On Class D Amplifiers Measurements

[dc655321]

And asking the best class D designers, when are you going to get rid of such circuit behaviour???

If you were designing a class D amp, what would you do to address (suppress) this?

 

[Matias]

1kHz and its harmonics is where the ear is most sensitive, therefore designers prioritize this region for lowest distortion, and reviews show it as well.

Distortion above 10 kHz is irrelevant because:
1. Musical signal is way below full scale, so its harmonics even lower than on mids and bass; and
2. Harmonics are above absolute hearing threshold (20+ kHz) and below low level audibility.

Good old Fletcher-Munson curves say it all.

 

[tmtomh]

Not quite following this. Are you saying that these artifacts, all under -100db are audible? 5Khz iz the fundamental here, right?

He's likely going to claim they are not all under -100dB and/or that they are indeed audible in some conditions. That's been his pattern before RE this issue.

 

[antcollinet]

He's likely going to claim they are not all under -100dB and/or that they are indeed audible in some conditions. That's been his pattern before RE this issue.

I'm sure @pma can speak for himself.

 

[tmtomh]

I'm sure @pma can speak for himself.

And he most assuredly will.

 

[pma]

I would like to see a valid evidence (a proper set of tests) that the class D distortion and mess above 20kHz is unimportant. Not because of thresholds of hearing (everyone knows Fletcher-Munson curves for some 70 years at least) but to make sure that the class D ultrasound distortion cannot create audible intermodulations in tweeters. Is this possibility excluded? Is there any proof?

 

[antcollinet]

I would like to see a valid evidence (a proper set of tests) that the class D distortion and mess above 20kHz is unimportant. Not because of thresholds of hearing (everyone knows Fletcher-Munson curves for some 70 years at least) but to make sure that the class D ultrasound distortion cannot create audible intermodulations in tweeters. Is this possibility excluded? Is there any proof?

How would you test for this?

Or perhaps more to the point, how many tweeter designs would you need to test, to rule out the possibility in all tweeters?

 

[jae]

I would like to see a valid evidence (a proper set of tests) that the class D distortion and mess above 20kHz is unimportant. Not because of thresholds of hearing (everyone knows Fletcher-Munson curves for some 70 years at least) but to make sure that the class D ultrasound distortion cannot create audible intermodulations in tweeters. Is this possibility excluded? Is there any proof?

Record your tweeters and show us the results !

 

[boXem]

Sorry I am back again with this thread . It was the Amir's review

Audiophonics HPA-S400ET Review (Stereo Amplifier)

This is a review and detailed measurements of the Audiophonics HPA-S400ET. It was sent to me by the company after member request and costs 1 490,00 € (US $1,638). I must say this is one of the best packaging of class D amplifiers (Purifi in this case) that I have seen! It is slick and has...

www.audiosciencereview.com

and especially this measurement

that has shown that even the best current class D module, Purifi 1ET400, suffers from quite severe high frequency non-linearity. Thank you @amirm for performing this kind of @45kHz BW distortion vs. power test as a rule. I have decided to accept it as my standard test method as well and I have tested 3 of my amplifiers under same test conditions, the comparative result I have posted here in this thread:

45kHz BW distortion study of 3 amplifier topologies (2xAB, 1xD) + A/B test

Following the discussion in the Audiophonics HPA-S400ET thread on Purifi HF non-linearity, I have decided to make a comparative measurement of 3 of my amplifiers. THD vs. power into 4ohm with @45kHz bandwidth. The amplifiers under test are: 1) A250W4R, 250W/4ohm class AB amplifier based on...

www.audiosciencereview.com

The only difference from Amir's test is that I plot THD (without N noise component) vs. power so I can better see distortions at low level which would be masked by noise in the THD+N test.

The result with Hypex NC252MP module is this>
View attachment 196111
It is a bit worse than the Purifi result however similar behaviour. And the result is worse than the class AB amplifiers that I have measured and posted in the thread linked here above.

It is an ultrasonic distortion, you will say. Correct, The poor behaviour results from distortion components above 20kHz. Does it matter, or is it OK? Though 1kHz spectrum from NC252MP is nice, the 5kHz and higher are NOT nice. There are not only ultrasonic harmonics, but also non-harmonic distortion products, like

View attachment 196114

This cannot be compared to a very good class AB amplifier with clean spectra free of non-harmonic products. So, I am taking up the glove and walking against mainstream again . At least raising an unpopular question. And asking the best class D designers, when are you going to get rid of such circuit behaviour???

1. @amirm measures THD+N. You measure THD. It could be just N and not THD on the Purifi. That is a measurement I should do when I have some spare time.
2. while I agree that the spectrum is a bit dirty, harmonics level is not higher than in the audible band. THD is higher because high order harmonics are still there while absent in audible band
3. I would not exclude that some of this is is linked to the output filter being "only" of second order type, and is a bit weak as an anti-imaging filter

 

[SIY]

I would like to see a valid evidence (a proper set of tests) that the class D distortion and mess above 20kHz is unimportant. Not because of thresholds of hearing (everyone knows Fletcher-Munson curves for some 70 years at least) but to make sure that the class D ultrasound distortion cannot create audible intermodulations in tweeters. Is this possibility excluded? Is there any proof?

So:

1. Prove a negative.

2. And do it for something that, when you look at impedances (electrical and mechanical) involved, is implausible.

But this isn't a jihad or anything, right?

 

[RichB]

I would like to see a valid evidence (a proper set of tests) that the class D distortion and mess above 20kHz is unimportant. Not because of thresholds of hearing (everyone knows Fletcher-Munson curves for some 70 years at least) but to make sure that the class D ultrasound distortion cannot create audible intermodulations in tweeters. Is this possibility excluded? Is there any proof?

There is also the possibility that these test measurements expose distortion that is not the worst case distortions.
The Stereophile simulated speaker measurements show some variance in load. Some amps dropping .5 dB into 4 ohms as well.

I don't think the transparency of amps is a given because we would have to test the into reactive loads, consider linearity issues as distortion, and measure the THD during such tests.

- Rich

 

[tmtomh]

I would like to see a valid evidence (a proper set of tests) that the class D distortion and mess above 20kHz is unimportant. Not because of thresholds of hearing (everyone knows Fletcher-Munson curves for some 70 years at least) but to make sure that the class D ultrasound distortion cannot create audible intermodulations in tweeters. Is this possibility excluded? Is there any proof?

This doesn't make any sense, because the source of IMD in tweeters (or any speaker driver) is not harmonic distortion from the amplifier, but rather simply the existence of ultrasonic frequencies in the signal the amp is sending to the speakers - regardless of how many or how few of those ultrasonic frequencies are pure signal vs distortion products.

The In fact, the quantity and amplitude of the ultrasonic content going to one's speakers depends on the frequency bandwidth of the amp, along with the amp's inherent distortion (THD and IMD) and the frequency extension of the musical source itself.

So a Class AB amp that has, say, higher THD and IMD than a Purifi-based amp, and higher bandwidth (say 100kHz as restorer-john is always advocating, vs. a typical Class D filtered limit of around 40kHz), could actually output more ultrasonic content than the Class D amp, and depending on the bandwidth of the tweeter could present a worse case for tweeter IMD than the Class D amp.

This is just another instance of @pma 's obsession with top-octave distortion in Class D amps being higher than their distortion at 10kHz ad below - even when that top-octave distortion is *lower* than that of a comparable Class AB amp across that amp's entire bandwidth.

 

[SIY]

I don't think the transparency of amps is a given because we would have to test the into reactive loads, consider linearity issues as distortion, and measure the THD during such tests.

I showed this data in my last review of the GaN Systems amp and the about-to-appear review of the Orchard Starkrimson Ultra.

 

[freemansteve]

As shown elsewhere, it surely comes down to bangs per buck. Or pops and clicks for vinyl guys. Or marginalities concerning stuff most of us can't hear.

In the UK I can get an amazing Class-D amp for £450, eg Nord. What AB (etc) for £450 is going to be better?

Sorry, but I'm on a budget that rules me out of the discussion! I certainly can't afford the test gear or the time trying to listen to sine waves all day - we have Amir for that!

 

[pma]

So:

1. Prove a negative.

2. And do it for something that, when you look at impedances (electrical and mechanical) involved, is implausible.

But this isn't a jihad or anything, right?

It is not a jihad. So prove, based on scientific test methods, that distortion >20kHz makes no problem and cannot create audible intermodulations in any tweeter. I am waiting.

The best way is not to create possible issues. Not to defend all the compromises, "because we like" the product, topology etc. The way to the hell. Even Bruno used to say that.

 

[SIY]

It is not a jihad. So prove, based on scientific test methods, that distortion >20kHz makes no problem and cannot create audible intermodulations in any tweeter. I am waiting.

"Prove a negative."

The one example I had on hand (NCore with a metal dome tweeter) showed zero problem. You were pretty quiet about that.

 

[restorer-john]

It's pretty simple on ASR. If you are a Class D amplifier, the rules of the game are different.

You don't have to answer for any sins outside the 20-20kHz frequency range. Phew. That's a relief! You get a free pass for anything untoward that 'humans' can't hear. You even get an expensive filter to hide your spurious output from analysis. Wow, that's awesome!

Because you have 'special' topology, the generally accepted testing parameters and procedures are skewed to suit your strengths, cover up your weaknesses and show you in your best light. How good is that?

How about the classic HPF at fc=400Hz for hum and noise? That helped tube and class A/AB amps and was used for decades? Where did that go?

If you are a class AB amplifier or class A, you are required to demonstrate your abilities up to at least the 10th harmonic when you specify your performance. But your 1kHz THD+N will be all that ranks you on the ASR leaderboard. Forget about any other frequencies- they don't count.

Sssh, nothing to see here. Move along. Yawn.

 

[SIY]

It's pretty simple on ASR. If you are a Class D amplifier, the rules of the game are different.

You don't have to answer for any sins outside the 20-20kHz frequency range. Phew. That's a relief! You get a free pass for anything untoward that 'humans' can't hear. You even get an expensive filter to hide your spurious output from analysis. Wow, that's awesome!

Because you have 'special' topology, the generally accepted testing parameters and procedures are skewed to suit your strengths, cover up your weaknesses and show you in your best light. How good is that?

How about the classic HPF at fc=400Hz for hum and noise? That helped tube and class A/AB amps and was used for decades? Where did that go?

If you are a class AB amplifier or class A, you are required to demonstrate your abilities up to at least the 10th harmonic when you specify your performance. But your 1kHz THD+N will be all that ranks you on the ASR leaderboard. Forget about any other frequencies- they don't count.

Sssh, nothing to see here. Move along. Yawn.

Still waiting for someone who frets about Class D (usually people whose decades of expertise have been rendered obsolete almost overnight) to show the horrible effects of millivolt ultrasonic signals running down a cable to a coil attached to a moving mass. You claim there's an issue, demonstrate it. I can't find it anywhere in any of the Class D amp tests I've done (though admittedly I haven't tested any super-cheap amps, not have I tested any super-cheap old-fashioned amps, either).

 

[RichB]

I showed this data in my last review of the GaN Systems amp and the about-to-appear review of the Orchard Starkrimson Ultra.

It may be me, but did not find in with Search. Do you have a link?

- Rich

 

[SIY]

It may be me, but did not find in with Search. Do you have a link?

- Rich

I don't think it's online yet, just in the print publication.