this is real APU noise + synthetic 1k sine mix.
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Topping D90LE - SINAD measurement (with E1DA Cosmos APU Notch)
- Thread starter Rja4000
- Start date
I posted the screenshot, you can see it, or do you need BW at -3db?
OP
the plot above is the ratio of ignoring weighting FFT bins. As I said, AP doesn't simply ignore some range of bins but has implemented analog-like notching and Butterworth HPF 4or at 22Hz + LPF 64ord at 20Hz.
Of course, the situation I've simulated is way nonrealistic because the only noise was taken into account but distortions were artificially set to zero. However, that simulation well shows a noticeable difference AP vs REW, and nearly no difference AP vs Matlab. With more or less real signals, the difference is less.
Attachments
Last edited:
OP
oh, sorry, the wrong plot was attached.
OP
I posted the noise only without artificial sine, and yes, it is fixed.
OP
only a polynomial form.
OP
- Thread Starter
- #33
After some "tuning",
I can now consistently reproduce REW "AES-17 2015 standard notch" results with Virtins Multi-Instrument 3.9.6.2
Here is a comparison of different SINAD measurement methods:
It's all for the RME ADI-2 Pro fs R
I measure the DAC, both channel merged in Mono (my usual setup for measurements) for the 13dBu range
The RME allows to go over 0dBFS, so I dialed in +2.5dB on the volume knob, for an output range of 15.5dBu (approx)
Except for the Red measurement, the ADC is the RME itself, in AutoRef mode (range switches automatically from +4dBu to 13, 19 and 24dBu)
The Black plot is with the APU notch (+6dB), with the AES-17 2015 notch compensation (fundamental computed on 1 full octave, as in REW)
The Blue plot is the same measurement, but using the standard SINAD measurement from Multi-Instrument (which measures on a much narrower range)
We see that there is a constant noise penalty of around 2.8dB with the APU notch in the loop.
The Orange and Pink plots are measured without the APU Notch (direct from DAC to ADC)
Interestingly, they are pretty much identical (the measurement of the fundamental over 1 octave is only important if you use the APU Notch)
Also, we see that the noise at low level is exactly matching the noise with the APU notch and the AES-17 1 octave fundamental.
That gives me good confidence that we can get accurate results this way.
Finally, the Red plot is measured by connecting directly from the DAC to the E1DA Cosmos ADC in 4.5V range.
Note that the noise is higher (we could get lower noise with another range, but then we'd saturate at the top level)
Also note that the distortion starts lower in Voltage and that DAC output is then limited, due to the low input impedance of the E1DA ADC.
I can now consistently reproduce REW "AES-17 2015 standard notch" results with Virtins Multi-Instrument 3.9.6.2
Here is a comparison of different SINAD measurement methods:
It's all for the RME ADI-2 Pro fs R
I measure the DAC, both channel merged in Mono (my usual setup for measurements) for the 13dBu range
The RME allows to go over 0dBFS, so I dialed in +2.5dB on the volume knob, for an output range of 15.5dBu (approx)
Except for the Red measurement, the ADC is the RME itself, in AutoRef mode (range switches automatically from +4dBu to 13, 19 and 24dBu)
The Black plot is with the APU notch (+6dB), with the AES-17 2015 notch compensation (fundamental computed on 1 full octave, as in REW)
The Blue plot is the same measurement, but using the standard SINAD measurement from Multi-Instrument (which measures on a much narrower range)
We see that there is a constant noise penalty of around 2.8dB with the APU notch in the loop.
The Orange and Pink plots are measured without the APU Notch (direct from DAC to ADC)
Interestingly, they are pretty much identical (the measurement of the fundamental over 1 octave is only important if you use the APU Notch)
Also, we see that the noise at low level is exactly matching the noise with the APU notch and the AES-17 1 octave fundamental.
That gives me good confidence that we can get accurate results this way.
Finally, the Red plot is measured by connecting directly from the DAC to the E1DA Cosmos ADC in 4.5V range.
Note that the noise is higher (we could get lower noise with another range, but then we'd saturate at the top level)
Also note that the distortion starts lower in Voltage and that DAC output is then limited, due to the low input impedance of the E1DA ADC.
Jimster480
Major Contributor
Mono? It has a specific Mono mode? Or are you just using one channel?
Jimster480
Major Contributor
Yes but are you merging it yourself? Meaning you are feeding the same signal to both channels and then doing an external merge? Or is there some mono mode?
I'm going to assume that you are doing an external merge because you mentioned "XLR Output"
OP
- Thread Starter
- #39
Fair point
Here we go
NB: One difference with the initial measurements in this thread is that I now use a test signal at 997Hz, as per AES-17.
The main difference is that the 20th harmonic falls below 20kHz.
It is at -147.09-4.23= -151.32dBr though.
I tuned my APU notch for 997Hz and -30dB.
Last edited:
Similar threads
