• 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!

Listening test of 2 power amplifiers - files recorded for download - disclosed

Can you hear a difference and which file do you prefer

  • I can hear a difference but have no ABX result

    Votes: 4 14.3%
  • I cannot hear a difference but have no ABX result

    Votes: 14 50.0%
  • I can hear a difference and have an ABX result

    Votes: 2 7.1%
  • I cannot hear a difference and have an ABX result

    Votes: 4 14.3%
  • I prefer witch1 file

    Votes: 1 3.6%
  • I prefer witch2 file

    Votes: 8 28.6%

  • Total voters
    28
  • Poll closed .

Langston Holland

Member
Forum Donor
Joined
Aug 5, 2021
Messages
58
Likes
119
Location
Pensacola FL
@Langston Holland your DAC and ADC are not synchronous, correct?

Correct, it's a shame too, but the USB bus assimilated all the good stuff long ago, like the Borg. OTOH, it stimulated people like Benchmark's John Siau to slay the asynchronous beast and he did. : )
This is great. What is the track you used? Also, what is the approx peak voltage in the recordings? Sorry if you already said what these are in the other forum.

You said you are not willing to do ABX, right?

The track I used is freely available on HDTracks in their 2020 Sampler, it's called I Put a Spell on You by Chantal Chamberland.

Voltages for the entire song, peak and otherwise (℅ Ocenaudio, latest version):

Directly out of the DAC3
DAC3.png


Out of the AHB2
AHB2.png


Out of the NC400
NC400.png


On ABX, I'm willing and able even though it's painful, I actually did my first set of trials last night (on these amps, I've done this kind of thing for over a decade). Anyone that's done it before probably won't volunteer again, but I'll take punishment to learn something. : )

God bless you and your precious family - Langston
 
Last edited:

Langston Holland

Member
Forum Donor
Joined
Aug 5, 2021
Messages
58
Likes
119
Location
Pensacola FL
So off we go. Sighted listeners hear differences and self confirm each other. So the assumption is made what they are hearing is real. Can't do or won't do any abx testing however.

I'd like a link to your ABX testing, I'm sure I could learn something to incorporate into mine.

God bless you and your precious family - Langston
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,524
Likes
37,057
I'd like a link to your ABX testing, I'm sure I could learn something to incorporate into mine.

God bless you and your precious family - Langston
I've posted some before. I normally use ABX plugin which is part of Foobar. Pkane has also included on in Deltawave if you wish to use it. I've not posted any for this particular pair of files as I've not done any.

Main thing is you have to have precise volume matching first off. The two file have to line up pretty close in time. Not necessarily sub sample alignment, but nearest sample or few is a good idea.
 

pogo

Major Contributor
Joined
Sep 4, 2020
Messages
1,242
Likes
382
This is news to me - thank you! I just sent him an email asking if I can download it after the site seemed to indicate it was going to mail it to me from Germany! : )
This could possibly support the subjectively made perceptions objectively and raise such comparisons to another level. You should really try it out.
 

pjug

Major Contributor
Forum Donor
Joined
Feb 2, 2019
Messages
1,775
Likes
1,561
Correct, it's a shame too, but the USB bus assimilated all the good stuff long ago, like the Borg. OTOH, it stimulated people like Benchmark's John Siau to slay the asynchronous beast and he did. : )


The track I used is freely available on HDTracks in their 2020 Sampler, it's called I Put a Spell on You by Chantal Chamberland.

Voltages for the entire song, peak and otherwise (℅ Ocenaudio, latest version):

Directly out of the DAC3
View attachment 147352

Out of the AHB2
View attachment 147353

Out of the NC400
View attachment 147354

On ABX, I'm willing and able even though it's painful, I actually did my first set of trials last night (on these amps, I've done this kind of thing for over a decade). Anyone that's done it before probably won't volunteer again, but I'll take punishment to learn something. : )

God bless you and your precious family - Langston
Thank you. For the recordings of AHB2 and NC400, do you know how to get voltage from the dB values?
In my question on synchronous, it was pointed out to me that I should have said shared clock. Is it possible to use an external clock with your DAC, or do you have another good one to allow the recordings to be made where ADC and DAC use the same clock?
 

Langston Holland

Member
Forum Donor
Joined
Aug 5, 2021
Messages
58
Likes
119
Location
Pensacola FL
do you know how to get voltage from the dB values?

Decibels are ratios of numbers, so there's no way to convert dB to an absolute voltage level unless you know one of the two numbers (voltages) that were used to compute the ratio (dB). The reason we bother with the decibel is because changes in their values closely approximate the way our senses work (hearing in this case). Thus if we increase the voltage going to our loudspeakers by 6dB, it sounds the same whether it increased the sound level from 70dB to 76dB, or from 80dB to 86dB. You have to double voltages to increase by 6dB, and this isn't intuitive. For example, going from 1V to 2V or from 50V to 100V will sound like the same increase, both are 6dB. This gets messy, decibels make the number scale human.

If relative voltages will work for you, just assume a reference of 1V and calculate the voltage from the change in dB like this: 10^(dB/20), which means you divide the dB number by 20, then press the 10x (the "x" will be superscript) key on your calculator. Example 1: a 3.2dB increase from 1V = 10^(3.2/20) = 1.45V. Example 2: a 3.2dB decrease (-3.2dB) = 0.69V.
to allow the recordings to be made where ADC and DAC use the same clock?

Each recording was made with the same clock, the one inside the APx515 analyzer, and the original file playback was made using the same clock, the one inside the DAC3 converter.

On synchronous (shared) vs. asynchronous (not shared) clocks, it really doesn’t matter with DeltaWave or our listening experience as long as the sample rate remains close to the same as the original recording, and more importantly, it doesn’t vary over the course of the playback and recording. An asynchronous converter decodes what the sample rate of the incoming audio is supposed to be and uses its own clock to determine when each sample is allowed to transition to the next step, whether it's recorded or allowed to exit a converter as audio. High quality converters, such as those used in this test satisfy each parameter; their clocks beat at the intended sample rate per second and stay that way during use. In this case my amp testing used a sample rate of 96,000 cycles per second and this frequency never drifted.

How do I know that? Straight lines in DeltaWave (thank you @pkane !):

Clock Drift.PNG


God bless you and your precious family - Langston
 
Last edited:

Langston Holland

Member
Forum Donor
Joined
Aug 5, 2021
Messages
58
Likes
119
Location
Pensacola FL
By way of contrast, the following DeltaWave Clock Drift plot is from the OP's amp test files (witch1 and witch2), where he used the same I/O device in duplex mode to playback and record his amp outputs. Given that it was the same device, the sample rate and clock drift will be identical during each playback/recording event, but as you can see even this "shared" clock scenario still did not prevent the I/O device from drifting differently during the 2nd playback/recording event.

DeltaWave tried to correct for the drift error, but since it varies with time it can only adjust the average around zero. Accurate measurement of anything requires LTI (Linear and Time-Invariant) behavior from both the measuring instrument and the device under test. Imagine trying to measure the height of an interior wall to verify if it's 8 feet, yet the wall keeps growing and shrinking as you attempt to measure it. Or say the wall is behaving itself and remaining constant, but your measurement tape keeps growing and shrinking while you try to read the numbers! This violates the time-invariant requirement and it's exactly what the OP's audio I/O device is doing, though on a much smaller scale, but one that becomes an issue when used for precision measurement.

All settings in DeltaWave are default except for "Auto-trim start & end", which I selected.

Clock Drift witch12.PNG


God bless you and your precious family - Langston
 
Last edited:

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,588
Likes
38,291
Location
Gold Coast, Queensland, Australia
People are overthinking the clock/timing issues. Just record both amplifiers' output at the same time fed into the one A/D converter. Either a 4 input A/D for stereo, or just the left channel of each amp into a 2 channel interface.
 

pjug

Major Contributor
Forum Donor
Joined
Feb 2, 2019
Messages
1,775
Likes
1,561
Decibels are ratios of numbers, so there's no way to convert dB to an absolute voltage level unless you know one of the two numbers (voltages) that were used to compute the ratio (dB). The reason we bother with the decibel is because changes in their values closely approximate the way our senses work (hearing in this case). Thus if we increase the voltage going to our loudspeakers by 6dB, it sounds the same whether it increased the sound level from 70dB to 76dB, or from 80dB to 86dB. You have to double voltages to increase by 6dB, and this isn't intuitive. For example, going from 1V to 2V or from 50V to 100V will sound like the same increase, both are 6dB. This gets messy, decibels make the number scale human.

If relative voltages will work for you, just assume a reference of 1V and calculate the voltage from the change in dB like this: 10^(dB/20), which means you divide the dB number by 20, then press the 10x (the "x" will be superscript) key on your calculator. Example 1: a 3.2dB increase from 1V = 10^(3.2/20) = 1.45V. Example 2: a 3.2dB decrease (-3.2dB) = 0.69V.
I understand dB, I only want to know if you were recording with the amp putting out a lot of power or not. I am hoping that the comparison would be made with tens of volts peak at least, as high as you can do without the amp clipping.
 

pjug

Major Contributor
Forum Donor
Joined
Feb 2, 2019
Messages
1,775
Likes
1,561
People are overthinking the clock/timing issues. Just record both amplifiers' output at the same time fed into the one A/D converter. Either a 4 input A/D for stereo, or just the left channel of each amp into a 2 channel interface.
This is a good idea. But I think some want stereo for the ABX so then you need to have the 4 channel ADC.
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,588
Likes
38,291
Location
Gold Coast, Queensland, Australia
I understand dB, I only want to know if you were recording with the amp putting out a lot of power or not. I am hoping that the comparison would be made with tens of volts peak at least, as high as you can do without the amp clipping.

Absolutely true. Audible amplifier differences become obvious at elevated levels into real speakers, less so @2.83V into a 8R resistor.
 

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
People are overthinking the clock/timing issues.

From my short experience, it is only an issue using DW (not saying that software works wrong). The higher the drift/jitter is, the worst the results are.

A fix I found: acceptable levels for recordings. Jitter gets better, PKM also. I assume incorrect levels for recordings adds a lot of quantization noise (DW shouldn't have to deal with).
 

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,632
Likes
10,207
Location
North-East
From my short experience, it is only an issue using DW (not saying that software works wrong). The higher the drift/jitter is, the worst the results are.

A fix I found: acceptable levels for recordings. Jitter gets better, PKM also. I assume incorrect levels for recordings adds a lot of quantization noise (DW shouldn't have to deal with).

I'm afraid you misunderstand the DW metrics. Jitter is the RMS of the null error signal in time domain, while RMS null is the error in amplitude. Both are different representations of the same error/difference, just two different ways of looking at it.
 
  • Like
Reactions: pma

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
I'm afraid you misunderstand the DW metrics. Jitter is the RMS of the null error signal in time domain, while RMS null is the error in amplitude. Both are different representations of the same error/difference, just two different ways of looking at it.

I recently opened files recorded at -10db vs original -0,5dB file. The first point I noticed was that. Just after, I noticed one match has a huge jitter, and one seems to be better according to PKM.

From my (very short) experience using DW, raising the level whilst recording (not clipping, but close to original) may improve both jitter and PKM.

And I assume using too low levels for recordings may add quantization noises, compromizing DW maths.

Sorry if I missunderstood all that :-(
 

Attachments

  • JitterLevel.PNG
    JitterLevel.PNG
    233.7 KB · Views: 73

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,632
Likes
10,207
Location
North-East
I recently opened files recorded at -10db vs original -0,5dB file. The first point I noticed was that. Just after, I noticed one match has a huge jitter, and one seems to be better according to PKM.

From my (very short) experience using DW, raising the level whilst recording (not clipping, but close to original) may improve both jitter and PKM.

And I assume using too low levels for recordings may add quantization noises, compromizing DW maths.

Sorry if I missunderstood all that :-(

The key is to understand that everything that DeltaWave does is by first computing a difference between the two waveforms. This difference is then measured in multiple different ways and metrics and charts presented to help analyze it.

Quantization noise is extremely unlikely to cause problems. DW math doesn't care about quantization noise. The error signal will be computed with the quantization noise included, if present.
 

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
Quantization noise is extremely unlikely to cause problems. DW math doesn't care about quantization noise. The error signal will be computed with the quantization noise included, if present.

I assume this noise has an effect if one wants to plot at below -90/-100dB with a limited ADC...
 

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,632
Likes
10,207
Location
North-East
I assume this noise has an effect if one wants to plot at below -90/-100dB with a limited ADC...

In a 24-bit DAC or ADC, a -10dB volume adjustment can affect the last two bits. That's well below -120dBFS level. Most modern DACs implement 32-bit processing internally, so the quantization error will be much, much lower.
 
  • Like
Reactions: pma

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
In a 24-bit DAC or ADC, a -10dB volume adjustment can affect the last two bits. That's well below -120dBFS level. Most modern DACs implement 32-bit processing internally, so the quantization error will be much, much lower.

My 96/24 ADC is at least 10 years old (the technology). And I'm sure the results wheren't the same depending on recording level. But i may redo a test, at 0 (relative), -10, -20, -30. I am almost sure it won't match the same with original.
 

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,632
Likes
10,207
Location
North-East
My 96/24 ADC is at least 10 years old (the technology). And I'm sure the results wheren't the same depending on recording level. But i may redo a test, at 0 (relative), -10, -20, -30. I am almost sure it won't match the same with original.

If done correctly, the difference should be far below -100dB. On the other hand, noise and distortions could change with level.
 
  • Like
Reactions: pma
Top Bottom