These are the measurements of a custom USB DAC using PCM1794 using loop back at 24-bit, 96 kHz . can someone comment on how good and bad they look? Also please give a comparison with a better DAC.
Hope to hear from experts on these objective data. I know anyone can say how sweet and mellow their DAC sound but when it comes to objective parameters long silence is the answer.
Hope to hear from experts on these objective data. I know anyone can say how sweet and mellow their DAC sound but when it comes to objective parameters long silence is the answer.
Attachments
Last edited:
For reference I have attached a measurement of the the noise from a PCM1794A based DAC that I built some time ago.
For the measurement I used a CS5381 based ADC with an input amp based on an OPA1612 to get a very low noise.
For the right channel 0dBFS is equivalent to -14 dBV (200 mVrms). The left channel of the ADC has less gain and more noise (and uses a different op-amp for the buffer), so it is not relevant in this case.
The maximum level from the DAC is 3.544Vrms, equivalent to 11 dBV.
The rms noise over a 20 kHz bandwidth is measured at -98.4 dBFS.
This all translates to a dynamic range of 123.4 dB non-weighted.
With A-weighting the dynamic range was 126 dB. This is close to the 127 - 129 dB specified for the PCM1794A. (The configuration of the output buffers probably adds 1 - 2 dB noise to get a balanced output).
Apart from the numbers it is important to note that there are no visible discrete frequencies in the noise spectrum.
In your measurements there is some noise, which seems to be related to mains hum. But whether this is due to the DAC or the ADC is difficult to know. And it may be difficult to get rid of if there is no galvanic isolation and if you use single ended outputs (non-balanced).
For the measurement I used a CS5381 based ADC with an input amp based on an OPA1612 to get a very low noise.
For the right channel 0dBFS is equivalent to -14 dBV (200 mVrms). The left channel of the ADC has less gain and more noise (and uses a different op-amp for the buffer), so it is not relevant in this case.
The maximum level from the DAC is 3.544Vrms, equivalent to 11 dBV.
The rms noise over a 20 kHz bandwidth is measured at -98.4 dBFS.
This all translates to a dynamic range of 123.4 dB non-weighted.
With A-weighting the dynamic range was 126 dB. This is close to the 127 - 129 dB specified for the PCM1794A. (The configuration of the output buffers probably adds 1 - 2 dB noise to get a balanced output).
Apart from the numbers it is important to note that there are no visible discrete frequencies in the noise spectrum.
In your measurements there is some noise, which seems to be related to mains hum. But whether this is due to the DAC or the ADC is difficult to know. And it may be difficult to get rid of if there is no galvanic isolation and if you use single ended outputs (non-balanced).
Attachments
Yes, there is a lot of noise to all your measurements, with many of peaks to all range of frequencies. Maybe your PSU, maybe you have overload the input of your sound card!
Look at the following Noise Floor captures.
All are from the same ESS9023 dac-chip, spdif input. The differences are from the different initial PSU. 7805 at the first, Salas V1.1 the better.
The captures are with EMU-0404(usb).
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.