[time-nuts] weird Raspberry PPS+GPS NTP server behaviour

[ASSI]

[sorry for breaking the threading, I'm replying from the list archive since 
gmane.io does not have this list anymore]

Hi David,

[ntpheat]
> I didn't know about that program and I'd like to try it here.  A neat idea.
> Is it available stand-alone - ready to run?  I could compile it here given
> the source, but not if it has dozens of dependencies on NTPsec, which I 
> don't use.

It's actually an implementation of a suggestion by me a few years back, 
however it doesn't take into account that the CPU temperature can not directly 
translate into the crystal temperature and it does not regulate even the CPU 
temperature very well.  I've since refined that concept to use the CPU load 
and CPU temperature to correlate that back to the actual crystal temperature.  
It takes a bit of time to tune it up correctly, but it turns out that the 
rasPi I have are all very similar, so if you start with a generic set of 
constants you don't really make much of an error.

I've posted results of that setup last year on July 22nd:
http://lists.febo.com/pipermail/time-nuts_lists.febo.com/2019-July/097218.html

The control loop is a Perl script that needs two standard Debian modules 
installed on top.  If you're interested, you can mail me directly.  I have two 
TinkerBoard that have better clocks (less noisy on the medium term) and I can 
usually keep them within 5ppb over a week or around 1ppb per two days (with a 
linear drift removed).  These however need a custom kernel since the turnover 
point for their crystal ios too close to the standard thermal throttling 
threshold.


Regards,
Achim.
-- 
+<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+

Samples for the Waldorf Blofeld:
http://Synth.Stromeko.net/Downloads.html#BlofeldSamplesExtra