Bass control differences between DAC? What to check with measurements?

[pma]

Audible at all if 1db louder?

YES.

 

[solderdude]

But will that change the perception of bass when there are no harmonics there ?
Will the difference in amplitude still be there when not using the default filter for the X16 to level the playing field ?

A simple experiment would be to use the X16 and determine if the bagpipes sound different using default and sharp filter '2' ?
That would be far more effective trying to determine if those frequencies actually play a role.

Also try to determine the Df of both DACs (with similar filters).

 

[Pdxwayne]

But will that change the perception of bass when there are no harmonics there ?
Will the difference in amplitude still be there when not using the default filter for the X16 to level the playing field ?

A simple experiment would be to use the X16 and determine if the bagpipes sound different using default and sharp filter '2' ?
That would be far more effective trying to determine if those frequencies actually play a role.

Also try to determine the Df of both DACs (with similar filters).

BTW, the bagpipe comparison part have little bass content. I switched to comparing high notes as it was showing the most strange issues when using the DeltaWave app.

I did try using x16 different filters to listen to bagpipe. I don't think I sense much different, but I was mostly looking for the E30 kind of sound and I did not find it. When I put back E30 to my stereo system to listen, the different was not night and day obvious. Again, unfortunately, my family members are always around. Wife and daughter can't stand loud music.... So I have little chance to do more comparisons again.

What is Df?

 

[solderdude]

Df is the difference signal but displayed differently.
There are issues with that one too.

There is nothing in your test results that would warrant different sounding bass.
The next thing to investigate is your bias (knowing what is playing).
For that you need one more helper. You can spread your testing over weeks if need be.
Just have someone swap DAC's every now and then when you are not in the room.
Let them log what they changed at what date/time. Make sure you don't see that log or got hints something changed or can hear typical 'hints' like a relay clicking etc.
Also make sure the cabling and DACs are hidden and don't cheat by looking what driver/DAC is active.
Try to determine with normal music listening which DAC s playing.
Log it per session (date/time) and after a few weeks compare notes.

That is the only 'test' you can do. Wave forms seem similar coming out of the DAC. Those wave forms dictate the sound. When the differences are small (and they seem to be) only one conclusion remains.
That one needs testing and will be extremely enlightening regardless of the outcome.

 

[Blumlein 88]

You don't need to do these 20 attempts the same day. I have done some tests over weeks (where the wife swapped or didn't swap connections when I was not home).
Statistical enough attempts are essential when you want to eliminate bias.



correct. There were no differences upto 10kHz though. There won't be much harmonics changing the tonality of a bass above 10kHz. That is more in the 150Hz to 2kHz range.
20Hz to 8kHz is very easy for any DAC.

I also don't agree. You can start slopes around 8 khz and above. Done right it can effect how solid the bass sounds or how warm it sounds. Often boosting or slightly shelving down 8 khz and above effects your perception of airiness. More air and softer lesser bass vs less air and firmer more bass.

Now in this case I'm not convinced we have an answer either way as those spectrum differences look to be noise.

What might help is to see a spectrogram. That is built into Deltawave so how about we see the spectrogram for the two DACs. A spectrogram can reveal such things that FFTs sometime hide.

 

[solderdude]

I also don't agree. You can start slopes around 8 khz and above. Done right it can effect how solid the bass sounds or how warm it sounds. Often boosting or slightly shelving down 8 khz and above effects your perception of airiness. More air and softer lesser bass vs less air and firmer more bass.

Also if the shelving above 10kHz is less than 1dB ?
The spectrogram is a good idea. I would see no reason as to why 2 DACs would do that differently and if that can resolve 1dB.
The tests done by @Pdxwayne with different filters showed he did not perceive the bagpipes differently yet these have higher harmonics.
More datapoints are needed.

Especially expectation bias has to be checked.
Once one is convinced 2 DAC sound different and one knows which DAC is used it is impossible not to be biased (speaking on personal experience)
The unconscious part is powerful and cannot be overruled by conscious thinking.

 

[Pdxwayne]

Here are the spectrograms. Not sure how to interpret them....

ASR web site keeps saying uploaded file too large, so I had to adjust the screen size multiple times, recapture, and tried upload multiple times. : (

Thanks for checking!

E30:

X16:

DAC comparison:

 

[solderdude]

It's time for the blind test.

Don't stress... span it over weeks.
Just teach someone which cables to swap and let them note what was connected when.

 

[Pdxwayne]

It's time for the blind test.

Don't stress... span it over weeks.
Just teach someone which cables to swap and let them note what was connected when.

Unfortunately it is not practical for my music enjoyment over long term. I have to use fixed volume in my node2i to get mqa decode to work via coaxial to my x16. It means I will need to use the remote of x16 to adjust volume. This alone defeats the blind test requirements.....

Still, I really appreciate your suggestions so far.

Can you tell me what the spectrograms mean to you?

 

[solderdude]

The delta shows that the differences are very small and noise dominated.
Basically the waveform is very similar.

Here is another test.
Record both DAC outputs and post the recordings without given away which DAC is what.
Maybe you will get some folks to have a listen.

 

[Pdxwayne]

The delta shows that the differences are very small and noise dominated.
Basically the waveform is very similar.

Here is another test.
Record both DAC outputs and post the recordings without given away which DAC is what.
Maybe you will get some folks to have a listen.

I would love to do that, but according to copyright law, I can't copy any portion of file and distribute it. I wonder if there is any copyright free high quality bagpipes music out there.....

 

[Pdxwayne]

The delta shows that the differences are very small and noise dominated.
Basically the waveform is very similar.

Here is another test.
Record both DAC outputs and post the recordings without given away which DAC is what.
Maybe you will get some folks to have a listen.

BTW, I checked the original file with a copy of it and the diff spectrogram is fully green.

Since you mentioned "noise", this makes me curious. I think I should check the noise of the playback and capture chain. This weekend I will capture same song a couple more times for each DAC. I will then compare the captured files of each DAC and see if I could get fully green diff spectrogram or not. Then I get good idea about the "noise" of the chain. Sounds reasonable?

Thanks!

 

[Blumlein 88]

I agree the spectrograms showed nothing unusual a bit of noise.

As for posting copies, generally you are okay if copies are no more than 30 seconds. PLEASE NOTE: I am no lawyer. And there is no formal rule that says 30 seconds and less is fine. But it is something of a accepted idea if you use song snippets for no more than 30 seconds it is okay. So you'll have to decide that for yourself.

So some blind testing is in order to see if there is anything actually being heard or just a result of placebo.

 

[solderdude]

I wonder if there is any copyright free high quality bagpipes music

https://www.bensound.com/royalty-free-music/2

 

[Blumlein 88]

I'm still curious why you seem to have so much noise. Here is a result from where I sent a song thru DA to AD conversion 8 times. And it isn't noisy they you are getting results. Yet it has 8 passes thru for noise to build up.

Here is the result of the same music from one pass thru. You see the frequency response ripple hasn't built up, but both are free of high frequency noise that seems to infect these recent posted results.

 

[Pdxwayne]

I'm still curious why you seem to have so much noise. Here is a result from where I sent a song thru DA to AD conversion 8 times. And it isn't noisy they you are getting results. Yet it has 8 passes thru for noise to build up.

View attachment 113590

Here is the result of the same music from one pass thru. You see the frequency response ripple hasn't built up, but both are free of high frequency noise that seems to infect these recent posted results.

View attachment 113591

What program do you use to capture audio? At what bit rate and frequency?

May I know how you connect your chain?

Mine is USB disk to Node2i, Node2i coaxial to DAC, DAC RCA to Scarlett Solo version 3, Scarlett Solo USB to laptop, laptop to Audacity capture at 44100 Hz, 32 bit floating point.

Thanks!

 

[Pdxwayne]

I agree the spectrograms showed nothing unusual a bit of noise.

As for posting copies, generally you are okay if copies are no more than 30 seconds. PLEASE NOTE: I am no lawyer. And there is no formal rule that says 30 seconds and less is fine. But it is something of a accepted idea if you use song snippets for no more than 30 seconds it is okay. So you'll have to decide that for yourself.

So some blind testing is in order to see if there is anything actually being heard or just a result of placebo.

I really hope is placebo.

I was listening to E30 for about 2 weeks with my downstairs stereo system before I put in the x16. My immediate impression was it is missing sparkles. Also, music notes seem to stop too early.... It could be just my hearings turned a bit bad during the time I started listening to x16.

Listening test is not likely with my stereo setup downstairs, unfortunately. Family members around all the time and they already sick of my blind tests I did last year between E30 and KTB.

 

[Blumlein 88]

I used Macbook over USB to March audio DAC 1 analog out over RCA to Antelope Audio Zen Tour which went over USB back to IBM laptop using Reaper to record. I have done the same thing using Foobar (over WASAPI) for playback and feeding a Scarlett 18i20 1st gen for the ADC with similar results.

If you are going thru Windows to record in Audacity the Windows sound mixer is probably in the way. Audacity has no ASIO capability. Windows is probably resampling or dithering your audio.

You can download Reaper and use it for free. It is more than you need, but take a minute to learn to record with it. You can then use ASIO between your Scarlett and Reaper. This direct connection will prevent Windows from mucking it up. Not familiar with your playback chain so don't have any suggestions there.

https://www.reaper.fm/

Free to download and trial forever, but I do buy it since I use it so much.

Oh, I think I was doing all this at 44.1 khz or it might have been 48 khz.

 

[AnalogSteph]

May I know how you connect your chain?

Mine is USB disk to Node2i, Node2i coaxial to DAC, DAC RCA to Scarlett Solo version 3, Scarlett Solo USB to laptop, laptop to Audacity capture at 44100 Hz, 32 bit floating point.

How did you adapt RCA to the Solo's balanced input? Monoprice RCA --> XLR cable? You may want to temporarily resort to battery operation on the laptop in order to defeat any ground loop issues.

I think Audacity uses WASAPI by default these days, but it's shared mode and as such settings for the recording device have to be explicitly chosen (the control panel dialog has gotten even more hidden since Windows 10 but is still there). Format might default to 16/44 for all I know.

In the CS4272's ADC as used in this device, digital filter stopband at single speed starts at about 0.58fs, so you really want to run that at 48 kHz at least (no aliasing until 0.42fs = 20.2 kHz). I would recommend 96 kHz (stopband from 0.68fs --> no aliasing until 0.32fs = 30.7 kHz). 192 kHz probably is diminishing returns territory (stopband from 0.78fs --> 0.22fs = 42.2 kHz, though ultimate rejection is a few dB better still).

BTW, in controlled listening tests level matching to 0.1 dB is advised. 0.3 dB is about the limit of human perception. A mismatch of 0.5 dB probably won't cut it.

 

[Pdxwayne]

I used Macbook over USB to March audio DAC 1 analog out over RCA to Antelope Audio Zen Tour which went over USB back to IBM laptop using Reaper to record. I have done the same thing using Foobar (over WASAPI) for playback and feeding a Scarlett 18i20 1st gen for the ADC with similar results.

If you are going thru Windows to record in Audacity the Windows sound mixer is probably in the way. Audacity has no ASIO capability. Windows is probably resampling or dithering your audio.

You can download Reaper and use it for free. It is more than you need, but take a minute to learn to record with it. You can then use ASIO between your Scarlett and Reaper. This direct connection will prevent Windows from mucking it up. Not familiar with your playback chain so don't have any suggestions there.

https://www.reaper.fm/

Free to download and trial forever, but I do buy it since I use it so much.

Oh, I think I was doing all this at 44.1 khz or it might have been 48 khz.

I installed Reaper and used ASIO4all. Cleaner, but still not as clean as yours. I wonder where is the issue....

Captured with Audacity vs original:

Captured with Reaper with ASIO vs original: