AD1372 - does DAC sample rate follow ADC rate?
I had a '1372 configured as 96K throughout and found the dB/Hz response plot as expected (effective low pass @ around 22K) but when I configure as 96K sampling and 48K I2S bus the response plot seems to have a low pass @ around 10K - this is sort of an aside to my question title but it was while looking for anything I could change to push that knee upwards in hertz that I was led to notice the lack of certainty about DACs - any tips here appreciated.
Whilst reviewing the datasheet I was challenged by a knowledgeable friend to confirm I was configuring the chip correctly as I truly intend and found myself at a loss to absolutely guarantee the DAC sample rate - ADC to DAC Talkthrough implies well enough that the sample rate is same but the DAC sample rate isn't mentioned in the details of any registers and I cannot find where it is mentioned elsewhere either. ADC can configure as 96K or 192K but
Happy enough to take the word of an AD rep but happier still if someone can point out documentation that clarifies this absolutely - I'll be a little embarrassed if it is in a document I have already referred to (or should have found easily enough) but terribly grateful either way.
adau1372
Integrated/Special Purpose A/D Converters
adau1372 sampling rate(s)
Audio CODECs
OK. DaveThib says the DAC sample rate is 192K - https://ez.analog.com/thread/81552#comment-244467
Hello robbo1970,
Thanks for finding the earlier post I wrote. It should have pretty much explained the things you were asking.
I am curious about the 10kHz LPF you mentioned. Some of our older parts have an emphasis bit that can cause these sorts of issues but this part does not have that feature. Did you figure out where this loss was coming from? It could be an output/input impedance issue. Are you using the PGAs?
Also, make certain you have the PLL and clocking registers setup properly. I did mention this in my earlier post. SigmaStudio has no way to actually know what the sampling rate is inside the part. This is all done with register configuration and the actual frequency of the clocks coming into the part.
Dave T
Thanks Dave, I use a 12.288MHz oscillator so PLL is not engaged. The (seeming) LPF was a result of the way I used a pair of Variable State Filter objects to make my tone controls; the Low Pass of the first one forms Bass, the High Pass of first one goes to second one and Low Pass of second one forms Midrange and then the High Pass of second one is forming Treble; when I removed 'tone controls' from my project I got to see the (nearly) brick wall LPF at about 23K I was expecting all along. I learned that IIR/Biquads (integer scaled coefficients, S1.30) have numerical precision problems at lower frequencies when I tried to make tone controls using those and now I have found out that State Variable Filters have problems at higher frequencies - I can put a Biquad in parallel with the SVF filters and derive a High Pass for Treble from that.
Please log in to reply.