[time-nuts] Help understanding an ADEV

Bob Stewart bob at evoria.net
Thu Sep 11 13:51:43 EDT 2014

Hi Bob,

After I posted last night, I realized that the loop was running, so I froze 
the DAC and did another sample.  Here is the voltage plot in 10s of uV, 
as well as the ADEV.  I can pull the op-amp out of the GSPDO engine and 
plot it without the DAC attached if that would be helpful.  There is a 
voltage-divider at the OCXO attached to the VRef output from the OCXO.  As always, I use Tom's adev1 software.


I don't understand what you are asking for to do an ADEV of the GPS.  What do I plot and under what conditions?

I've been wondering if it would be better to look in the frequency domain.  
I'll have to look at Tom's site to see if he has code to do that.

 From: Bob Camp <kb8tq at n1k.org>
To: Bob Stewart <bob at evoria.net>; Discussion of precise time and frequency measurement <time-nuts at febo.com> 
Sent: Thursday, September 11, 2014 6:10 AM
Subject: Re: [time-nuts] Help understanding an ADEV


The plot pretty much confirms what you already know - there is a noise spectrum and it’s not what you want to have. In a closed loop system, it’s all going to add up. The big unknown is - do you get noise out of the DAC trying to correct the OCXO’s noise or just noise out of the DAC? Is the DAC noise from the DAC it’s self or from the loop?

One source of noise is the LSB on the DAC. If you have a 1 ppm EFC and a 1x10^-12 OCXO and a 1 LSB change at 1 second (yes that’s a lot …) - That’s a 1x10^6 delta, you must have a ~20 bit DAC. If you have 1x10^-8 out of your GPS,  your loop (or filter) must have a 10,000:1 attenuation at 1 Hz. That’s 80 db. If you have a simple loop, you get 20 db / decade. That would put your tau below 0.0001 Hz.

Yes those are fast and simple numbers. The real stuff has a lot more if’s and’s and but’s in it.  

Simple way to check things:

1) Do an ADEV on the OCXO + EFC and make sure it looks right outside the loop.
2) Do an ADEV on your GPS and figure out where it’s at. 
3) Consider checking the noise in the frequency domain, it’s easier to understand.

So much fun.


On Sep 10, 2014, at 11:06 PM, Bob Stewart <bob at evoria.net> wrote:

> After spending a lot of effort trying to get some useful tuning figures for the PID in my GPSDO engine, I decided to capture the DAC output with my 3456A.  And, of course I've made an ADEV plot.  If I understand this correctly, it means that there is mostly justnoise out to 2 tau, and an area from about 4-5 tau is mostly noise, as well.  Could someone tell me if I have this right?  It matches what I'm seeing on the delta plot, and would explain why we couldn't get anything approaching reasonable stability from the OCXO, since the noise is larger than the increments being made to the DAC.
> I've ordered the PWM version of the PIC, and hopefully, since it's the motor control version (as opposed to the audio version) it will have much better noise performance.
> The red scatter is the EFC measured at the OCXO in tens of microvolts, and the blue line is the ADEV.  I'm using a short shielded twisted-pair with mini-clipsas the probe.  Hopefully I've got things scaled properly and have run the test properly.
> http://evoria.net/AE6RV/TIC/DAC.wander.png
> Bob - AE6RV
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.

More information about the time-nuts mailing list