• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

DAC measurements using DeltaWave

The is no way to de-embed the effect of the ADC other than using a "reference DAC"...

The SU-10 sits at the very top of ASR's SINAD table :).

Besides the DC filter, the ADC also has effects coming from the anti-aliasaing filter and these effects are magnitude and phase changes near fs/2 (so one should choose a steep linear-phase filter) and a passband ripple which is small but not irrelevant.

Yes. But these will be 'applied' equally to all comparison files (which have been through the DA/AD chain). If large differences in nulls between comparison files exist, then it must be down soley to the DACs.

Assuming, of course, that the R and C files are all correctly matched. I'm confident of this because of the very little variability between samples from the same DAC.
 
My first inclination was to think that the DC filter should only be applied to the reference file, as the comparison file has already had a similar filter applied by the ADC's analogue input caps. And although applying the DC filter to R+C may give better nulls, I wouldn't want to mess around with the comparison file, as I can't differentiate between the effects of the DAC and the ADC.

I'll play around with things. If applying the DC filter to R+C improves the nulls, but tends to homogenize them too, then I think for the purposes of this thread, that it should only be applied to the reference file.

Mani, unless you can be sure to reproduce the exact filter that was applied in the loopback process, filtering just the reference file is a no-no. You're potentially corrupting the original reference in a way that the comparison was not, and then trying to compute the differences between them. That's why I recommend applying the DC filter to both. I also suggest that you change DC filter application to DC/start+end. This way the filter will be applied twice and will have a greater effect, I suspect the null will be a bit deeper if the differences are in that region of the frequency bandwidth.
 
The SU-10 sits at the very top of ASR's SINAD table :).

And what does that have to do with its frequency response below a few Hz? :)

By the way, to @KSTR's point, here's your SU-10 track filtered with just a simple HP 8Hz FIR filter available in DeltaWave since time immemorial:

1736763572991.png

1736763648811.png
 
Mani, unless you can be sure to reproduce the exact filter that was applied in the loopback process, filtering just the reference file is a no-no. You're potentially corrupting the original reference in a way that the comparison was not, and then trying to compute the differences between them.

Yes, exactly replicating the ADC's filter would be ideal.

Applying a 6dB/octave MPHP filter at 0.1Hz improves the RMS difference by 8dB. However, it doesn't make much difference to the A-weighted RMS, the PK Metric or the RMS of the delta spectrum.

I'll play around with a few things...
 
Yes, exactly replicating the ADC's filter would be ideal.

Applying a 6dB/octave MPHP filter at 0.1Hz improves the RMS difference by 8dB. However, it doesn't make much difference to the A-weighted RMS, the PK Metric or the RMS of the delta spectrum.

I'll play around with a few things...
But the filter you applied is a minimum phase one. That changes the phase response of the reference across the whole frequency range, not just in the low frequencies. You are altering the reference in a major way, hoping that it might match the comparison.
 
And what does that have to do with its frequency response below a few Hz? :)

I'm really not interested in what happens below a few Hz. The effects of the MPHP filter go all the way up to 20kHz - clearly seen in the spectrum of delta charts.

(My mention of the SU-10 was in response to @KSTR 's comment about "reference DAC".)
 
But the filter you applied is a minimum phase one. That changes the phase response of the reference across the whole frequency range, not just in the low frequencies. You are altering the reference in a major way, hoping that it might match the comparison.

Yep, that's the idea :).

It'd useful to be able to set the frequency of the filter more accurately (better than my current 0.1Hz). I'd then use DW to determine the optimum.
 
Yep, that's the idea :).

It'd useful to be able to set the frequency of the filter more accurately (better than my current 0.1Hz). I'd then use DW to determine the optimum.
The corruption that has occurred to the comparison in the loopback process is a complex function. You guessing at it is fine for experimenting, but the actual measurement results you get as the result are not admissible in the court of law ;)
 
You could go the other direction. Put a boost to those very low frequencies for both reference and comparison file. See how much that worsens the null. Get an idea how much of the difference is there versus elsewhere in the band.

When you listen to the difference file is anything audible? If you boost it is anything audible? If the difference file is inaudible, then you can be sure it is an inaudible difference in simple play back. If you can only hear it with a boost to the difference file you can be sure it is inaudible in playback. If the difference file is audible, it still may be an inaudible difference for simple play back.

I think Deltawave is not being sensibly employed here in some respects.
 
We cannot compensate for the ADC totally. But can we compensate for the effects of its coupling caps? We know that the ADC:
- has a finite capacitor value (47 µF)
- has a finite input impedance (47 kΩ)

The high-pass “knee” should be at around 0.07 Hz theoretically.
 
I don't know how many times I have to repeat this... I am NOT trying to match the waveforms for the sake of getting a better null <10Hz. What I'm trying to do is to compensate for the effects of the ADC's coupling caps, in the audible range: https://audiosciencereview.com/foru...asurements-using-deltawave.59822/post-2195642

Best explained by o1: https://audiosciencereview.com/foru...asurements-using-deltawave.59822/post-2196025

"So, while you’re mainly compensating for a subsonic roll-off in the RME’s analog input, the act of matching that filter’s minimum-phase curve across the entire audio band is what tightens your null residual all the way to 20 kHz."
 
We cannot compensate for the ADC totally. But can we compensate for the effects of its coupling caps? We know that the ADC:
- has a finite capacitor value (47 µF)
- has a finite input impedance (47 kΩ)

The high-pass “knee” should be at around 0.07 Hz theoretically.

Don't even think about asking me to build a circuit emulator into DeltaWave! ;)
 
Don't even think about asking me to build a circuit emulator into DeltaWave! ;)

All I need is to be able to apply the MPHP filter to the ref file more accurately (hundredths of Hz), and then play around until I get the best nulls.

Not sure it's worth adding any such functionality to DW.
 
I think Deltawave is not being sensibly employed here in some respects.
I think it is an interesting use. It shows us how various effects from interpolation filters and DC blocking (in the ADC) show up, even if they are not audible. I think in the end this will be a demonstration that all decent DACs that use sharp linear phase filters give great nulling results.
 
The SU-10 sits at the very top of ASR's SINAD table :).
It does, but that table is for 1kHz. We have no idea what the table would look like if Amir made measurements at 10Hz, 1Hz and 0.1Hz. Almost every audio specification I've seen does not include test points below 20Hz so commercial products can do whatever they want. If they are sensible they will filter DC, IMHO.
 
Yes, exactly replicating the ADC's filter would be ideal.

Applying a 6dB/octave MPHP filter at 0.1Hz improves the RMS difference by 8dB. However, it doesn't make much difference to the A-weighted RMS, the PK Metric or the RMS of the delta spectrum.

I'll play around with a few things...
Paul mentioned that you provided one the recording files but don't see it. Do you have a kink (edit: ha ha I meant link)?

Also I can filter the reference file with a 0.07 Hz or 0.072 Hz high pass filter if you are not familiar with using sox (I am not very familiar with it either). You just put the reference file in the soxinput folder. The command I used to operate on the Oringinal2.wav file to make O2_0_072.wav was:
"sox original2.wav o2_0_072.wav highpass -1 0.072"
 
Last edited:
I think in the end this will be a demonstration that all decent DACs that use sharp linear phase filters give great nulling results.

Something I said from the beginning of DW: Phase response can have a significant impact on the RMS null value. Which is why RMS of null isn't a value I'd use all by itself to compare two devices, just like SINAD, it can help but it can also mislead.

Removing the effect of both, ADC and DAC filters in the loopback will have an even greater impact on the null value. Here are Mani's files, again, but this time engaging the non-linear EQ function that's designed to remove the effects of filters. In this case, all the filters in the loopback chain.

Notice that compared to Mani's result of about -81dB RMS null with his MPHP filter, this null is even deeper, at -89dB:

1736767919611.png


1736767975381.png


1736767990941.png
 
Something I said from the beginning of DW: Phase response can have a significant impact on the RMS null value. Which is why RMS of null isn't a value I'd use all by itself to compare two devices, just like SINAD, it can help but it can also mislead.
You can say it in the lecture part of the course, but sometimes we students have to do the lab work to really get the message.
 
Something I said from the beginning of DW: Phase response can have a significant impact on the RMS null value. Which is why RMS of null isn't a value I'd use all by itself to compare two devices, just like SINAD, it can help but it can also mislead.

That's why I'm not using it in the 'league table'.

Removing the effect of both, ADC and DAC filters in the loopback will have an even greater impact on the null value. Here are Mani's files, again, but this time engaging the non-linear EQ function that's designed to remove the effects of filters. In this case, all the filters in the loopback chain.

Notice that compared to Mani's result of about -81dB RMS null with his MPHP filter, this null is even deeper, at -89dB:

Yes, but non-linear EQ homongenizes the results: https://audiosciencereview.com/foru...asurements-using-deltawave.59822/post-2194642

I want the effects of the DACs' filters to be reflected in the nulls.
 
Back
Top Bottom