On Class D Amplifiers Measurements

[Matias]

I don't think so until amps are measured into stressful reactive loads.
I have read that some negative feedback are sensitive to load as well a frequency where they become less effective.

- Rich

A measurement would confirm this for sure. Until then we only have Hypex's claim: "Flat, fully load-independent frequency response".

 

[RichB]

A measurement would confirm this for sure. Until then we only have Hypex's claim: "Flat, fully load-independent frequency response".

If not explicit, I expect that "load-indepent" refers to resistive loads.

Stereophile has a standard reactive load. They show frequency variance (linearity) into 8, 4, 2 ohms + the simulated speaker.
Distortion is not measured into reactive (Speaker) loads.

- Rich

 

[tmtomh]

I don't think so until amps are measured into stressful reactive loads.
I have read that some negative feedback are sensitive to load as well a frequency where they become less effective.

- Rich

How has that not been addressed above?

 

[RichB]

How has that not been addressed above?

The Stereophile simulated speaker tests show frequency response not distortion.
I don't know how such measurements are taken, if they a band limited to the frequency, there may distortions in other frequencies not shown.

At one time, HometheaterHifi measured mega amps while driving Carver Ribbon speakers and these amps did very well.
I never saw and AVR or Class-D measured in this way. Maybe they are better than the A/B amps.

I have not seen amplifier distortion measurements into reactive loads.

- Rich

 

[dkfan9]

NC252MP is below 0.004 ohm, and 1ET400A, although it varies more than the NC252MP, is below 0.001 ohm, both through the audio range.

Putting them into context, for example:

Hegel H160: below 0.1 ohm.
"The output impedance was very low over most of the audioband, at 0.07 ohm, rising to 0.1 ohm at 20kHz. (Both figures include 6' of speaker cable.) As a result, the modulation of the H160's frequency response with our standard simulated loudspeaker (fig.1, gray trace) was minimal."
https://www.stereophile.com/content/hegel-music-systems-h160-integrated-amplifier-measurements

Krell KSA-50S: below 0.28 ohm.
"its output impedance was highish for a solid-state design at a calculated 0.28 ohms. This was the same at 20kHz as it was at 20Hz, however. "
https://www.stereophile.com/content/krell-ksa-50s-power-amplifier-measurements#

CH Precision M1.1: below 0.1 ohm best case.
"The output impedance with 100% feedback, including a 6'-long, spaced-pair speaker cable, was a low 0.1 ohms across the audioband. With 20% feedback, it rose to 0.24 ohm at low and middle frequencies, and 0.27 ohm at the top of the audioband."
https://www.stereophile.com/content/ch-precision-m11-power-amplifier-measurements#

McIntosh MC275: below 0.38 ohm best case.
"The McIntosh's output impedances were on the low side for a transformer-coupled tube design, at 0.33 ohm from the 4 ohm tap, 0.57 ohm from the 8 ohm tap, and 0.87 ohm from the 16 ohm tap. These figures applied at low and midrange frequencies; the impedance at 20kHz was a little higher, at 0.38 ohm (4 ohm tap), 0.7 ohm (8 ohm tap), and 1.1 ohms (16 ohm tap)."
https://www.stereophile.com/content/mcintosh-mc275-power-amplifier-measurements

JA measures output impedance with 6 feet of speaker cable attached, and the measurement rig itself is different, so it's not exactly 1 for 1 with those measurements from Purifi and Hypex. Better would be a Stereophile review of an amp using one of these modules. This would account for any additional parts used prior to the output itself that can add up at such a low level. But to your broader point the output impedance itself is low enough to be considered a non issue on its own. I still think it's worthwhile to explore what happens with speakers because the output impedance doesn't necessarily tell the whole story on what happens when driving reactive loads.

 

[pma]

Alan @March Audio is right that the natural music signal almost always does not contain high frequency components above 5kHz in a full amplitude, so it is not needed to test the amp this way . Similarly he would argue about the 19+20 kHz test. OK, I may agree, however, the high frequency components are almost never present alone, but they are superimposed on a signal of lower frequency and the resulting sum of all components may reach up to full scale, even if the individual high frequency sine waves have low amplitude. It is the truth easily defendable that energy of many many music sample frequency components remains same up to midrange, several kHz, typically 2 - 3 kHz. So I am proposing another test, twin tone IMD 2kHz + 17kHz with ratio of amplitudes 100:1, in other words 2kHz component is 40dB higher than the 17kHz component. I have tested the AIYIMA A07 class D amplifier with this signal - and - as expected, it does not like it. Even if there is a negligible, 100x smaller 17kHz superimposed on the 2kHz wave and is not visible by eye observation, the amp does not like it, because, it sees the sum of amplitudes of the 2kHz and 17kHz sines, so the resulting amplitude is high with fast 17kHz ripple.

This is the test signal, clean spectrum of 2 components 100:1 amplitudes

And this is the amplifier response, high IMD of 0.3% (-50dB)

The IMD products will be definitely more audible than the 17kHz tone itself.

 

[Matias]

Leaving continuous power aside (good), the IMD components depend on how well refined the amplifier technology is. The Aiyima A07 is a budget amp. Here is the other side of the spectrum, Purifi 1ET400A. Artifacts below -120dB.

 

[dkfan9]

Alan @March Audio is right that the natural music signal almost always does not contain high frequency components above 5kHz in a full amplitude, so it is not needed to test the amp this way . Similarly he would argue about the 19+20 kHz test. OK, I may agree, however, the high frequency components are almost never present alone, but they are superimposed on a signal of lower frequency and the resulting sum of all components may reach up to full scale, even if the individual high frequency sine waves have low amplitude. It is the truth easily defendable that energy of many many music sample frequency components remains same up to midrange, several kHz, typically 2 - 3 kHz. So I am proposing another test, twin tone IMD 2kHz + 17kHz with ratio of amplitudes 100:1, in other words 2kHz component is 40dB higher than the 17kHz component. I have tested the AIYIMA A07 class D amplifier with this signal - and - as expected, it does not like it. Even if there is a negligible, 100x smaller 17kHz superimposed on the 2kHz wave and is not visible by eye observation, the amp does not like it, because, it sees the sum of amplitudes of the 2kHz and 17kHz sines, so the resulting amplitude is high with fast 17kHz ripple.

This is the test signal, clean spectrum of 2 components 100:1 amplitudes
View attachment 107369



And this is the amplifier response, high IMD of 0.3% (-50dB)
View attachment 107370
The IMD products will be definitely more audible than the 17kHz tone itself.

What does 2kHz solo look like? 6kHz is third harmonic. Obviously there's more than harmonic distortion going on here with the spikes in between the harmonics, in particular at 13kHz. I'd be curious to see a cheap A/B amp/AVR subjected to this test.

 

[March Audio]

Leaving continuous power aside (good), the IMD components depend on how well refined the amplifier technology is. The Aiyima A07 is a budget amp. Here is the other side of the spectrum, Purifi 1ET400A. Artifacts below -120dB.

View attachment 107372

As you show the Hypex and Purifi amps have superb IM ( which is fundamentally related to THD) performance and will beat most AB amps.

The Aiyma is a budget low performance amp and should not be used for any kind of reference or judgement on the fundamental performance of class d.

Amirs Aiyma thd + n measurement.

PURIFI

 

[March Audio]

Leaving continuous power aside (good), the IMD components depend on how well refined the amplifier technology is. The Aiyima A07 is a budget amp. Here is the other side of the spectrum, Purifi 1ET400A. Artifacts below -120dB.

View attachment 107372

Oh another thing to point out with the IM plot you posted is the the power level shown is 100 watts. This is way above any level you would expect to see with music close to 20kHz. Look at just how low the imd products are.

Again it's a demonstration of the idea that class d has high distortion at high frequencies is a total myth.

 

[Matias]

Hyped NCore NC500OEM has IMD artifacts below -95 dB in 4 ohms too. Not as low as Purifi Eigentakt 1ET400A, but still excellent.

 

[restorer-john]

Again it's a demonstration of the idea that class d has high distortion at high frequencies is a total myth.

Amir's latest review of a switching (Class D) amplifier. Check this out:

https://www.audiosciencereview.com/...icon-dd8-review-multichannel-amplifier.19576/

Another example of your "total myth" actually being total reality in the real world. Although the results appear rather strange to me.

All class D unfortunately gets tarred with the same brush due to implementations such as the amplifier reviewed. Sure, that's not fair to the good examples, but it's reality, nonetheless.

 

[tmtomh]

All class D unfortunately gets tarred with the same brush due to implementations such as the amplifier reviewed. Sure, that's not fair to the good examples, but it's reality, nonetheless.

From what I can tell you and pma are the main sources of that kind of overgeneralized tarring of all Class D. So you could just stop doing it and the problem would be cut in half.

 

[SIY]

Alan @March Audio is right that the natural music signal almost always does not contain high frequency components above 5kHz in a full amplitude, so it is not needed to test the amp this way . Similarly he would argue about the 19+20 kHz test. OK, I may agree, however, the high frequency components are almost never present alone, but they are superimposed on a signal of lower frequency and the resulting sum of all components may reach up to full scale, even if the individual high frequency sine waves have low amplitude. It is the truth easily defendable that energy of many many music sample frequency components remains same up to midrange, several kHz, typically 2 - 3 kHz. So I am proposing another test, twin tone IMD 2kHz + 17kHz with ratio of amplitudes 100:1, in other words 2kHz component is 40dB higher than the 17kHz component. I have tested the AIYIMA A07 class D amplifier with this signal - and - as expected, it does not like it. Even if there is a negligible, 100x smaller 17kHz superimposed on the 2kHz wave and is not visible by eye observation, the amp does not like it, because, it sees the sum of amplitudes of the 2kHz and 17kHz sines, so the resulting amplitude is high with fast 17kHz ripple.

This is the test signal, clean spectrum of 2 components 100:1 amplitudes
View attachment 107369



And this is the amplifier response, high IMD of 0.3% (-50dB)
View attachment 107370
The IMD products will be definitely more audible than the 17kHz tone itself.

What are the voltages?

 

[restorer-john]

What are the voltages?

34W@4R according to his plot legend. ~11.5V.

 

[restorer-john]

Another misrepresentation, or lack of understanding.

As you well know John you cant hear above 20kHz. The second harmonic of a 20 kHz signal is at 40 kHz. The 3rd is at 60kHz. The harmonics of fundamental signals above 10kHz (2nd) and 6.7kHz (3rd) fall out of the audible range.

You cant hear them. Your speakers cant reproduce it (dont get pedantic about speakers having some response at 40kHz).

It is therefore irrelevant. It has no bearing on sound quality.

So yes you are perpetuating myths.


Also, as @pma did, you are using a mediocre/poor performing amp as your subject and erroneously extrapolating that as representative of *all* class d performance.

There are poor performing class D amps just as there are poor performing class A/B amps.

View attachment 107496



Purifi

View attachment 107502

I don't think you get it Alan. Surely you understand @amirm's plot I've posted has a 45KHz bandwidth and there is an AUX-40 sitting up front of the AP. Class Ds need that help and this one still is giving a dreadful showing, even with the two filters in place.

There are no harmonics after the 2nd of 20kHz in that result. It's also getting a helping hand due to the rapid falloff in FR above 20kHz as well. The 15kHz has 2nd and third, the audible 10kHz has 2nd, third and fourth and the 5kHz (very audible) has 2nd potentially through to the 8/9th.

So not a myth in this case, just reality for this TI implementation.

 

[restorer-john]

I fully get it thanks.

If you did "fully get it", you wouldn't have gone off on a tangent about inaudible harmonics at 60kHz. There are none in that result. Shall we exclude all the harmonics of 5KHz at circa 0.1% too?

You accused me of spreading a myth about HF THD in Class Ds. I've just shown you an example, reviewed today, on ASR by our host. Clearly not a myth in this case is it?

ou must stop trying to misrepresent any particular example as representative of class D as a whole.

Nobody was. If you took the time to read my post, you would see I was being very specific about the amplifier in question and included the link. I made zero representations as to the results applying to all Class Ds and unlike you, I do NOT make a habit of editing my posts after the event either.

Did you actually read this?

This particular thread is also a general Class D discussion about measurements. It is not an exclusive Hypex and Purifi thread. HF distortion was being discussed, my post, comments and link was, and is highly appropriate.

So please, put me back on ignore like you said you had. Then you won't feel the desperate need to come and attack everything I may write. It's so tiresome.

 

[Blumlein 88]

So is everyone going to endlessly repeat what was already in the other thread that got locked?

You know if measurements can't answer the question what is required? Listening tests. The final arbiter.

So can someone find an example where listening to even test tones that a class AB amp outdoes a class D? One in which the class D is not fully FTC compliant, but the class AB is. If you can provide those examples, then next is music. Results with unrealistic test tones at least shows it is possible. Then if you can't show it with music then someone probably provided the correct dividing line between what was worth doing. OTOH if you also can hear the difference with music then there is some minor superiority to FTC compliant AB amps.

I for one don't want to be on the listening panel using even mid-frequency IMD pairs of signals at max level. The only speaker which would withstand such a signal I believe is a big ESL. I think you are going to fry everything else. Maybe if some sort of short toneburst can be devised it might be possible. Then there is the issue of frying the ears of the listening panel.

Think about how the listening tests limits what you can use due to human ear limits. Which only applies if you can get the signal out without blowing the speakers in use. And those speakers have to have low enough distortion not to be the limiting factor. Again probably only a big ESL has a chance at that.

 

[pma]

It is rather music than sine waves listening that shows the perceived difference. It is then much more difficult to prove by ABX. I do not think that the reason is full power response, rather the content of high harmonics and intermodulation products.

 

[restorer-john]

I do not think that the reason is full power response, rather the content of high harmonics and intermodulation products.

Exactly, Nobody is talking FTC ratings in this part of the discussion. It's HF THD and IM products in the case of this product (the review) from low power to high power across the spectrum.