[time-nuts] Re: ADEV interpretation

[marek dorsic]

Hello,

    thank you all for valuable comments. Luckily I can read also german :)

In between I have done a second, longer measurement. I see some serious
diurnal cycles that may be related to temperature changes (sunny
south-facing room) or GNSS, but I suppose this will affect the measurement
at 12h period not 24h. The counter is not calibrated, having an 0.6Hz
offset from nominal, but this should not compromise the stability
measurement as of my understanding.
Nevertheless I can see the noise floor of my internal counter oscillator
against ZED-F9T at 2x10^10. I see can see this figure again and again by
offsetting an 50k sample window. (Have you ever seen an animated the
ADEV?). So the lastly posted 5x10^10 figure must have been with not well
heated counter or some other error. However I still do not fully understand
why I can’t see any changes when applying the quantisation error correction
from UBX-TIM-TP message. I use allantools python module for calculating
ADEV.
Please find attached also proposed noise floor measurement.

Cheers,
    .marek


On Sun, Apr 4, 2021 at 3:32 PM Magnus Danielson <magnus at rubidium.se> wrote:

> Hi,
>
> The 1/tau slope in ADEV is... a bit messy to explain. Often it ends up
> being an limitation in the instrument. The simple rule is that it is
> single-shot resolution limit. The fine-grained detail is that it is a
> combination of resolution quantization and that of noise. I did research
> on it, and it does some funky stuff. I should redo that one into a more
> readable article some day, to many loose ends to clean up and explain.
>
> Anyway, most often this instrument noisefloor is of great annoyance.
> Often, as ADEV breaks away from it you start to see the ADEV of the
> reference and source. The actual instrument will also have some
> systematics that show up, but usually they are not too strong.
>
> As you do MDEV and TDEV, the slope of white noise and indeed
> quantization noise goes steeper, so you can get real measurement values
> for lower taus that way.
>
> Cheers,
> Magnus
>
> On 2021-04-04 13:52, Marek Dorsic wrote:
> > Hello,
> >
> > I would like to ask your knowledge to help me interpret this basic ADEV
> result.
> > Assume following measurement setup:
> > PPS output of Ublox RCB-F9T connected to channel 1 of a HP53131A counter.
> > 10MHz output of the counter's internal oscillator (opt. 010, medium
> stability oven) divided with PICDIV to 1 PPS connected to channel 2 of the
> counter.
> > Counter set to measure time interval from channel 1 to channel 2.
> > After processing the readings, the floor of ADEV is 5x10^-10 at about
> 30-100 seconds interval.
> > Which of the following is true?
> > a) ADEV shows the real performance of the internal oscillator,
> > b) ADEV floor is 5x10^-10 because of the restricted counter time
> interval resolution (500ps),
> > c) ADEV floor is 5x10^-10 because of the ZED-F9T PPS jitter (after saw
> correction) of the gate start input channel.
> >
> >
> >
> > Thanks in advance
> >      .marek
> >
> >
> > _______________________________________________
> > time-nuts mailing list -- time-nuts at lists.febo.com -- To unsubscribe
> send an email to time-nuts-leave at lists.febo.com
> > To unsubscribe, go to and follow the instructions there.
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com -- To unsubscribe send
> an email to time-nuts-leave at lists.febo.com
> To unsubscribe, go to and follow the instructions there.
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: adev.png
Type: image/png
Size: 102238 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20210405/873f5974/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: tdev.png
Type: image/png
Size: 128925 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20210405/873f5974/attachment-0001.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: NoiseFloor.png
Type: image/png
Size: 68055 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20210405/873f5974/attachment-0002.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: animated-adev-50k.gif
Type: image/gif
Size: 1912759 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20210405/873f5974/attachment.gif>