Kiwi v1.554 adds GPS clock correction synchronization with WSPR cycles
|
Rob Robinett
I owe much thanks to John Seamons for his quick response to my suggestion of this new feature
Since corrections to the Kiwi's clock which occur during a WSPR cycle would appear to be drift and/or interference to the wsprd decoding algorithm, I suggested that John make it possible for the Kiwi administrator to limit such corrections to the interval between WSPR cycles. On the Admin->Config page you can now select the cadence of when such corrections are applied. It will take some testing to determine if, or how much, moving changes to those WSPR cycle gaps improves SNR and spot counts. It seems likely that if there are improvements with its use they will be most noticeable on 20M and above and more on the FST4W-300/900/1800 modes than WSPR-2 Those of you with multiple Kiwis attached to the same antenna are well equipped to experiment and report to us and John the results.  |
|
|
|
I've always wondered about the correction algorithm and whether it using a floating average to make adjustments, or just the average of every single measurement since the last reboot. In other words, it seems like a floating average of say, the most recent 100 GPS measurements would be more effective at tracking any gradual drifts in the oscillator frequency, rather than a cumulative average from every single measurement..
Regardless, it is nice to know there is a new menu option to fine tune when the adjustments are made. |
|
|
|
WA2TP - Tom
Following. My question would be how many gps fixes would be adequate for an accurate lock? Then it would become a matter of your GPS Antenna setup which would determine just how long it would take for a specific site to acquire that number of fixes (the more sats you can acquire the faster you reach this number).
Do we know if the “refresh” of this reapplication of the floating average interrupts the wave file audio in any way? Perhaps an audio artifact or hiccup of sorts would interrupt/corrupt the cycle when it updates?
Sent from Mail for Windows
From: Bruce KX4AZ
Sent: Wednesday, August 17, 2022 6:06 AM To: wsprdaemon@groups.io Subject: Re: [wsprdaemon] Kiwi v1.554 adds GPS clock correction synchronization with WSPR cycles
I've always wondered about the correction algorithm and whether it using a floating average to make adjustments, or just the average of every single measurement since the last reboot. In other words, it seems like a floating average of
say, the most recent 100 GPS measurements would be more effective at tracking any gradual drifts in the oscillator frequency, rather than a cumulative average from every single measurement..
|
|
|
|
Glenn Elmore
Bruce, From JKS' recent comments, the Kalman filter that is already implemented in the Kiwi is very useful for filtering typical GPS output. I made measurements of it's positional accuracy and approximate BW here at the house by stepping the physical location ~100'. I found that it took on the order of 20 minutes to settle but came very close to the same (I presume, correct) answer every time. JKS mentioned that this filtering keeps the Kiwi from making sudden frequency jumps on bad data. If you are interested perhaps you should study the source code created to implement it. Having said that, It is becoming apparent that the Kiwi GPS-aiding is not sufficient to produce adequate short term stability in a Kiwi for some FST4W modes. They short term errors end up increasing spectral spreading so much that SNR plummets and decodes become impossible. Perhaps more to come on this subject. Glenn n6gn
On 8/17/22 04:40, WA2TP - Tom wrote:
|
|
|
