[time-nuts] BBB DDMTD - analysis
subscriptions at burble.com
Wed Oct 29 16:16:17 EDT 2014
This is the second post related to my attempts to create a BBB DDMTD,
the first post had a load of graphs showing details of what was
happening around glitch periods. This post is about interpreting those
Pretty obvious from only a cursory glance at the graphs are that
transitions are not continuous and are happening at very specific
intervals. I would appreciate some help in understanding how this might
be occuring or how to find out what might be going on as I'm currently
at a loss.
I've tried a few different combinations of oscillators to try and rule
them out and this seems to have no impact. It would seem the effect is
coming from the sampling / BBB bit.
There are 169 clocks/samples between the larger spikes, this corresponds
to ~16.9ms real time or a ~110ps interval at the DUT. The smaller spike
is 94 clocks/samples after the larger spikes (=~9.4ms / ~62ps). The dead
time with no transitions following a smaller spike is 75 clocks/samples
I'm not aware that there is any relation of these intervals to anything
else in the setup.
I'd welcome any thoughts on what might be going on here.
More information about the time-nuts