Is the UcD modulation scheme less than optimum?

[sovadk]

Yes and no. Yes, the gain is not unity for frequencies right up to the switching frequency, but no, it still must be unity for the switching frequency itself. This is the surprising gain jump I mentioned earlier.

Attached is a graph of the measured gain of the comparator in UCD type of modulator.


The modulator is oscillating at 1Hz. The measurement is done in the following way:
The input to the modulator is swept with a sine wave and the ration between the amplitude of the fft tones for the carrier and the pwm is plotted (at the current frequency of the sweep).

The graph shows that the gain is constant below the modulation frequency and rises just at the modulation frequency. This is where you get the two gains from. The low frequency gain is what we're interested in when we want to examine the loop/noise-transfer-function/frequency-response.
This low frequency gain is typically 6dB lower than what is needed to make the loop gain unity and explain why the whole thing will oscillate.