Thanks Paul!

I had played around with gain a lot before adding that last bit to the 
differential algorithm.  A word about that:  I 
developed my code without a lot of reference to the PID work by 
Wescott.  I had learned how to do the integral part a different way and 
hadn't realized it.  So, it makes sense that things would go badly wrong 
without an "iTerm".

So, I cut the dTerm in half and divided the pTerm by about 3.5.  I left my 
integral factor the same.  This is only a 30 minute run, but the difference is 
remarkable.  I'll play with the gains a bit more to see what happens.

http://evoria.net/AE6RV/TIC/GSPDOe.2.png
http://evoria.net/AE6RV/TIC/ADEV2.png 

I'd still like to know whether the tools I have at hand can measure the phase 
noise on my OCXO.

Bob

(And yes, I accidentally responded directly to you again so this is a repost to 
the list.)



________________________________
 From: Poul-Henning Kamp <[email protected]>
To: Bob Stewart <[email protected]> 
Sent: Wednesday, August 27, 2014 2:02 PM
Subject: Re: [time-nuts] OCXO Phase Noise Measurement in Primitive Conditions
 

--------
In message <[email protected]>, Bob Ste
wart writes:

Your PLL (PID) has (far) to high gain:  You are yanking the EFC around
to every single little wiggle in the GPS signal.

The curve relative to the Rb should be very, very straight, any
wavyness should be on timescales of at least 20-30 minutes, not
within single minutes as your first plot shows it.

This is a very common beginners fault:  I made it myself too :-)

Lower the PID gain, until the DAC/EFC line only wiggles the minimum
amount necessary to keep the OCXO from drifting away from the Rb...


-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
[email protected]         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.
_______________________________________________
time-nuts mailing list -- [email protected]
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Reply via email to