Looking at Mediumwave Carriers, and Frequency stability
I have been experimenting with generating high resolution frequency spectra of MW channels using recordings generated with kiwirecorder.py, for identifying stations based on frequency offset, a la my Carrier Sleuth program. I was originally looking at making a KiwiSDR extension, but decided my own python script was the path of least resistance 
It's promising, and so far it looks like the KiwiSDR frequency stability will be adequate for the task, ideally 0.1 Hz or better. So yeah, that is 0.1 ppm or better. Hoping for better. I suspect to a large degree the stability is going to depend on the GPS signals used correct A/D clock timing drift. In the case of my KiwiSDR I have fairly good GPS reception (after many trials and tribulations which I have documented here), I have a high gain external GPS antenna and usually have 11 good satellites.
I recall several threads in the past here regarding KiwiSDR frequency stability, but have lost track of what the current expected value(s) are. I'm curious as to what I can expect, under ideal (or nearly ideal) conditions.
It's promising, and so far it looks like the KiwiSDR frequency stability will be adequate for the task, ideally 0.1 Hz or better. So yeah, that is 0.1 ppm or better. Hoping for better. I suspect to a large degree the stability is going to depend on the GPS signals used correct A/D clock timing drift. In the case of my KiwiSDR I have fairly good GPS reception (after many trials and tribulations which I have documented here), I have a high gain external GPS antenna and usually have 11 good satellites.
I recall several threads in the past here regarding KiwiSDR frequency stability, but have lost track of what the current expected value(s) are. I'm curious as to what I can expect, under ideal (or nearly ideal) conditions.
Comments
That said, I spent some time this morning gathering data from various online KiwiSDRs, and it seems roughly half have zero or perhaps 1 or 2 GPS satellites, so I did not use them, figuring they would be far too inaccurate.
Could something like this also be used to sort multiple NDB's using the same nominal carrier frequency, but using slightly different modulation frequencies ?
Regards,
Martin - G8JNJ
To verify that accuracy, I use the 20M ground wave transmissions of K6PZB which are locked to 14.097090 by a GPSDO.
The 'wsprd' utility program in WSJT-x outputs spot frequencies to .1 hertz resolution, but wsprnet.org rounds the uploaded spots to the nearest 1 hertz, so I can't show the exact rx frequency uploaded by the wsprdaemon program running at KPH.
To retain the full frequency accuracy of wsprdaemon rx sites, my next version will include the options to also upload spots to a logs.wsprdaemon.org research strength cloud database server which users can query and graph using SQL.
Glenn N6GN has achieved even greater Kiwi accuracy by feeding his Kiwi from a Leo Bodnar GPSDO. With that clock one can view on the Kiwi's IQ display milliherz drift at 30 Mhz. Hopefully, with more GPSDOed Kiwis and a database which records spots to full accuracy, we can learn more about propagation from the spot database.
====================
pi@KPH-Pi4b-85:/tmp/wsprdaemon/uploads.d $ grep K6PZB successful_spot_uploads.txt | head; awk '/K6PZB/{print $6}' successful_spot_uploads.txt | sort | uniq -c
200120 0420 3 -23 0.2 14.097090 K6PZB CM88 37 0 1 0
200120 0432 1 -21 0.2 14.097090 K6PZB CM88 37 0 1 0
200120 0442 3 -23 0.2 14.097090 K6PZB CM88 37 0 1 0
200120 0454 3 -23 0.3 14.097090 K6PZB CM88 37 0 1 0
200120 0502 3 -22 0.2 14.097090 K6PZB CM88 37 0 1 0
200120 0512 3 -24 0.2 14.097090 K6PZB CM88 37 0 1 0
200120 0522 3 -22 0.2 14.097090 K6PZB CM88 37 0 4 0
200120 0540 3 -23 0.2 14.097090 K6PZB CM88 37 0 1 0
200120 0550 3 -21 0.2 14.097090 K6PZB CM88 37 0 1 0
200120 0612 3 -24 0.2 14.097090 K6PZB CM88 37 0 1 0
135 14.097090
5 14.097091
pi@KPH-Pi4b-85:/tmp/wsprdaemon/uploads.d $