#### Measuring frequency stability - simplest approach?

Bruce KX4AZ

All of the recent discussion about FST4W decoding and the need for an external GPSDO clock has me thinking about the easiest way to measure the "native" stability of the various SDRs I have at hand, including the Kiwis.  The simplest approach I can think of is to set a Bodnar GPSDO to output a standard frequency in a quiet area of the HF spectrum (something close to 30 MHz) and let enough of that into the SDR to get a good S/N ratio.  Then set the receive frequency to be something like 1 kHz lower than the Bodnar, in USB mode, and pipe the audio into Spectrum Lab software.  I have barely used Spectrum Lab previously, but I assume it has the ability to make graphs of the received audio frequency over time. In the case of the KiwiSDR I also assume that I would need to disconnect the GPS antenna if I want to observe the native stability/accuracy of the oscillator.

Appreciate any feedback about whether this simple setup is reasonable to measure the oscillator stability, or if I am missing any key aspect.

Rob Robinett

Hi Bruce,

I think your technique will work well   Gwyn has used fldigi to obtain high resolution audio spectra and I’m told that Spectrum Labs can also perform those calculations

However to evaluate an SDR for its suitability for FST4W reception, the sprectral width value which can be obtained from WSJTX is the most meaningful metric.

Rob

On Thu, Aug 18, 2022 at 11:45 AM Bruce KX4AZ <bruce@...> wrote:
All of the recent discussion about FST4W decoding and the need for an external GPSDO clock has me thinking about the easiest way to measure the "native" stability of the various SDRs I have at hand, including the Kiwis.  The simplest approach I can think of is to set a Bodnar GPSDO to output a standard frequency in a quiet area of the HF spectrum (something close to 30 MHz) and let enough of that into the SDR to get a good S/N ratio.  Then set the receive frequency to be something like 1 kHz lower than the Bodnar, in USB mode, and pipe the audio into Spectrum Lab software.  I have barely used Spectrum Lab previously, but I assume it has the ability to make graphs of the received audio frequency over time. In the case of the KiwiSDR I also assume that I would need to disconnect the GPS antenna if I want to observe the native stability/accuracy of the oscillator.

Appreciate any feedback about whether this simple setup is reasonable to measure the oscillator stability, or if I am missing any key aspect.

--
Rob Robinett
AI6VN
mobile: +1 650 218 8896

Gwyn Griffiths

Hi Bruce

The technique you propose will work, but the short-term jitter may depend on the exact detail of how you intend to "pipe the audio".

If you were to use a browser to access your Kiwi and pipe its audio into another application, e.g. using Pulseaudio, then be aware that jitter will occur due to buffering and non-exact audio rates between the modules in the application chain. If you have to use a browser then setting abuf=0.5 in the Kiwi url will reduce, but not eliminate, this jitter. One better route is to use kiwirecorder.py to record a wav file with a narrow bandwidth, say +/- 20 Hz and replay that into the frequency determining app.

The frequency cal mode within WSJT-X is also useful, listing to 1 milliHz on the screen, but is not necessarily at regular intervals.

regards
Gwyn G3ZIL

Bruce KX4AZ

Gwyn/Rob,

Thanks for the feedback and for the alert about audio jitter.  I have used the WSJT-x cal mode in the past to check frequency accuracy but I surmised that only Spectrum Lab would have the ability to graph the data and summarize the results.  This gives me a good reason to figure out how to use it for that; it seems very "feature rich", indeed.

Bruce

Jim Lill

Depending on what you're doing.... sound card stability can be an issue. On my day job, we look very critically at related stuff and have been thwarted at times by drifty sound capture gear. The current Focusrite gear, aimed at the professional musical community is very stable.

-Jim

WA2ZKD

On 8/19/22 07:28, Bruce KX4AZ wrote:

Gwyn/Rob,

Thanks for the feedback and for the alert about audio jitter.  I have used the WSJT-x cal mode in the past to check frequency accuracy but I surmised that only Spectrum Lab would have the ability to graph the data and summarize the results.  This gives me a good reason to figure out how to use it for that; it seems very "feature rich", indeed.

Bruce

Bruce KX4AZ

On Fri, Aug 19, 2022 at 09:11 AM, Jim Lill wrote:

Depending on what you're doing.... sound card stability can be an issue. On my day job, we look very critically at related stuff and have been thwarted at times by drifty sound capture gear. The current Focusrite gear, aimed at the professional musical community is very stable.

https://focusrite.com/en/audio-interface/scarlett/scarlett-solo

Jim,
Excellent point, and I appreciate the link to that stable sound interface.  What occurs to me is that I can set the Bodnar GPSDO to the same audio frequency I intend to use (later) for oscillator stability measurements, connect it directly to the sound card input, and observe the stability/accuracy over a period of time,  That should give some good information about its performance before I put it to work with the SDR measurements.
Bruce
Bruce

Jim Lill

The Bodnar to soundcard method is how we assessed the Focusrite which is relatively inexpensive. Most internal soundcards are not very good, nor are dongles.

On 8/20/22 08:49, Bruce KX4AZ wrote:

On Fri, Aug 19, 2022 at 09:11 AM, Jim Lill wrote:

Depending on what you're doing.... sound card stability can be an issue. On my day job, we look very critically at related stuff and have been thwarted at times by drifty sound capture gear. The current Focusrite gear, aimed at the professional musical community is very stable.

https://focusrite.com/en/audio-interface/scarlett/scarlett-solo

Jim,
Excellent point, and I appreciate the link to that stable sound interface.  What occurs to me is that I can set the Bodnar GPSDO to the same audio frequency I intend to use (later) for oscillator stability measurements, connect it directly to the sound card input, and observe the stability/accuracy over a period of time,  That should give some good information about its performance before I put it to work with the SDR measurements.
Bruce
Bruce

 1 - 7 of 7