To: rsgb_lf_group@blacksheep.org
From: Paul Nicholson <vlf0403@abelian.org>
Message-ID: <dca12b9c-c401-1338-38ff-feed1f7088db@abelian.org>
Date: Thu, 25 Jan 2018 06:12:14 +0000
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101
 Thunderbird/52.3.0
MIME-Version: 1.0
Content-Language: en-US
Subject: LF: Testing Audioinjector Octo with RPi3
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 7bit
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org


I've got an Octo/RPi3 combination under test at the moment
with vlfrx-tools version 0.8b.

Running with 6 input channels at 96k frames/sec 24 bit samples.

For testing the stability, I'm using vtnspec to measure the
phase of a 19531.25 Hz signal from a Trimble GPSDO.  It logs
the phase once per second.

To make it a tough test of vttime, I've hobbled the GPS that
provides the PPS to the Octo.  I've configured it to not output
a PPS if it doesn't get a fix, and I've put the antenna in a
poor location so that it often doesn't get a fix.

By default the GPS will 'hold-over' and in the absence of
a position fix it will output its best guess.  With this
disabled, vttime frequently sees (or rather doesn't see)
missing PPS pulses, often several in a row.  It reports
'insufficient PPS peak/mean ratio' in its log file.

Effectively, vttime is having to do the hold-over instead of
the GPS.  Which is better? I'm not sure yet.

Here's a plot of measured phase on an 18 hour run

  http://abelian.org/vlf/tmp/180125a.gif

Overall the measured phase has standard deviation 5.9 degrees
and vttime is reporting a mean absolute difference of PPS
intervals of between 250 and 350 uS.  Not particularly good,
we would like to see sub 100uS here, and less than 0.5 degree
jitter on the reference signal phase.

On a couple of occasions, vttime failed to maintain the timing
during long periods without PPS and the phase went a bit wild
for a minute or two but then recovered quickly.

Actually the results are good enough for most work, including
amateur radio.   A professional SID monitor would not like
the residual standard deviation though.

This particular Octo (the 3rd one I bought) is showing quite
good sample clock stability.  Here's the plot from the program
vttimeplot,

  http://abelian.org/vlf/tmp/180125b.gif

So it is no surprise that vttime is holding over missing PPS
without much trouble.  I need to test the other two Octo and
arrange more temperature variation to make it a tougher test.

As it stands, this particular Octo could be used with longer
vttime time constants for reduced phase jitter.

Conclusion: encouraging but more work needed.

--
Paul Nicholson
--