAC to FS B Write 1 Set DAC MS bit V Write 2 Enter DDS amplitude C Command 1 Select closed loop operation W Write 1 Set EEPROM state D Read 2 Get discriminator error Read 1 Get system S N E Read 1 Get EEPROM state Read 1 Get oven set point temperature F Write 1 Enter QCAD
8. are brief C Liat Filename scan frd Cenfiaure EasyPlat Start lasting only a couple of seconds Plot Ereg and within the dynamic range of the Plot TG tt Points 63146 Hels Deis 6 temperature controller The resulting frequency changes are Current integrated by the relatively large start 30 00 C End 65 00 E 5 00 L thermal mass of the oven and Initial i Scan Hun become cyclic fluctuations on the D yell 0 00 Baie 2 0 LHE Tine 17 50 Hrs order of 2x10 so that even the un averaged frequency record is State smooth Temperature scan Scan Progress Figure 30 shows the QCAD temperature scan screen after the completion of a scan with resulting red TC data and superimposed green cubic fit A EasyPlot plot of the TC data 1s shown in Figure 31 and an excerpt from the data file 1s shown in Figure 32 which includes the cubic fit coefficients and TC Plot Y Ass 4 6027 1e OF to 2 39249e 06 3 6b4e 0 diy derived crystal parameters 9 TC Plot Axis 30 00 to 65 00 deg C 4 38 deg Cdiv Figure 30 QCAD Temperature Scan Screen PerkinElmer QCAD System TC Data for Crystal S N Bliley 040 Crystal S N Bliley 040 Start Temperature 30 00 End Temperature 65 00 Scan Rate 2 00 Tau 9 961472e 01 30 00 2 044269e 06 1 009 64 98 2 392494e 06 1 010 Plot crystal TC cubic fit coefficients Give Botsl rtid2 rt nqdd3 L 3 a0 7 164359e 07 2 al 3
9. j 35 ligu 2 3 6 4 7u 10V VVVV 5 31 rF Signal Monitor Ri4 Ci BNC 30k Bias Adj RI6 A7D Converter 1 k RF nput 1 25 C8 A 5y 0 01 dde m 6 dB Pad Xtal Network Ed Detector Network voc T p Di MMBD4148 12 bits R6 R8 R9 C3 3 RI 1 k 8 9V FS VQ lu gt RLS AD 7457 22 1 22 1 lu x amp mV 1 k Dc ATS 13 401 MHz 50 v 5 _ Amplitude dip 7 7 R12 Serial aDC Dac 338956 001 AE mi R3 ead 7 i min Z freq C4 RIL 56 2K mU L k Interface LW 200 Hz lu 1 k d D2 To PIC MMBD4148 R13 Microcontroller j 3 1k u Ci V lu 158 Quen Temperature D A Converter TEMPREF gt YOuT SCLK 12 bits AD5320 DIN Sy FS Note DAC and thermistor bridge referenced to 5V supply SYNC QCAD Crystal Section Figure 13 QCAD Detector Schematic A Su R45 A tl2u 5 10k Thermistors cea R46 R47 2 R48 R49 T Balu V 4 Cantherm MF5S2C15 3F395a 50 kohm 7 1 at 25C R50 TEMPREF 3 90k From Temperature Set DAC Thermistor Bridge Amplifier D3 MMBD4148 Set Thermal Gain R53 61 9k Demand Pouer Limiter Atl2v cia Voltage Controlled 8 lu Current Sources Cit al Quen 1 U4 a 4 V R17 QQk 2RI8 1 08k 1 R22 Current R19 R28 R21 Sense 180 108 18 13 Q2 Quen 0 001u miga Heater HTRSUP TP1 cas lu to 5V l2 bits 10 mC Resolution 25 to 65C Range Steinhart Hart Model Coefficients lt E B Current R26 R27 R28
10. matter because an anomaly has occurred and its severity is judged by the size of the frequency change Example An example of a I month QCAD measurement run is shown in the figures below This run resulted in about 2 6 million t 1 second data points Figure 23 shows the system configuration for this run which was conducted between 08 05 2009 and 09 05 2009 on Bliley Crystal S N 040 at its upper turnover temperature of 51 C and maximum drive power of about 8 dBm The absolute frequency at fm the crystal s minimum impedance point is the average to the two DDS words 13 400342325 MHz about 1 002 kHz below its nominal operating value in the RAFS QCAD 1 Configuration S W Rev 0 22 UTC 2009 09 05 13 03 14 MJD 55079 543912 Crystal S N B040 50C OdB Oven Temperature Setpoint 0347 50 99 Deg C DDS HF Word 1C966DA0D16F 13 400442325084470 MHz LF Word 1C9651AA8AO0E 13 400242325084548 MHz Deviation 1BF64761 200 0000 Hz Amplitude Maximum Data Stream Stream Rate 13 Tau 0 9961 sec Frequency 6DAOD1 5 851635e 05 Servo Mod Rate E000 19 0735 Hz Gain 08 2 x 256 Loop Closed EEPROM Read No Detector Voltage 6948 1 028 VDC Signal 8718 28 3 Error 0001 Status 2 Green Closed Loop Small Error OK Signal DAC MS Bit 18 24 Reading 66B 2 0 401 VDC Figure 23 QCAD System Configuration Date 09 05 09 Time 09 58 48 Data Points 1 thru 2602025 of 2602025 Tau29 961470
11. system S N lt Command 1 Disable timer interrupts G Write 1 Enter servo gain gt Command 1 Enable timer interrupts H Write 4 Enter HF DDS sub word Configuration DDS Frequency E vr Detector E T or 2 or alue The QCAD Windows user interface includes a HF 1C96 6DESAE6F 13 400444249102321 Voltage 671D 1 007 VDC Configur ation screen shown in Figure 22 that LF 1096 51EF670E 13 400244249102400 Signal B4C1 28 7 can be used to display and change set all of the Deviation 1BF64761 200 0000 Hz Enor 0000 i system S opera ng parameters Please consult dps VHS lend Signal Waveform the Help file for the details about using the Staue HN QCAD configuration screen poss Mme Amplikade Maximum Attenua Data Stream i tion dB Word Value EEPROM Stream Rate 13 0 9961 Sec Read at Reset Save All Frequency BDE5AE 5 855334e 05 Servo Change Setting Word Value cers i HF DDS Sub Word Mod Rate E000 19 0735 Hz Beremeter DDS Sub Word Gain 08 256 x Hex Value EDESAEEF Bytes 4 Loop Open Closed Sweep C Rx Update Send New Value to QCAD DAC Marker r Defauls p PIC SZW Rev Word Value FS 0 22 MSBit fis 24 B Load Reading 73 0 841 VDC 0 von Reset Full Scale 4170e09 Calibrate 5 0e 10 Oven Temperature Setpoint D4BD 45 00
12. value that 1s integrated to become the DDS tuning word The signal samples are taken after a blanking interval to exclude the FM switching transients which in principle would cancel while additional samples are taken during the peak of the transient for monitoring purposes as shown in Figure 20 from the QCAD user interface application These signal waveforms closely resemble those of a passive cesium beam tube atomic frequency standard Detector Calculations are performed during the various sub intervals word Value Ls to perform additional averaging and prepare the DDS Voltage 20E1 0 595 VDL tuning word RS 232 and stripchart recorder data streams siml mp ET x A relatively fast 115 200 RS 232 baud rate is used to maximize the allowable data stream rate Error aooo TITT LIA A Signal waveform Ul LIN fJ LLLI Lg gn Status Q U da Figure 20 Signal Monitor Panel User Interface The QCAD processor has a simple comi COM Status 115200 8 1 N Start T hardware interface based on single See GLAD acad frd stel 6 040 QCAD ASCII character commands to enter SUE Ver 1 10 C 155 Stable32 exe Crad f o autol Em CIE Path ress Start to parameters and perform queries as read data via shown in Table 2 These Li AvgFactor fi H Points comm pa commands are then used to C Plot Configure ament Scan Stable32 Help Close implement
13. 042312e 07 a a2 1 252024e 08 3 a3 1 265733e 10 E Standard crystal TC cubic fit coefficients df f Al T To A2 T To 2 A3 T To 3 To 26 4 deg C Al 9 218778e 08 E EE EE MEE A2 2 495631e 09 poete a e Sa bed A3 1 265733e 10 au c a E UK E CRM GE UU EU NE UE ND SUE MUK NUN Delta Theta 1 090 min of arc pop wrt nominal angle of 35 25 degrees 0 p o 0g po ce EORUM i E TRE RE Upper turn point 49 88 deg C 30 40 50 60 70 Lower turn point 16 06 deg C Temperaure C TC slope at 45 deg C 5 37e 08 per deg C Figure 31 Crystal TC Plot with EasyPlot Figure 32 Excerpt from scan frd Data File Scan Rate A bidirectional scan may be used to determine the optimum scan rate which should be as fast as possible while avoiding hysteresis and to minimize the initial temperature lag An example of a fast scan rate 1s shown in Figures 33 and 34 which used a rate of 20 C per hour x10 the recommended value Even at that high rate the cubic fit determined by the average of the two directions 1s reasonably accurate For example the upper turnover temperature is estimated as 49 97 C instead of 49 88 C f List Filename C 5QCADSscan Configure EasyPlat Start C Plot Ereg s Pia TO H Paints 1 2630 Help Defaults Close Scan Type Up Low to High C Down High to Low Retrace L H L Scan Temperature High es D Current iq es 65 00 C 20 00 s 0 00 l n fo oo ogh P asg Hes Scan Progress Initial Dwell
14. 0e 01 File qcad 66c frd The entire freq PRE 0 ISo n DATA uency record is 11 0 iL oe shown in Figure 24 Te ee er O A abes after averaging the data by a factor of 3 twelve grOSS Dr outliers caused by 70 power line aa i OM TIERE NERA ee ENS HELPER ER S LU N disturbances during thunderstorms 7 Several frequency E AOL jumps are apparent 54 as is some cycling iL d and wandering but Mae esso rift The crystal l I was operated near E point at a drive 4o 25 50 75 100 125 150 175 200 225 250 275 30 0 level of about 160 Time Days uW Figure 24 Overall 1 Month Frequency Record Zooming in on the frequency record shows only a small effect 1x10 10 p p from air conditioner temperature cycling with a period of about 40 minutes The cause of the slower semidiurnal cycling is probably thermal but its exact origin has not been established and may follow power line variations Date 08 31 09 Time 14 15 42 Data FREQUENCY DATA File qcad 56 001 Stable32 includes Largest Frequency Jum P TOVicIONS i 8 00 s d r detecting and analyzing frequ OL TR T Figure 25 shows the ES largest jump of ri 6 50 about 3 1x10 at ar Day 5 7 Zooming a 6 00 in on the jump e shows that it O 5 50 occurred during a S single pair of l1
15. 13 4 MHz AT cut fundamental mode quartz crystal resonators for anomalous frequency jumps in critical GPS rubidium atomic clock application 1 While crystal oscillator static frequency error 1s removed by the atomic clock s frequency lock loop a crystal frequency jump can cause a permanent phase time error as the loop corrects a relatively large gt pp10 frequency change The mechanisms behind such quartz crystal resonator frequency jumps are complex and beyond the scope of this paper But it is well known that they depend on both the individual device and its operating conditions in particular temperature and drive level Itis therefore necessary to test every resonator over a fairly long period of time e g 1 month under its actual operating conditions to avoid anomalous behavior Furthermore it 1s more effective to perform that screening prior to installing it into flight hardware The equivalent circuit parameters for these crystals are shown in Table 1 Table 1 Equivalent Circuit Parameters of 13 401344 MHz Crystal in PerkinElmer GPS RAFS load capacitance at nominal 13 401344 MHz a 0 004030346 Value so that the crystal is resonant with its specified 25 pF The QCAD system operates the crystal under test as the shunt element in a two port network excited by a frequency modulated RF source The network transmission 1s sensed by an AM detector whose output is a minimum at the minimum impedance frequency of the crystal T
16. 2 24e 11 2 O4e 11 bibe deedebbbi HeeededeiebibHeeesbebebbbbReeedeebebebe Reese dedebder 2 15e 11 2 5 e 11 5 47e 11 4 92e1 T ie be dopE a L3ge 10 fi PP Pei Pp bp Pe IER 35 15e 10 periRbeeeeedeegerPdeppeeeeemeededenreeRReeeeedeeepBeReddep Be FeeeredeeBRiedqeeeexreedeierevdi 3 54e 10 Eius i iiio doi PPPH Fb PPM Po diio FOPIN 4 B4e 10 e uL cc cuu d cc 7 S51e 10 9 36e 10 ZEE Data Points 1 thru 1283410 of 1283410 FREQUENCY STABILITY Ba ua Tau23 9614700e 01 File qcad 66 005 Figure 27 shows a driftremoved stab ility plot for the second half of the record The stab ility improves from 6x10 at 1 second to 2x10 at 30 seconds and then gradually rises at longer averaging times remaining below 1x10 10 out to about 2000 seconds The data show white Second Half of 1 Month d Run E EM noise at short averaging times which becomes flicker EM noise and then random walk FM noise at long averaging times ppin Overla 10 102 10 104 10 Averaging Time t Seconds Figure 27 Stability Plot Overall this crystal s behavior 1s marginally acceptable for the GPS RAFS application Jump Limit The acceptable limit for a crystal frequency jump depends on the application In the case of the GPS RAFS where the crystal oscillator is locked to the rubidium reference with a relatively fast 10 msec frequency lock loop and the most important requirement is to maintain sub
17. The Quartz Crystal Anomaly Detector QCAD System W J Riley Hamilton Technical Services Beaufort SC 29907 USA bill wriley com Keywords Quartz crystal resonator frequency jumps crystal anomalies drive level sensitivity temperature frequency characteristic crystal resistance Abstract This paper describes a Quartz Crystal Anomaly Detector QCAD system for detecting jumps and other anomalies in quartz crystal resonators The crystal under test is operated at its minimum impedance frequency in a shunt network excited by a FM modulated DDS synthesized RF source The resulting AM signal is detected and used in a frequency lock loop servo to lock the frequency of the RF source to the crystal resonance The DDS tuning word is a direct measure of the crystal frequency and the DDS amplitude sets the crystal drive level Relatively fast 1 second low noise pp10 and high resolution pp10 frequency measurements are supported by the associated embedded software and user interface and the crystal behavior can be observed as a function of time temperature drive level and other variables in a convenient and inexpensive way Introduction The Quartz Crystal Anomaly Detector QCAD is a measuring system for detecting frequency jumps and other anomalies and characterizing the general behavior aging temperature coefficient and drive level dependence of quartz crystal resonators The QCAD system was developed as a means to screen
18. al Services QCAD Processor PIC Section H Riley ee Figure 16 QCAD Processor Schematic C21 c22 c25 C86 lu lu lu lu 3 3 M Power cea 43 3 4 47uF 10V C 1lu E b ielesleales zatzs zoio ai zal az aalea salsalea o AN From PIC 10 73 74 79 80 31 37 65 DYDD AYDD M VY 17 al SsDO S1BB 13 F LIORESET A2710 RESET A3 2 Reverse Polarity l po i5 a4 Uu m Protect 14 as Diade AD9852 V C15 C16 C1 E 0 UD CLK 48 Bit DDS 43p 22 p 150p eg o Q 1u RESET QUT1 7 IQUT1 Cl1 ci gt c13 C14 430p 470p 9 338p GND auT2 n 7 1 dB Ripple Elliptic tauT2 2 5 ohms DAC O P Cuttoff 15 MHz Atten gt 6 dB above 18 MHz ry DIFF CLK ENABLE VOUT PLL FILTER Ma S P SELECT ipaum R2 R15 7WR SCLK C27 2 00k 249 9 RD CS NC 13 35 57 58 63 lu V V FSK BPSK HOLD iis rid s Ka a A TEATE t V SY Comparator A Su C9 u4 Bypass 3 5 p lt amz To PIC e 1 MHz V Reference V V J R12 amp R13 at end of 137 ahm lines at DDS pins 3 3v Hamilton Technical Services QCAD Processor DDS Section i Re B PKI aaa28 Figure 17 QCAD DDS Schematic Firmware The firmware of the QCAD processor has two main functions processing the discriminator signal to tune the DDS frequency and realizing the user interface that controls the system and outputs its readings The former is implemented using multi phase timer interrupts while the latter utilizes a loop that waits
19. ce Ohms M 0 13 4001 13 4002 13 4003 13 4004 13 4005 Frequency MHz Cy 25 rfi xis Capacitive Figure 9 Crystal Impedance Circle Figure 10 Crystal Impedance Plot Detector Hardware The QCAD detector board holds the 6 dB 50 Q RF drive input attenuator and crystal network followed by a forward biased temperature compensated diode AM detector a x25 signal preamplifier and the Analog Devices AD7457 12 bit analog to digital converter ADC It also has the oven temperature controller with its thermistor bridge DC amplifier voltage to current converters heater transistors demand current limiter and an Analog Devices AD5320 12 bit digital to analog converter DAC to set the oven temperature Those circuits are shown in Figure 11 The oven uses four thermistors and heaters on each side to minimize thermal gradients as shown in Figure 12 Schematic diagrams of the QCAD detector and temperature controller circuits are shown in Figure 13 and 14 Crystal AM Attenuator TE eric Detector Preamp ADC Syo To s M uC RF punge Crystal Ven Sere spears From uC Thermistors Figure 11 Detector Circuits Figure 12 Crystal Oven A 12u 12 VDC Power Input C HTRRET Regulator A 5v C HTRSUP a LM78L 5 Preamplifier Size D C5 In _ Out ua C Size A pin C9 9 001
20. ctors on the QCAD Processor and Detector boards Temperature Scan QCAD Crystal Dissipation Versus Crystal Resistance for 0 dBm Detector Board Drive Crystal Dissipation mW 10 15 20 25 30 Crystal Resistance Ohms Figure 29 Crystal Dissipation Versus Resistance The QCAD Windows user interface application includes provisions for performing a temperature scan on the crystal under test Temperature scans anywhere between 25 C and 65 C are supported in steps of about 10 m C with an adjustable initial dwell time and ramp rate Temperature and averaged frequency readings are written to a disk file at each temperature increment The resulting temperature coefficient data are fitted to a cubic equation and estimates of the crystal TC parameters turnover temperatures AT cut angle and TC at the nominal 45 C RAFS operating temperature are provided These estimates are satisfactory even when based on a fairly narrow temperature scan The TC and fit data are plotted on screen the former as the scan progresses A 30 C to 65 C scan at a rate of 2 C takes 17 2 hours and produces about 3500 data points where each frequency value is the result of averaging about 18 1 second measurements as shown in the example below The ambient temperature must be at least 5 10 C below the lowest scan temperature to maintain proper oven temperature control The 10 m C temperature steps cause transients in the crystal oven heater power but they
21. er and its associated firmware can be rather easily adapted for other frequencies in this general range Similarly the FM modulation rate and deviation can be adjusted for other resonator types The physical arrangement of a single unit for each crystal under test could be changed to a multichannel architecture with a single RF source and processor by switching the detector signals as was done in the original 1960 s system The on board crystal oven could be replaced by an external temperature chamber perhaps for multiple crystals under test The emphasis could be changed from observing the behavior of a device under constant operating conditions to making TC measurements or even characterizing static g sensitivity or radiation sensitivity As presently implemented it is easy to sweep the crystal resonance or scan the unit versus temperature to find its turnover point Specifications The specifications for the QCAD system are shown in Table 3 Table 3 Quartz Crystal Anomaly Detector QCAD System Specifications Frequency Range 13 399658203125 To 13 401489257812 MHz aao WS sts pom rt nominal 13400400 Mis At minimum impedance frequency of quartz crystal resonator under test Frequency Resolution 48 Bit internal DDS LSB 5 03x10 3 Byte 24 bits numeric data stream with 137 ppm range and 8 14x10 resolution DDS Output Power 0 dBm nominal max adjustable with 12 bit resolution Quartz Crystal Compatibility HC 37 U 1 TO 8 Ho
22. for and responds to characters from the RS 232 I O Signal Processing The QCAD system uses a relatively low squarewave FM modulation rate 19 Hz and an FM deviation approximately equal to the crystal resonance linewidth 100 Hz This produces a detected signal that has switching transient followed by DC levels that indicates the frequency offset from the center of the resonance where they are equal for the two half cycles Oscilloscope traces of the QCAD discriminator signal are shown in the upper portion of Figures 18 and 19 for open and closed loop conditions The lower traces are the squarewave FM waveform Two types of signal averaging are used to reduce the noise level The first utilizes multiple signal samples during each modulation cycle and the second averages those samples over the data stream interval Both of those actually use summation rather than averaging to increase the servo gain and reduce quantization The net effect 1s that further averaging does not significantly improve the short term stability which 1s dominantly flicker noise from the diode detector and preamplifier Figure 18 Open Loop Signal Figure 19 Closed Loop Signal Each modulation half cycle 1s divided into 16 sub intervals and the discriminator signal is sampled eight times during eight of those sub intervals to reduce noise and ADC quantization Those samples are then summed and the half cycles differenced to produce a frequency error
23. hat discriminator response is amplified synchronously detected and integrated to produce a frequency control voltage for the source The overall arrangement shown in Figure 1 resembles the frequency lock loop of a passive atomic frequency standard No test oscillator 1s required Ta Frequency i M Quart Crystal Resonator Measurement System Frequency Lock Servo vael Tug 3 RF Source C rystal Envelope Preamp VD XO Hebu o ri D etectar P D i criminatar Tur Frequency Control E Synchronous E 5 3 Detector Modulation Reference Signal Mio dulation signal Generator Integrator Error Signal C antral Voltage Figure 1 QCAD System Concept This process of locking a frequency source to a passive resonance response is similar to that first described by R V Pound for microwave oscillator 2 and an arrangement for locking an oscillator to a passive quartz resonator described by Walls and Stein 12 In particular the QCAD system is based on a similar one that the author developed in the late 1960 s for measuring the frequency temperature characteristics of quartz crystals used in uncompensated room temperature counter time bases That system used a single frequency modulated source that excited a bank of crystal networks and AM detectors whose outputs were multiplexed into a single frequency lock servo That system which required an external frequency measuring system operated at a relatively slow sampling
24. lder 5 0 017 diameter leads on 0 1 grid 0 2 to 0 6 long Quartz Crystal Oven Temperature 25 C to 65 C settable with 0 02 C resolution Crystal Drive Power Range 8 dBm nominal max adjustable with 12 bit resolution Measurement Interval Adjustable between 1 to 13 seconds shorter intervals possible but not recommended Modulation Rate 19 Hz nominal can be adjusted but not recommended Short Term Stability ADEV lt 1x10 at 1 second 2x10 typ at t 30 sec Temperature Coefficient TC lt 2x10 C Strip Chart Recorder Output 0 to 1 VDC 12 bit resolution with range adjustable in octave steps from 5 03x10 P to 1 71x10 FS update rate same as modulation rate w o ext analog LPF Frequency Reference 10 MHz 7 dBm nominal 0 to 10 dBm Signal Monitor 0 to 2 5 VDC shows DC detector bias 1 0 VDC nominal e and recovered discriminator signal Power Supplies 12V 1 5A and 3 3V 9 4 0A 12 VDC 300 mA max 2 5 mm inner conductor with 5 5 mm OD connectors supplied with unit 43 3 DVC Q 350 mA 2 1 mm inner conductor 3 W SS at 25 C and 45 C setpoint from 120V AC line Size LxWxH 7 01 x 7 01 x 2 64 Hammond 1590F Enclosure 2 8 pounds DC Power Frequency Reference BNC Outputs Strip Chart Recorder dual binding posts Signal Monitor BNC Oscilloscope Sync BNC Crystal Oven Heater Voltage Monitor 2 banana jacks RS 232 bipolar 115 000 Baud 8 data bits 1 stop bit no
25. lyzer measure save list clear list Hz do 3 3O0 00 JL H jc F RO fice of merit 13407222 08 188460 0 03447632543 4 087344866615 15 41 1 207761133E 12 0 00821 2 13407221 63 188301 0 034466695569 4088490819E 15 1542 1 219487648E 12 0 00826 3 13407221 86 188380 0 03447153738 4087916516E 15 1542 1212821104E 12 0 00821 4 1340722192 188342 0 0344746189 4087550964615 1542 1213052321E 12 0 0083 5 1340722174 188374 0 03447231443 4087824447E 15 1542 1215387356E 12 0 00829 G 13407221 54 188422 0 03447504504 408750078E45 15 41 1 208103787E 12 0 00825 Figure 37 Measured Crystal Equivalent Circuit Parameters Alternative Crystal Networks A resistive reflection coefficient bridge could be used in place of the shunt crystal network with the crystal 1n the unknown bridge arm 11 The resistance of the other equal bridge arms would have to close to that of the crystal resonance resistance to obtain a good null condition and a balun transformer would be required at the output to the detector The response of the reflection coefficient bridge would have essentially the same FWHA and require higher source power for the same crystal drive and it is not obvious that that arrangement would offer any advantage over the shunt network Other Applications While the present embodiment of the QCAD system is directed specifically to measuring 13 4 MHz AT cut fundamental mode resonators the DDS synthesiz
26. nanosecond timing the issue becomes what time error results as the loop corrects the frequency jump As a rough estimate a 1x10 frequency change that persisted for 10 msec would cause a time error of 0 1 nsec a reasonable limit for this application FREQUENCY DATA Crystal Frequency vs RF Drive Level Drive Level The QCAD system can easily measure ssl the drive level sensitivity of a crystal by measuring the frequency as a function of RF drive power by varying the internal DDS amplitude via the QCAD Windows application or changing the setting of an external step attenuator The latter Relative Frequency pp10 method was used to obtain the data of a 2 Figure 28 which shows the frequency Ey et a ee mE variation for RF attenuator steps from 10 a rdg 33 7 457 35 735 45 100 Time Minutes dB to 0 dB in 1 dB steps with retraces to 10 db and 7 dB As expected the Figure 28 Frequency versus Crystal Drive Level sensitivity increases with increasing drive power estimated to be a maximum of 500 microwatts in these tests where the RF drive sensitivity was about 2x10 dB The variation in crystal dissipation versus crystal resistance is shown in Figure 29 for a nominal 0 dBm drive level to the QCAD detector board That level can be reduced by DDS attenuator setting in the QCAD Windows interface Configuration screen or increased by adding an optional amplifier between the DDS O P and detector RF input conne
27. parity USB with adaptor circuit programming User Interface Software QCAD Windows application provides parameter setting data monitoring and capture frequency and temperature sweeps supplied with unit Analysis and Plotting Software Stable32 supplied with unit EasyPlot recommended Analysis Software The Stable32 program for frequency stability analysis can be launched by a button on the QCAD Windows user interface at any time during a run Besides its general capabilities Stable32 includes several features of particular interest for use with the QCAD system An automation script may be invoked when the program is launched to automatically perform an analysis During manual file reading the Comments button of the Multicolumn Data Details function can be used to display comments that are automatically or manually entered into the QCAD data file during a run Stable32 also includes a frequency jump detection and analysis feature that can be used to check the record for jumps EasyPlot or another general purpose scientific plotting program can be used to plot the results of QCAD frequency sweeps and temperature scans from their respective data files Conclusions A simple yet effective method has been developed for making frequency measurements on quartz crystal resonators that is well suited for evaluating their general behavior aging frequency jumps temperature coefficient drive level sensitivity and the like The QCAD sy
28. rate but performed very well for its intended purpose for crystal TC measurements An important advantage of the arrangement is that no phase sensitive RF switching is required The calculated and actual resonance characteristics for the crystal of Table 1 are shown in Figures 4 6 The actual resonance plot is inverted by the preamplifier The current QCAD system also operates in a similar fashion But rather than multiplexing a single frequency source it uses separate DDS synthesizers clocked by a frequency reference to excite each crystal whose tuning words directly indicate the crystal frequency That parallel architecture can operate at a faster rate and is easily expanded to the number of channels required No external frequency measurement system is needed and it retains the feature that the RF phase is not critical thus permitting the crystal network to be located remotely e g in a temperature chamber from the RF source and processing circuits The present embodiment includes a programmable crystal oven temperature and crystal drive power both of which are important factors related to frequency jumps Photographs of the QCAD system are shown in Figures 2 and 3 r o B Figure 2 QCAD System Exterior of Box A circuit model of the QCAD crystal network is shown in Figure 4 The QCAD Windows applications includes means for sweeping the DDS frequency through the crystal resonance and recording the detector response as shown in Fig
29. rk Analyzer 11 R C Smythe An Automated Resonator Measurement System Using a Reflection Coefficient Bridge Proceedings of the 35 Annual F requency Control Symposium May 1981 pp 280 285 12 F L Walls and S R Stein A Frequency Lock System for Improved Quartz Crystal Oscillator Performance IEEE Transactions on Instrumentation and Measurement Vol IM 27 No 3 September 1978 pp 249 252 d Appendix 1 QCAD RF Amplifier Voltage Regulator D1 S1BB 13 F U1 LM 8LO9 4 In Out 4 7u 20V Com 12 to 15 VDC B Size We GND Wb Omit R1 for initial reguletor testing TO 92e All Unmarked Rs and Cs 0805 SMD lu RS 10 R3 Ue LMH6 03 ia RF Level Adj IN SMB Cable from DDS O P W4 R9 GND o 50 Bournes 3296Y 1 500 Use Either R9 or Re PWB P N 121509 1 Has R2 PWB P N 121509 2 Has R9 Nominal Signal Levels Input dBm 1 mW into 50 ohm input Z 0 224 Vrms Op Amp O P 1 39 Vrms 3 94 Upp Amp O P 0 7 Vrms 10 mW 10 dBm into 50 ohm load Xtal Drive Pur 1 1 mW 15 ohm Rx 1 5 mW 30 ohm Rx R can be raised to increase amplifier gain Figure Al 1 Photograph of QCAD RF Amplifier W is Regulator TP or Regulated Supply Input al 11 mA tup CS lu A RF Amplifier C6 1lu R8 WS B OUT SMB Cable to Detector Input W6 GND 392 R6 Op Amp VG Overall Gain 49 9 8 86 12 9 dB 75 0 6 23 9 9 dB One of Four Amplifiers on PWB Panel 2 1 Version 2
30. so has a 12 bit DAC that is used to produce an analog output to drive a stripchart recorder Schematics of the QCAD processor and DDS circuits are shown in Figures 16 and 17 Reverse polarity protection diades Scape Sunc J J8 SMB SMB Q amp si 1 efenn si 1 efenn TEE EI c2 Ji S1BB 13 F 12 to 15 VDC Power Control Suitches amp LEDs L Ji Sua SW LED Up Hi SW3 I Doun TX QUT SY Regulator 5u RX IN from PC RLIN R2IN Cl Cil C2 INVALID Sleeve Tip MAX3235E RS 23e 70 uses H W UART ed MHz Clock From U4 H J bese Ground z FORCEON D tT 19 rogcEorF 3 5 mm Stereo GND Phone Jack V Tip of stereo plug is data from PC JU Middle contact of stereo plug is data to PC Signal TP ACAD DDS Sep 11 09 49 28 REF 0 0 dBm ATT 10dB wrt B blnk dBE A EZPLLLLI UESESSEE El i REESE ELI E CENTER 13 400 MHz RBEM 10 kHz VBW 10 kHz SPAN 3 00 MHz SWF 60 ms Figure 15 DDS Spectrum J3 To QCAD Crustal Section DB9 Female CLK DAC SYNC DATA GND ADC CS 12V SIGNAL HTR RTN Logic Level HTR SUP Conversion C6 lu V Separate Trace to Pur Jack Gnd ea To DDS SCLK UDCLK DIO CTORESET_ Ubb2 2 bd For ISP Connect cable to PICSTART Plus programmer Apply 5V power no 3 3 V power required 35 Unplug J3 external connections to RB6 and RB SP No reference clock required Hamilton Technic
31. stem concept has been found effective for screening resonators for frequency jumps and other anomalies and it could be adapted for a variety of measurements on quartz crystal resonators Acknowledgements The author wishes to acknowledge the support of the PerkinElmer Rubidium Department for the development of the QCAD system References 1 GPS Rubidium Atomic Frequency Standard RAFS PerkinElmer Inc Salem MA 01970 USA 2 R V Pound Electronic Frequency Stabilization of Microwave Oscillators Review of Scientific Instruments Vol 17 pp 490 505 November 1946 3 J R Vig Quartz Crystal Resonators and Oscillators for Frequency Control and Timing Applications A Tutorial 4 E Hafner The Role of Crystal Parameters in Circuit Design Proceedings of the 22 Annual Frequency Control Symposium April 1968 pp 269 281 5 W S Metcalf Practical Crystal Measurements and Standardization Proceedings of the 27 Annual Frequency Control Symposium June 1973 pp 55 62 6 V E Bottom Introduction to Quartz Crystal Unit Design Van Nostrand Reinhold 1982 ISBN 0 442 26201 9 7 User Manual Stable32 Program for Frequency Stability Analysis Hamilton Technical Services Beaufort SC 29907 USA W J Riley Algorithms for Frequency Jump Detection Metrologia 45 2008 S154 S161 9 EasyPlot Program for Scientific Plotting and Data Analysis Spiral Software Brookline MA 02146 USA 10 T Baier DG8SAQ Vector Netwo
32. tector bias or temperature R estimates of about 16 ohms were obtained for the Bliley 040 over a 30 C to 65 range a value that implies Q 184 000 JDG8SAQ Vector Network Analyzer Software licensed to William Riley File Measure Settings Tools Options Help Network Analyzer Measurements lox 20ohm 1 13 407222MHz 15 45ohm 0 53 i 0 01 The validity of the crystal resistance measurement was confirmed with reflection coefficient measurements made on a vector RF network analyzer VNA 10 as shown in Figure 36 The plot shows the magnitude of the crystal impedance and the reflection coefficient Smith chart near the crystal resonant frequency The marker at the minimum Z point shows a resistance of 15 4 ohms in good agreement with the QCAD value of 16 ohms Figure 37 shows the crystal equivalent 9 ss iaxsune Cente 1340725 MHz Stop 13 408 MHz ohm circuit parameters extracted by the same ws ui Continuous instrument by fitting the measured impedance Z circle The measured motional inductance value is Y in good agreement with that shown in Table 1 Figure 56 VNA Crystal Impedance Measurement Crystal Analyzer Analysis will be performed into Mem X Equivalent Circuit L 344 504 mH L cC R C 40975 fF Al 15 41 316028 Ohm BUE 1208103787 pF suto optimize 9 f 3e4pe 240722153 MHz gt Figure of Mart 0 00825 Ware 2304 179773 1000 O 188422 Batch Crystal Ana
33. ure 5 It is somewhat asymmetrical perhaps because of the finite sweep speed The data can be replotted and fitted as shown in Figure 6 The Lorentizian fit has a full width at half amplitude of 287 Hz and inflection points at 83 Hz nearly the same as the normal FM deviation of 100 Hz The single tuned circuit fit implies a Q 98 000 the typical specified value Neither fit follows the resonance data at lower amplitudes probably because of detector nonlinearity QCAD Frequency Sweep for Bliley Crystal S N 040 Detector Voltage VDC 13 4000 13 4005 13 4010 13 4015 0 13 3995 Frequency MHz Figure 5 Actual Crystal Resonance Figure 3 QCAD System Interior of Box Micro Cap 8 Evaluation Version xtalmeas cir 00013 45M 13 452M dix v 4 50 2 22 Olu lu 2 C4 ul r Ml Ley i 10k 1M TC2 a D2x Fea ij ind 511p Olu Figure 4 Model of Crystal Resonance QCAD Frequency Sweep for Bliley Crystal S N 040 T Lorentizian Fit Single Tuned Circuit Fit i Center frequency 354 Hz Center Frequency 352 Hz d FWHA 287 Hz Q 98 000 2 Peak 1 45 Peak 1 54 o gt 5 o o D a 0 200 100 0 100 200 300 400 500 600 700 800 900 1000 Lorentzian Fit Single Tuned Circuit Fit y b 1 4 x c x c a a Frequency Hz wrt 13 4 MHz y a sqrt 1 2 b x c c 2 b x c c a 286 715 b 1 45403 c 354 095 a 1 53634 b 2 57402 c 352 215 Figure 6 Crystal Resonance Replotted and Fitted
Download Pdf Manuals
Related Search