• 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!

Replace OP amps. Completely pointless, or not?

antcollinet

Master Contributor
Joined
Sep 4, 2021
Messages
7,408
Likes
12,291
Location
UK/Cheshire
Maybe it could be as you say but this is what an EE who manufactures amplifiers said about the measurement Stereophile performed on the NAD C 298 amp , so I don't know what to believe:
Well he seems to be talking about THD there, not IMD.

What he is saying is that since, harmonic products are always a multiple of the "generating" frequency.the lowest frequency harmonic distortion product is 2x the frequency it is generated from. So if you limit measurement bandwidth to 25kHz, the highest frequency you will get a harmonic of is 12.5kHz.

THD is supposedly a measure of the "total" harmonic distortion but you are not measuring distortion of signals AT ALL above 12.5kHz, and are cutting off a lot of the harmonics of lower frequencies.

In reality it doesn't matter much - because the harmonic products that are ultansonic (that you are not measuring) are not audible.

However, this is only fair if you are comparing like with like. If you compare one amp measured with a wide bandwitdth, with another amp measured with a lower bandwidth then the higher bandwidth measurement will always give a higher THD figure, even if the actual distortion is the same.

So to compare the THD of two amps, make sure the measurement bandwidth is the same. If it is always 25kHz, then fine.
 

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,383
Likes
24,749
Location
Alfred, NY
Maybe it could be as you say but this is what an EE who manufactures amplifiers said about the measurement Stereophile performed on the NAD C 298 amp , so I don't know what to believe:

Morello:

Note that the measurements on this class-D device are carried out with the low-pass filter (f=25 kHz) Audio Precision AUX-0025 connected between the output of the power amp and the input of the analyzer, which is why all THD measurements are irrelevant above about 6 kHz. I'm surprised Atkinson doesn't limit the graphs to just this.
....
The reason for the measurement procedure is that the analyzer from Audio Precision that Atkinson uses does not handle copious amounts of high-frequency garbage typical class-D spews out. What I object to is that THD data is still reported up to 20 kHz, even though the data is not relevant above 6-7 kHz. It is also clear that many readers of testers do not reflect on this, which is, of course, completely understandable.
(because they don't understand how this thing with measurements should be done, if I interpret what he says)
.....
THD is a measure of the step nonlinearities - quadratic, cubic and higher order terms in the transfer characteristic. The same terms entail intermodulation distortion. If you want a correct picture of the amp's distortion, the bandwidth when measuring THD up to 20 kHz must be at least 80 kHz, which means that harmonics up to and including order four are included.

The Nad C298 shows high THD at treble frequencies above 10 kHz, despite the fact that the numbers are heavily embellished due to the low-pass filter used in the THD measurement.

(with reservation for inaccuracies via google translate)
Note I'm just the messenger so don't shoot me down.

Edit:
I'll translate, or let google do it, a little more from that thread what Morello says:

...measuring THD up to 20 kHz with 25 kHz bandwidth is nothing to have opinions about - if you understand the implications, namely that the harmonics in the top octave are not included, it is easily realized that it is de facto an inappropriate procedure as the result is nonsense. ..

..I don't think you understood what I wrote. If an amplifier exhibits THD of 1% at 20 kHz with a dominant third tone, it means that the third tone has a frequency of 60 kHz. If you measure with 25 kHz bandwidth, the third tone will be attenuated by, for example, 40 dB (depending on the filter's flank attenuation) and the measurement will show a THD of 0.01%. The measurement thus gives a false result. ..


I-or in that thread:
What Morello has tried to convey above (and in many older threads) is that the nonlinearities that give rise to inaudible spectral components above 20kHz when measuring THD also give rise to audible intermodulation distortion (in the audible range). If you measure the IMD instead, this will be clear.


I don't know who "Morello" is, but there's a lot of half-true and thus misleading stuff in there.
 
OP
DanielT

DanielT

Major Contributor
Joined
Oct 10, 2020
Messages
4,750
Likes
4,631
Location
Sweden - Слава Україні
Well he seems to be talking about THD there, not IMD.

What he is saying is that since, harmonic products are always a multiple of the "generating" frequency.the lowest frequency harmonic distortion product is 2x the frequency it is generated from. So if you limit measurement bandwidth to 25kHz, the highest frequency you will get a harmonic of is 12.5kHz.

THD is supposedly a measure of the "total" harmonic distortion but you are not measuring distortion of signals AT ALL above 12.5kHz, and are cutting off a lot of the harmonics of lower frequencies.

In reality it doesn't matter much - because the harmonic products that are ultansonic (that you are not measuring) are not audible.

However, this is only fair if you are comparing like with like. If you compare one amp measured with a wide bandwitdth, with another amp measured with a lower bandwidth then the higher bandwidth measurement will always give a higher THD figure, even if the actual distortion is the same.

So to compare the THD of two amps, make sure the measurement bandwidth is the same. If it is always 25kHz, then fine.
I don't know who "Morello" is, but there's a lot of half-true and thus misleading stuff in there.
Interesting. It is fascinating that two technically driven forums can have such different views on class D amplifiers and measurements on them. Now, I'm not the first to note that. Even Belker, who writes here on ASR sometimes, has noted it in the thread I refer to earlier:

.... the level of knowledge here (on the forum Faktiskt ) has high peaks, but so does it on ASR. A significantly larger forum with many designers and educated people. That is why it is so strange that there is no consensus.

I'm just curious but other than that I have nothing to add on the subject.:)

 

Matias

Master Contributor
Forum Donor
Joined
Jan 1, 2019
Messages
5,029
Likes
10,794
Location
São Paulo, Brazil
More measurements.

 

antcollinet

Master Contributor
Joined
Sep 4, 2021
Messages
7,408
Likes
12,291
Location
UK/Cheshire
Interesting. It is fascinating that two technically driven forums can have such different views on class D amplifiers and measurements on them. Now, I'm not the first to note that. Even Belker, who writes here on ASR sometimes, has noted it in the thread I refer to earlier:

.... the level of knowledge here (on the forum Faktiskt ) has high peaks, but so does it on ASR. A significantly larger forum with many designers and educated people. That is why it is so strange that there is no consensus.

I'm just curious but other than that I have nothing to add on the subject.:)

There is one point mentioned that I didn't address.

If you rely only on THD, then limiting the bandwidth ingnores distortion products that can then inter-modulate with other frequencies back into the audio band.

This might be a problem. I don't care about ultrasonic harmonic distortion products, but I do care if we get resultant audible IMD products.

Luckly we don't rely only on the THD measurement. We also have the multitone measurement - which is somewhat representative of a complex real music signal.

The multitones will not only generate Harmonc distortion products at all multiples of all the tones. But will also inter-modulate between each other AND between themselves and all the other harmonic and intermodulation distortion products. Even the ones we don't measure in the THD measurement.

The result of all that distortion going on is the "grass" at the base of that measurement. What we see with competent designs is that grass is well below the level of audiblity.

So - who cares if we miss some of the ultrasonic harmonics, if they don't create any audible problem?
 
Last edited:

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,383
Likes
24,749
Location
Alfred, NY
Interesting. It is fascinating that two technically driven forums can have such different views on class D amplifiers and measurements on them. Now, I'm not the first to note that. Even Belker, who writes here on ASR sometimes, has noted it in the thread I refer to earlier:

.... the level of knowledge here (on the forum Faktiskt ) has high peaks, but so does it on ASR. A significantly larger forum with many designers and educated people. That is why it is so strange that there is no consensus.

I'm just curious but other than that I have nothing to add on the subject.:)

There's a vocal minority who seem to have philosophical objections to Class D. Maybe it's because it commodifies high power and high performance? Maybe because it obsoletes their expertise on linear A/AB amps? But I've certainly had problems verifying their dark views (and sometimes measurements) on my lab bench and in my listening room. I have yet another high power moderate cost Class D amp on my workbench right now which outperforms the AP analyzer regardless of the load I throw at it. And it's far from the first. My challenge is finding something new to say for yet another stellar performer- it seems to be the norm now, not the exception.
 

DonR

Major Contributor
Joined
Jan 25, 2022
Messages
2,968
Likes
5,611
Location
Vancouver(ish)
There's a vocal minority who seem to have philosophical objections to Class D. Maybe it's because it commodifies high power and high performance? Maybe because it obsoletes their expertise on linear A/AB amps? But I've certainly had problems verifying their dark views (and sometimes measurements) on my lab bench and in my listening room. I have yet another high power moderate cost Class D amp on my workbench right now which outperforms the AP analyzer regardless of the load I throw at it. And it's far from the first. My challenge is finding something new to say for yet another stellar performer- it seems to be the norm now, not the exception.
If people knew less about the Class D theory of operation they might not have so many objections (a little knowledge is a dangerous thing). Much like the advent of digital music which found many critics in the early days despite the fact that studios were using the technology widely up to a decade earlier.
 
OP
DanielT

DanielT

Major Contributor
Joined
Oct 10, 2020
Messages
4,750
Likes
4,631
Location
Sweden - Слава Україні
What we see with competnt designs is that grass is well below the level of audiblity.

So - who cares if we miss some of the ultrasonic harmonics, if they don't create any audible problem?
And to be able to sniff out amp distortion (if it is not driven into clipping) you need ultra-low distortion speakers and a lot of training in detecting distortion. Of which I have none, luckily for my wallet, I'm just saying.;):D
There's a vocal minority who seem to have philosophical objections to Class D. Maybe it's because it commodifies high power and high performance? Maybe because it obsoletes their expertise on linear A/AB amps? But I've certainly had problems verifying their dark views (and sometimes measurements) on my lab bench and in my listening room. I have yet another high power moderate cost Class D amp on my workbench right now which outperforms the AP analyzer regardless of the load I throw at it. And it's far from the first. My challenge is finding something new to say for yet another stellar performer- it seems to be the norm now, not the exception.
Now that's more like it. It's just a matter of measuring at different bandwidths and then seeing how it looks with IMD, THD in the higher frequencies, within the audible range. Preferably comparative measurements for higher order harmonics.
When I say It's just a matter, I mean for those who have measuring equipment and the knowledge, which you seem to have.:)
If people knew less about the Class D theory of operation they might not have so many objections (a little knowledge is a dangerous thing). Much like the advent of digital music which found many critics in the early days despite the fact that studios were using the technology widely up to a decade earlier.
I like class D and so do a lot of people on that other forum. There may be a greater skepticism towards cheap load-dependent class D amps there than here at ASR, but good Hypex, for example, is appreciated and used by many on that forum.:)
There has been a development of class D amps in recent decades, which you are clearly aware of, but I still mention it for others who read this thread. Those who are notoriously skeptical of class D may have missed this development, what do I know?:oops:
Gone are the days when it was said that class D amps are only good for subwoofers. Well, maybe there are those who say that now, but missed that this is not the case nowadays.

Bruno Putzeys talks about the class D development in this interview::)

 

Attachments

  • Screenshot_2023-08-30_071935.jpg
    Screenshot_2023-08-30_071935.jpg
    502.5 KB · Views: 31
  • Screenshot_2023-08-30_072001.jpg
    Screenshot_2023-08-30_072001.jpg
    815.4 KB · Views: 32
Last edited:

MaxwellsEq

Major Contributor
Joined
Aug 18, 2020
Messages
1,627
Likes
2,424
So to compare the THD of two amps, make sure the measurement bandwidth is the same. If it is always 25kHz, then fine
That's where I think the objections are. I've seen it expressed as "Class D has this high frequency problem and lab measurements are covering it up by including a filter that blocks this high frequency problem, but Class AB are not being filtered so are getting a tougher ride than Class D".

This is potentially valid. If all the historical reviews of Class AB amps have TOTAL harmonic distortion including harmonics we can't hear, then THD as a number is not a valid comparator if one is with a filter and one isn't.
 

NTK

Major Contributor
Forum Donor
Joined
Aug 11, 2019
Messages
2,656
Likes
5,819
Location
US East
That's where I think the objections are. I've seen it expressed as "Class D has this high frequency problem and lab measurements are covering it up by including a filter that blocks this high frequency problem, but Class AB are not being filtered so are getting a tougher ride than Class D".

This is potentially valid. If all the historical reviews of Class AB amps have TOTAL harmonic distortion including harmonics we can't hear, then THD as a number is not a valid comparator if one is with a filter and one isn't.
Amir's headline SINAD numbers are measured with a 1 kHz test signal. If the measurement bandwidth is limited to 20 kHz, that covers to the 20th harmonic. Do we expect the H21, H22, ... etc., harmonics to make any significant differences?

For THD frequency sweeps, Amir increases the measurement bandwidth to 45 kHz.
 
OP
DanielT

DanielT

Major Contributor
Joined
Oct 10, 2020
Messages
4,750
Likes
4,631
Location
Sweden - Слава Україні
Amir's headline SINAD numbers are measured with a 1 kHz test signal. If the measurement bandwidth is limited to 20 kHz, that covers to the 20th harmonic. Do we expect the H21, H22, ... etc., harmonics to make any significant differences?

For THD frequency sweeps, Amir increases the measurement bandwidth to 45 kHz.
Sure, but measuring a few watts at 1 kHz is one thing. Measuring half power or even closer to the clipping point at 6kHz and above in frequency is another matter.
 

NTK

Major Contributor
Forum Donor
Joined
Aug 11, 2019
Messages
2,656
Likes
5,819
Location
US East
Sure, but measuring a few watts at 1 kHz is one thing. Measuring half power or even closer to the clipping point at 6kHz and above in frequency is another matter.
You listen to music with half of its power at 6 kHz or above? Or you are just interested in measurement for measurement's sake?
 
OP
DanielT

DanielT

Major Contributor
Joined
Oct 10, 2020
Messages
4,750
Likes
4,631
Location
Sweden - Слава Україні
You listen to music with half of its power at 6 kHz or above? Or you are just interested in measurement for measurement's sake?
But it might be interesting to know what it looks like. Sooner or later, probably later, the high-frequency distortion becomes audible. Amir includes it in his measurements. An example:
Topping PA5 II Stereo Amplifier Audio Balanced Power 4 ohm vs frequency Measurement.png


 

MaxwellsEq

Major Contributor
Joined
Aug 18, 2020
Messages
1,627
Likes
2,424
Amir's headline SINAD numbers are measured with a 1 kHz test signal. If the measurement bandwidth is limited to 20 kHz, that covers to the 20th harmonic. Do we expect the H21, H22, ... etc., harmonics to make any significant differences?
I don't think anyone expects to hear variations in H21 etc. :)
 

egellings

Major Contributor
Joined
Feb 6, 2020
Messages
4,004
Likes
3,244
That is not needed. If IM Distortion components appear in the audible frequency range, they will be there whether or not we measure higher (ultrasonic) frequencies.

To clarify, the high frequencies that inter-modulate and create audible band component are there in the signal whether we measure them or not. Those components will be visible in a measurement limited to 20kHz.
Agree!
 

egellings

Major Contributor
Joined
Feb 6, 2020
Messages
4,004
Likes
3,244
There is one point mentioned that I didn't address.

If you rely only on THD, then limiting the bandwidth ingnores distortion products that can then inter-modulate with other frequencies back into the audio band.

This might be a problem. I don't care about ultrasonic harmonic distortion products, but I do care if we get resultant audible IMD products.

Luckly we don't rely only on the THD measurement. We also have the multitone measurement - which is somewhat representative of a complex real music signal.

The multitones will not only generate Harmonc distortion products at all multiples of all the tones. But will also inter-modulate between each other AND between themselves and all the other harmonic and intermodulation distortion products. Even the ones we don't measure in the THD measurement.

The result of all that distortion going on is the "grass" at the base of that measurement. What we see with competent designs is that grass is well below the level of audiblity.

So - who cares if we miss some of the ultrasonic harmonics, if they don't create any audible problem?
Agree; so long as the 'grass' does not cause amplifier misbehavior, then it does not matter sonically.
 
Top Bottom