LF: WOLF test results, soundcard trouble

[DL4YHF@aol.com]

Hello group,

An update on the WOLF tests on easter weekend:

During more WOLF tests (which finally resulted in a 2-way WOLF QSO) with   
DF6NM, Markus found out what was wrong:

It was the TX sample rate on my side (operating from DF0WD). Here is the 
reason (..water on the mills of the DSP guys ;-) :

The precise sample rates of the soundcard, running at a nominal sample rate 
of 8000 samples per second, were **DIFFERENT** for TX (D/A conversion) and RX 
(A/D conversion)  !

First suspecting a bug in Spectrum Lab, I finally verified the sample rates 
for "record" and "play" with Windows Sndrec32.exe. I recorded 60 seconds of a 
sine wave from an audio generator, then replaying it with the same program. 
An oscilloscope in X/Y mode, with X connected to the generator and Y to the 
soundcard's audio output did NOT show a steady Lissajous figure (hope that's 
the correct english word for it)....  the Lissajous wasn't steady so the two 
input signals had different frequencies. 

The true sample rates of the soundcard were then measured by other means, 
they were:

8002.87 samples / second for the DAC (used to transmit)
7938.09 samples / second for the ADC (used to receive)

8002.87 is very close to the frequency Markus proposed during the test (he 
transmitted this via "backup link"in PSK31 by the way). I changed the sample 
rate in the "-r" argument passed to the WOLF software - et voila, immediate 
copy, with the center frequency of the WOLF signal where it should be (on 
137.5000 kHz).

This surprising fact occurred on a notebook with built-in soundcard (an "ES 
1869 Audio Drive", whatever that is). 


Conclusion

If you can receive other "reference" WOLF transmitters (like M0BMU) 
perfectly, but no-one can decode your transmission, dont be fooled by the 
soundcard (and don't share my fate hi). Verify the proper A/D AND the D/A 
conversion rate (if you use the soundcard's output for transmission). Btw the 
nasty notebook didn't cause trouble when using 11025 samples/second. I don't 
know if this a common problem with soundcards.

I hope this info is useful for other software authors. If your program must 
TRANSMIT something very precisely, use "individual" calibration factors for 
both A/D and D/A conversions. You may use the high-resolution timer to verify 
the sample rates by software, assuming the accuracy of the high-resolution 
timer is better than the accuracy of the soundcard's internal clocks.

Good luck, and thanks to everyone involved. Hope to be QRV on Wednesday 
evening but not sure yet.

 from Wolf (DL4YHF).