[time-nuts] BBB DDMTD - analysis
subscriptions at burble.com
Fri Oct 31 07:11:34 EDT 2014
Many thanks to everyone for your suggestions and patience.
Short answer, the problem was power related, but more accurately, it was
a shielding problem.
Long answer, it was not immediately obvious as power in the setup itself
didn't seem to be a problem. I need 12v (for the mv89a or 8663 OCXO), 5v
(for the MicroCrystal OCXO and to feed the BBB) and 3.3 for logic. Each
section has its own LDO with a sprinkling of bypass capacitors and these
are fed from a single power supply in a star. I'd already tried
different power supplies and monitoring the LDO outputs for noise.
After pretty much swapping everything out yesterday and not seeing any
real change to the problem, it was becoming clear that I needed to look
a bit wider. The hint that ~60khz related to switching power led to
tracking down a nearby DC step-down converter as the cause.
No doubt there are lots of thoughts of 'well, that shouldn't be a
surprise working on a shoddy breadboard', but note that I only did this
to narrow down what was causing the problem after initially seeing it
with the proto cape. The cape setup should have been more robust, so I
need to look at this closer to see where I'm lacking appropriate shielding.
I'm happy to have found the problem, and all the analysis yesterday has
given me lots of data about the impact of different components and
stability of the edge transition, so it's all good stuff :)
More information about the time-nuts