2) Calibration Method and Example

Transcription

1 Calibating Receive Fequency and Soundcad Sampling Rate uing an O-i Standad Fequency Jim Moitz, M0BMU ) Intoduction When uing a yntheied eceive and PC with ound cad with weak-ignal otwae uch a WOLF [ o to make high-accuacy meauement o ignal equencie, it i neceay to calibate the eceive equency and ound cad ample ate. O-ai equency tandad ae eadily acceible, uch a MSF on 60kHz and the Doitwich 98kHz tanmitte. But the equency meaued by the RX/ound-cad combination i aected both by eceive equency eo and ound cad ampling ate eo, o it i not poible with a ingle meauement to epaate and calibate both ouce o eo. The ound cad ample ate can be eaily calibated i a peciion audioequency ouce i available, and the calibated ound cad ued in conjunction with the o-ai ignal to detemine the eceive eo, but many amateu hack do not contain a uitable audio ouce. n elegant way aound thi poblem uing the eceive in M mode to eceive Loan-C pule ha been developed by John ndew, WTG [ n altenative method that imultaneouly calibate both ample ate and eceive eo, uing the eceive in SSB mode and any convenient LF tandad equency i given below. The baic idea i that making two equency meauement allow imultaneou equation to be contucted and olved that detemine both the RX equency eo and ampling ate eo unknown. Thee ae eveal way thi could be done; e.g. by meauing two dieent o-ai ignal, by etting the eceive to poduce dieent audio output equencie om the ame ignal. But a convenient method with mot eceive i to meaue the audio equency output poduced by the ame ignal in USB and LSB mode. Fo example, i the eceive equency i et.5khz below the ignal equency in USB mode, and.5khz above ignal equency in LSB mode, then with no eo the meaued audio pitch will be.5khz exactly in both cae. But i the actual RX equency i low, the output pitch on USB will be highe, while the LSB pitch will be lowe and vice-vea. Howeve, i the RX equency i coect, any eo meaued will be due to the ound cad ample ate, and will be the ame iepective o whethe USB o LSB i ued. In thi cae, a low ample ate will lead to highe appaent audio pitch and vice vea. t an intuitive level, the eceive equency eo i about hal the dieence between the audio pitche meaued in USB and LSB, while the eo poduced by the ound cad i oughly the dieence between the expected audio pitch and the aveage o the meaued USB and LSB pitch. In thi way, eo due to eceive equency and ound cad ample ate can be epaated out. Section 2 give the method and omulae ued to peom calibation. Beoe uing it, howeve, ead Section 3 on the limitation o thi method. I you ae inteeted to know how the omulae wee deived, ee Section 4. 2) Calibation Method and Example Select a uitable LF tandad equency ignal thi hould have known accuate equency, a good ignal-to-noie atio and a tong caie component example in Euope ae tandad equency tation like MSF (60kHz), HBG (75kHz), DCF77 (77.5kHz), and ome LW boadcate uch a DLF (53kHz) lloui (62kHz), Doitwich (98kHz). 98kHz i ued a an example hee, and DL4YHF WOLF GUI otwae aumed. The example numbe quoted ae the actual igue obtained uing an FT-87 a the eceive, and an old 300MHz laptop PC. Set the eceive to USB mode and a caie equency.5khz below the tandad equency (96.500kHz). In the WOLF coniguation ceen, et the eceive option ample ate to the nominal value (8000Hz) and the cente equency to 500Hz. Fom the mode menu elect equency meauement. Check that the ignal caie i cented in the WOLF pectum diplay. te 96 econd you hould have omething like:

2 :38:07 >WOLF m 96 -w t: 96 : 0.98 a: 0.5 dp: 8.7 : 0.98 i the equency oet meauement. In thi cae, Wol meaued a equency oet om the nominal 500Hz o 0.98Hz. Call thi oet. Now et the RX to LSB, et the caie equency to.5khz above the tandad equency (99.500kHz), and epeat the WOLF equency meauement. Fo thi example, the equency oet wa now 0.060Hz. Call thi oet. Now the ample ate and eceive equency eo can be calculated: equency o the o-ai tandad, e.g Hz expected audio equency, e.g. 500Hz Meaued oet in USB mode, e.g. 0.98Hz Meaued oet in LSB mode, e.g Hz nominal ample ate, e.g 8000Hz actual ample ate / actional eo in the eceive caie equency The actual ample ate i: ample/ec in the example The actional eo in the eceive equency i: x 0-7, o pat pe million (quite good ) The actual ample ate value can now be enteed into the ample ate boxe in the WOLF coniguation ceen. Calibating the ample ate uing a peciion audio ouce yielded a value o ample/ec; the eult agee within about 2 pat pe million. In nomal ue, the actual ignal equency will be dieent to the tandad equency ued o calibation. Fo WOLF eception, we would like to conigue the otwae with an accuate value o the expected audio equency o the ignal being poceed by the ound cad. Suppoe we wih to eceive a WOLF ignal at 503.5kHz, and we et the eceive to 502.5kHz USB to obtain a nominal khz audio output. The actual audio output equency can be calculated om: udio actual audio equency o eceived ignal caie equency o ignal, e.g Hz Caie eceive equency etting in USB mode, e.g Hz udio Caie Caie Hz, in thi cae cloe enough to 000Hz When doing thee calculation, take cae to include any minu ign and powe o 0; alo, make ue all equencie ae in hetz, not kilohetz! The method and omulae above ae intended o ue with WOLF, howeve the ame technique could be ued with othe ound cad baed otwae uch a Spectum Lab, go and othe. In thi cae, bea in mind that the audio equency meauement will be o the actual equency, athe than the oet om the expected equency a meaued by WOLF. So i imila meauement to thoe above wee done uing go, o example, on USB you might meaue a equency o Hz, and on LSB 2

3 499.75Hz. Then Hz, and Hz. To get wothwhile accuacy in the ample ate calibation, eolution o about 0.0Hz in the equency meauement i equied. Thi mean uing pectogam etting o QRSS20 o imila. Be awae that the ample ate calibation will have to be epeated o each dieent ample ate ued. It i not neceay to ue an audio equency o 500Hz; highe audio equencie will impove accuacy o ample ate calibation (and lowe equencie make it le accuate). Jut ente the appopiate value o expected audio equency,, into the omulae and WOLF coniguation, and et the eceive equencie accodingly. Mot eceive diplay caie equency in SSB mode, but ome (e.g. elective level mete) diplay the cente equency o the eceive paband intead, a do eceive in CW mode. Thi need to be taken into account when etting the eceive equency and deciding what i. Dit in eceive equency between the two meauement will ceate an eo in both ample ate and eceive equency calibation. dit o < 0.0Hz ove the peiod equied o the meauement i acceptable. With untabilied cytal ocillato a eceive eeence, and o-ai equency tandad in the LF ange, I have ound that dit can tay within a ew milli-hetz ove eveal minute, povided the equipment i given plenty o time to tabilie, i kept away om heat ouce and apidly changing ambient tempeatue, etc. Equipment with TCXO o OCXO eeence ocillato i conideably bette. It i a good idea to epeat the meauement a ew time to make ue igniicant dit i not occuing. 3) Limitation Thee ae ome igniicant limitation to the accuacy and application o thi calibation technique: The equency calibation utility in WOLF doe not appea to give coect eult when the oveall equency eo i geate than Hz. t 98kHz, a Hz eo i about 5 pat pe million, and at 60kHz it i about 7 pat pe million, howeve calibation o mot eceive will be within thi. I in doubt, check the equency oet detemined by WOLF i conitent with the equency o the caie in the WOLF pectum diplay. Thi method cannot be ued when the eceive ha moe than one independent ocillato detemining the eceived equency, ince thi intoduce additional unknown vaiable. Example o thi ae a eceive ued with an LF HF convete with it own cytal ocillato, eceive having a epaate, tunable BFO, o ome type o equency yntheie uing a epaate intepolation ocillato. Howeve, the vat majoity o moden eceive ue a ingle eeence ocillato. Mot moden yntheied eceive ue DDS yntheie. The tuning tep o a DDS i a binay action o the eeence equency, i.e. it i not an exact whole numbe o hetz. Theeoe, the actual output equency doe not nomally coepond exactly to the indicated dial equency, even when the eeence ocillato i peectly accuate; the tuning micocontolle elect the cloet available equency tep. The actual tuning tep i typically a ew ten o millihetz, o eo o thi ode can be expected in equency meauement, which will vay with the equency etting. In pinciple, thi eo i pedictable, but in pactice pediction i not nomally eaible becaue it equie knowledge o the algoithm ued to calculate DDS phae accumulato value, a well a the eeence equency, equency multiplication cheme, etc. udio equency meauement accuacy to a ew ten o millihetz limit the accuacy o ample ate detemination to a ew ten o pat pe million, howeve thi i good enough o WOLF and mot othe thing. The eceive equency calibation accuacy i o the ode o 0. ppm, which again i good enough o mot thing, including WOLF. 3

4 4) Deiving the Fomulae The oveall eect o the equency conveion inide a adio opeating in SSB mode ae to poduce an audio output equency equal to the dieence between the incoming tandad ignal equency and the caie equency et on the eceive dial, i.e: udio udio Caie Caie in USB mode, o in LSB mode The meaued value o udio i modiied by two ouce o eo. The it ouce o eo i due to the yntheie eeence equency not being exactly equal to it nominal value; i the nominal eeence equency i R and the actual eeence equency i R, then the actual caie equency i Caie ( R / R ). The atio ( R / R ) i the ame whateve equency the eceive i et to. The abolute equency eo i theeoe popotional to the et caie equency. (Note that thi only applie to adio in which all conveion ocillato equencie and the BFO equency ae deived om a common eeence; thi i tue o the vat majoity o moden yntheied eceive). The econd ouce o eo i the ound cad ample ate. Thi eult in the audio equency udio meaued by the otwae being dieent om the actual audio equency by an amount udio ( / ) udio, whee i the nominal ample ate, and the actual ample ate. The meaued audio equencie obtained in USB and LSB mode can theeoe be witten a imultaneou equation: U USB () L LSB (2) Whee: nominal ample ate actual ample ate / atio o actual eceive caie equency to indicated equency equency o the o-ai tandad, e.g kHz USB eceive caie equency etting in USB mode, e.g kHz LSB eceive caie equency etting in LSB mode, e.g kHz U meaued audio equency output in USB mode (appox. 500Hz) L meaued audio equency output in LSB mode (appox. 500Hz) The unknown vaiable we wih to ind ae / and /, which can be ound by elimination o othewie ( an execie o the eade, a they ued to ay!) to be: L U USB L U LSB (3) L USB U LSB (4) ( LSB USB ) 4

5 The WOLF calibation utility output the oet between the meaued audio equency and the expected, nominal audio equency, and the nominal equency i the dieence between the o-ai tandad and the eceive caie equencie. The eult that need to be enteed into WOLF (and uually othe otwae whee ample ate calibation can be done) i the actual ample ate. lo we want to know what the actual audio equency o the ignal we ae tying to eceive will be when the eceive i et to ome dieent eceive equency. To poduce convenient omulae to do thi, we can make the ollowing ubtitution in Equation 3 and 4: expected audio equency (e.g. 500Hz), oet in LSB and USB mode epectively meaued with WOLF equency check utility (a action o a hetz). L, U USB - LSB Then, ate ome moe execie o the eade, the omula o the actual ample ate become: (5) 2 2 The eceive caie equency eo i: 2 2 ; But, ince 2 <<, (6) Fo ignal in the LF ange, and audio equency eo o a ew hetz o le, the eo involved in thi appoximation i unlikely to be moe than a ew pat pe billion. moe ueul quantity to know i uually the actional eo in equency, / : - (7) 2 The actual audio equency the wanted ignal will poduce with a paticula etting o caie equency, auming USB mode i ued, i then: udio Caie udio Caie Caie Caie (8) Veion t Octobe 2009 Jame Moitz, M0BMU 5