-
-
Save ericfont/86d55909d121bbe087f6b11d9b569a93 to your computer and use it in GitHub Desktop.
retry exactly 200kHz oscillation peak
(obviously need longer simulation time to get more accuracy for above)
raw PWM output without integration:
if apply sinc filter with nulls every multiple of 275kHz, and then resample at twice that, then the 275kHz components will be very very small when folded into baseband audio.
Alternatively could try to null the peaks at 200kHz and 370kHz.
Or could have nulls at 100kHz, 200kHz, 300kHz, 400kHz, etc. Need to investigate and see what noise distribution looks like.
if apply a simple-moving average with the null point being 275kHz:
if instead null multiples of 200kHz:
sinc1:
sinc2:
sinc3:
sinc4:
sinc5:
Possibly the lowpass enveloped matched an integrator in low freqs, then wouldn't need to apply integrator again. Maybe a lower null would match it better.
Could send raw pulse widths or could downsample before transmitting.
To do a DAC from 40kHz first could zero expand to 200kHz, and then could simulate the circuit digitally, which would generate a PWM that could then in analog domain be integrated.
sim