ADAU1761 output noise problem
We have an ADAU1761 project, which is being booted by an external uC over I2C. We're using line level, differential inputs to the ADCs, with a passive U-pad to bring input signals in under the ADCs max input level. The output from the DACs is also differential, the signals pass through a differential analog filter stage which also converts the signals to single-ended and sends it to the output gain and driver stage. The ADAU1761 internal volume controls are all set to 0dB and for testing purposes the SigmaStudio schematic is simply a straightwire in-out connection through the DSP core. The problem is that the output noise floor is fairly high; quite audible on an average set of headphones. The noise seems fairly unaffected by alterations to the input configuration of the CODEC, but in the output configuration, the line out volume control seems to make a big difference, and each mute / volume control further back (closer to the DSP core) seems to affect the noise level less and less. Is this high noise floor normal for the ADAU1761? Are there any specific methods (close impedance matching etc.) to reduce it? Any help / suggestions would be appreciated greatly.
(Also, the analog section when physically bypassing the CODEC (differenital in bridged to differential out) has an extremely low noise floor, and we're positive is not the source of this noise issue.)
Audio Signal Processors
SigmaDSP Audio Processors
adau1761
As a follow up note, the output noise measured directly at the chip pin on the ADAU1761 evaluation board is ~20mV. Muting or turning the DACs off reduces this noise by about half. This is with or without anything connected to the input of the eval board.
I will try a noise measurement in the lab today or tomorrow to confirm the issue. We test our evaluation boards for analog-to-analog SNR and THD+N, and the test limits are high enough that the test would definitely fail for 20 mV (i.e. -44 dB SNR).
In the meantime, perhaps you could try the same test using the included DC power supply (or even better, a high quality lab supply) to power the board instead of powering it by USB. Some computers have very high noise that is transferred over the USB's +5 V line to the board. Perhaps that could be contributing to the problem.
Message was edited by: BrettG 110104 (added the word "SNR" to -44 dB)
The board is being powered via battery, as designed for the production version of our project, and the CODEC is being booted by a microcontroller, so there are no connections to a computer present in the system. In further testing we've noticed a significant reduction in noise by taking the audio (in line out mode) from the headphone outputs instead of the line outputs, almost as though the noise is coming from the mixers 5/6 or from the output stages themselves.
In your tests, were you able to get a quantitative measurement of the noise floor?
I tried to reproduce a the problem you're seeing, but so far I haven't been able to on my initial evaluation board tests. I'm using a DC-powered eval board and looping the differential outputs back to the differential inputs using fairly long 1/4" TRS audio cables. There is a tone generator in SigmaStudio that I'm using as a test source. The sine tone goes out of the DACs to the differential outputs, and loops through the cables back to the differential ADC inputs. I'm detecting an noise floor of under -90 dBFS. Muting or un-muting mixers 5 and 6 don't seem to have a noticeable effect on the noise floor.
I was able to go over the system today, and double check everything. Measuring the noise floor out of the ADAU1761 directly, I was getting about -84 dB, which is probably fine. The reason the noise is so audible is that our output gain stage is currently set very high, as we aren't getting enough signal level from the output stage of the ADAU1761. Checking the differential input signal to the CODEC (hot - ground) against the differential output signal from the CODEC (hot - ground) I'm seeing signal attenuation of about -14dB. All the volume controls in SigmaStudio are set to 0dB, and as far as I can tell from the data sheet this should produce an attenuation of only -1.38dB. Additionally, adjusting the differential input volume control via SigmaStudio in an attempt to correct this, has no effect on the output level whatesoever. Any ideas on why there's such a significant drop in signal level or, why the differential volume controls do not work? (P.S. Thanks for all your help so far.)
This question has been assumed as answered either offline via email or with a multi-part answer. This question has now been closed out. If you have an inquiry related to this topic please post a new question in the applicable product forum.
Thank you,
EZ Admin
Please log in to reply.