For a streamer based on a Raspberry 4, I bought two DAC hats - a HifiBerry DAC2 pro and an AUDIOPHONICS DAC I-Sabre ES9038Q2M.
I measured THD and noise on both hats with a Motu M4 and an E1DA COSMOS ADCiso. Software: REW and AudioTester. For playback I used LibreELEC and Moode. The test signal was a wav file with a 24-bit sine (sampling frequency 44.1 kHz, level 5 dB below FS).
While the noise floor was quite low with both DAC, I was a little surprised with the comparatively high THD value of around 0.04% for both hats. For comparison: the distortion of an old Roland duo capture EX DAC (same measuring setup) was more than an order of magnitude lower (0.003%) and the value published by ASR for the AUDIOPHONICS converter was even lower than that. Both ADC and programs delivered very similar results, BTW.
OK, 0.05% is (for me) not too bad, but I'm interested in what the reason for the high THD values could be. Any ideas?
Peter
I measured THD and noise on both hats with a Motu M4 and an E1DA COSMOS ADCiso. Software: REW and AudioTester. For playback I used LibreELEC and Moode. The test signal was a wav file with a 24-bit sine (sampling frequency 44.1 kHz, level 5 dB below FS).
While the noise floor was quite low with both DAC, I was a little surprised with the comparatively high THD value of around 0.04% for both hats. For comparison: the distortion of an old Roland duo capture EX DAC (same measuring setup) was more than an order of magnitude lower (0.003%) and the value published by ASR for the AUDIOPHONICS converter was even lower than that. Both ADC and programs delivered very similar results, BTW.
OK, 0.05% is (for me) not too bad, but I'm interested in what the reason for the high THD values could be. Any ideas?
Peter
I don't know the reason in your setup, but here are my observations:
Each ADC has an optimum input level range. Anything above will quickly increase THD.
USB powered devices can spoil the signal. I use a home-made Y cable. The D- and D+ go straight (so does the GND/shield), but the 5V (red) is disconnected, and comes from a power bank to the ADC.
It is best to perform an analog loopback test to verify clean noise floor (signal OFF) and low THD (signal ON).
I also use an USB isolator that is a tiny board for signal pass-through and galvanic isolation.
Now I have a medium quality sound card (Sound Blaster X-fi HD, SB1240) with a -130 dB noise floor and 0.00050% THD in loopback.
Lastly, the LED desk lamp caused serious interference until I turned it off.
I am not sure but the RPi might emit stray field to the HAT, there might be a ground loop, etc. I don't like sensitive audio circuitry in the vicinity of digital electronics.
Each ADC has an optimum input level range. Anything above will quickly increase THD.
USB powered devices can spoil the signal. I use a home-made Y cable. The D- and D+ go straight (so does the GND/shield), but the 5V (red) is disconnected, and comes from a power bank to the ADC.
It is best to perform an analog loopback test to verify clean noise floor (signal OFF) and low THD (signal ON).
I also use an USB isolator that is a tiny board for signal pass-through and galvanic isolation.
Now I have a medium quality sound card (Sound Blaster X-fi HD, SB1240) with a -130 dB noise floor and 0.00050% THD in loopback.
Lastly, the LED desk lamp caused serious interference until I turned it off.
I am not sure but the RPi might emit stray field to the HAT, there might be a ground loop, etc. I don't like sensitive audio circuitry in the vicinity of digital electronics.
Thanks for your reply.
I have done some more tests with the AUDIOPHONICS DAC and found that - as you mentioned - the output level is rather critical. I had to reduce it to 250 mVeff to get a THD value of 0.005 %. THD+N was 0.0065 %. That's ok for me, but an USB DAC like a Topping D10 would have been the better solution (with a similar prize). With an additional case on the desk, however.
Peter
I have done some more tests with the AUDIOPHONICS DAC and found that - as you mentioned - the output level is rather critical. I had to reduce it to 250 mVeff to get a THD value of 0.005 %. THD+N was 0.0065 %. That's ok for me, but an USB DAC like a Topping D10 would have been the better solution (with a similar prize). With an additional case on the desk, however.
Peter