Noise target for ADAU1701 (AVDD)
Hello,
I understand that less noise is always better. But is there any target level how low the noise in supply pins should be to be 'sure' that it does not degrade the performance?
I assume that the noise in AVDD is the most important to have under control and the noise in IOVDD, PVDD and DVDD does not cause problems so easily. Is that right?
Basically I try to figure out how good filtering I need to create for SMPS powering up ADAU1701 and MCU. It works on 100-150kHz range and creates saw tooth kind of ripple of course.
I'm doing my own investigation also on this and will post some results as I have them.
Thanks,
Ari
Audio Signal Processors
noise
SigmaDSP Audio Processors
adau1701
AD's eval board uses not merely a garden variety 3.3V regulator to power the -1701, but a ADP3336 low-noise linear regulator. Apparently they felt that low power supply noise would show off the -1701 with best advantage. It will be interesting to see how much of that is really necessary, thus with you I'm looking forward to both what ADI has to say as well as what you find.
Very interesting question Bob. Last but not least because the ADP3336 is a nightmare to find in the UK and it costs a small fortune.
I look forward to reading AD comments about your post.
3.3V
Hi Ari,
Your question is quite an interesting one, and merits a more-insightful answer from ADI's designers. I'll share my experience as a first guess. Buried in the -1701 data sheet is a PSRR spec of 50dB from the AVdd supply. We can set the full-scale DAC output of approx. 2.5V p-p as the zero dB reference. This allows for the following estimation:
Desired signal to noise ratio: -100 dB
PSRR from AVdd to ouput: +50 dB
Max. allowable AVdd ripple: - 50 dB below 2.5Vp-p = 8 mV p-p
Thie PSRR is specified only at 1KHz so this estimate may not be exact for other ripple frequencies, although of course no one will hear ripple that's out of the audio range.
I tried some measurements on a prototype pcb where a "simple switcher" buck regulator powers both the ADAU1701 and a PIC microcontroller. Standard filter components were used except for a slightly oversized inductor to keep its p-p current ripple lower than strictly necessary. No second-stage filter was used other than the usual bypass caps at appropriate chip pins.
3.3V supply ripple, about 260 KHz (2mV, 10uS/div):
Output (at product RCA jack) (2mV, 10uS/div)
The output hash shown is way out of the audio range (I forgot to BW limit my scope). It also shows no feature of the regulator ripple, so it appears that the buck regulator's ripple is well-tolerated here. It appears to show instead a 50 KHz ripple from outside my board -- perhaps noise from the switching power supply in my scope?
My application is in telephone conferencing; likely you're looking for much better noise performance in your application. AD's eval board uses not merely a garden variety 3.3V regulator to power the -1701, but a ADP3336 low-noise linear regulator. Apparently they felt that low power supply noise would show off the -1701 with best advantage. It will be interesting to see how much of that is really necessary, thus with you I'm looking forward to both what ADI has to say as well as what you find.
Best regards,
Bob
Thanks, Bob, for doing this test. I think your calculation is accurate.
We basically only test PSRR at one frequency, 1 kHz, so that spec is certainly not guaranteed at other frequencies, but in general the PSRR is pretty stable across the frequency range.
The ADP3336 was chosen for this board to provide the user with the best performance possible so that the ADAU1701 would be the "weak link" in the system performance. It's also free advertising for an Analog Devices product!
I think if you limit the power supply noise to -50 dB, as Bob mentioned, you should not see any significant noise components on the audio signal.
Please log in to reply.