Re: VLF: DF6NM on 8970.002 - samplerate detector seems ok

[Markus Vester <markusvester@aol.com>]

In the meantime, I could reproduce the long transient in SpecLab's samplerate lock. It is apparently brought up by a thermal samplerate drift on my netbook after turnon. In the attached screenshot you can see that the samplerate was initiaIly close to +2 ppm, which is outside my 1 ppm capture range. After the SR has drifted into range, the lock indicator turns green, indicating a correct lock. However the "Error Integral", whose magnitude is normally much less than 1, suddenly jumped to an erroneous high value (663552.1). As this number is interpreted as elapsed phase, the output frequency is steered high for about two hours, during which the error creeps slowly back towards zero. 

 

To avoid this situation, you can simply make sure that the capture range is large enough (eg. 5 ppm) to include the "cold" samplerate at startup. In case you do encounter an unplausibly large Error Integral, it can be reset immediately by restarting the samplerate detector (temporarily tick measure only), which brings the system back to the correct frequency within a few seconds. 

 

The effect seems to be peculiar to the MSK detector, I did not see it happen with the Goertzel filters tracking a straight carrier.

 

Best 73,

Markus (DF6NM)

-----Ursprüngliche Mitteilung-----
Von: Markus Vester <markusvester@aol.com>
An: rsgb_lf_group <rsgb_lf_group@blacksheep.org>
Verschickt: Do, 29 Mrz 2012 12:49 pm
Betreff: VLF: DF6NM on 8970.002 - samplerate detector seems ok

Just a short update: I tried a quick local test of the SR detector between two instances of the latest SpecLab 277 b22. The SR detector engaged, and the generator output followed within seconds. So it seems SpecLab is not the culprit for my transient frequency errors, and it's likely I did something wrong during the transmissions (yes I'm sure I had not ticked "measure only" box...).

 

Best 73,

Markus

 

 

-----Ursprüngliche Mitteilung-----
Von: Markus Vester <markusvester@aol.com>
An: rsgb_lf_group <rsgb_lf_group@blacksheep.org>
Verschickt: Do, 29 Mrz 2012 12:49 pm
Betreff: Re: VLF: DF6NM on 8970.002 - again tonight

My transmission on 8970.002 went off air at 6:35 this morning, and the antenna was taken down for the next couple of days because of windy wx.

 

Unfortunately, the LF harmonic showed that again my frequency was not correct during the first 90 minutes until about 22:30, with a similar behaviour as yesterday, and also during an earlier MFSK transmission (see attached). Immediately after starting SpecLab, the samplerate detector correctly measured the samplerate versus the DHO reference. The lock indicator turned green after the first three blocks, and plausible samplerates were displayed at all times. However the corrections were apparently not applied, and the carrier from the signal generator still drifted with the soundcard crystal. After more than an hour, the carrier suddenly jumped up by about 2 ppm, and within the next 20 minutes settled to the final correctly locked frequency. At first I thought it might be due to some glitch in my LF harmonic measurement or even at DHO38. But it seems to be reproducible so there may actually be a subtle error in the samplerate tracking function. I was using SL 2.77 b08, and will try to look more deeply into this - a simple workaround may be to start and lock SL a couple of hours before operation.

 

Thanks again for watching - it's fun to see the Dreamer's band so populated!

 

73, Markus (DF6NM)

ErrorIntegral.png
Description: PNG image