Hi,
We are using the McAsp on the Am335x to drive a '3104. 3104-master, 335x-slave.
The output from the '3104 has a dead band/area/gap that is ~5ms wide and repeats every ~85ms (12Hz).
The mclk is 24.756M, wclk is 48K, bclk is 1.536MHz and are continuously running.
The 85ms period is ~4096 periods of the 48K wclk.
The signal is active/dead for ~3844/252 periods (o-scope measurements) - 3840/256 more likely.
Can't be sure as signal returns to zero - could be forced by DAC vs data.
No error flags are set in the 335x McAsp status registers.
Same results with 48k or 8k sampled signals. Used 1KHz test tone pattern/file, and the portion that shows up appears to be at freq.
To me it seems like we are loosing data - but not sure how this could be the case. It is very consistent.
And if data were being lost, I would think that it would affect the signal more - not sure...
Maybe it is and I'm just not using enough resolution to detect the difference. I will try to measure it more accurately.
I'm hoping someone has seen this behavior before and/or might be able to clue me in as to what to look for.
Not sure what other info is pertinent. But ask away if your need more info.
Thanks,
Matt
