I am using a PCM5122 clocked with a 24.576MHz oscillator in an application that must produce fairly short bursts (<5ms) of broadband, ultrasound (15-90kHz) audio at 192kHz sample rate. The design I'm working on is modeled on the HifiBerry DAC Pro. I've noticed a correlation between the frequency of the external oscillator I use and both the spectral resolution of the output and the length of the smallest output of any kind I can produce. In fact, below about 3ms, the DAC will not produce any meaningful levels of sound at all when driven at 24.576MHz. This is true whether or not I pad the output signal with relatively long stretches of silence.
My testing setup uses a Raspberry Pi with several layers of software abstraction from the DAC hardware, so a solution to my entire problem is probably out of scope for this forum. My question for this forum is whether there is a factor besides oscillator frequency at the hardware level of the DAC that fundamentally determines the minimum length sound it can produce. This might take the form of internal click protection, a ramp-up/ramp-down setting, etc. This might apply to this specific IC or to DACs as a class of device.
Thanks for any help you can provide — I am RTFM but hoping to get some guidance if possible as I am a student and fairly new to this topic.