I had a bit of time to play again with the Octo, to investigate
where the excess timing jitter comes from.

The stereo audio injector is giving around 20nS jitter on the
PPS from the neo-m8t.

With the stereo card replaced by the Octo, all else the same,
the jitter varies between 300nS and over 1uS.

A close look at the shaped PPS seen through the Octo shows some
distortion or interference

 http://abelian.org/vlf/tmp/180217a.gif

Not sure if these are additive or little step changes in gain.

Here's worse example

 http://abelian.org/vlf/tmp/180217b.gif

Always the interference is on the rising and/or falling slopes
about half way.

Not present on the PPS when viewed on a scope at the Octo inputs.

Similar artifacts are visible on a plain old sine wave fed into
an Octo input.

 http://abelian.org/vlf/tmp/180217c.gif

For some reason on the sine wave the artifacts are always near
the turning points.

Meanwhile I monitored the 49.152 MHz master oscillator on a
SSB receiver and it was giving a nice pure note with no obvious
frequency or phase jitter.

So it looks like the timing jitter on the Octo is due to its
analog inputs, not the A/D conversion timing.

Strangely though, there's no sign of these artifacts when there's
no input to the channel.  It isn't just background interference.

I think the next step might be to bypass the Octo's input board
and take the PPS straight to the pins of the ribbon connector.

Wondering if anyone else sees this behaviour on a PPS or sine wave.
--
Paul Nicholson
--