NC252MP (class D) vs. A250W4R (classAB) burst measurements into 4ohm//2.2uF load

[Vuki]

The whole amp is oscillating at its switching frequency which is part of its functionality. So finding oscillations is like saying day is brighter than night. We don't worry about the switching frequency oscillations in exchange for superb performance elsewhere.

But shouldn't it oscillate at ca. 450kHz?

 

[amirm]

Another one:

MartinLogan Montis loudspeaker Measurements

Sidebar 3: Measurements I used DRA Labs' MLSSA system and a calibrated DPA 4006 microphone to measure the MartinLogan Montis's frequency response in the farfield, and an Earthworks QTC-40 mike for the nearfield and spatially averaged room responses. A note on these measurements is in order:

www.stereophile.com

So? These loads are difficult for any amps, regarding of topology. Here is the note from the measurement page:

"Either way, that this speaker is a difficult load for the partnering amplifier to drive is compounded by the high electrical phase angle, and ameliorated only by the fact that music rarely has high levels of energy in the top octaves."

As noted though, in practice it is not an issue since music energy is so low at 20 kHz. But yes, amps of any topology can be forced into protection, and even damage with such loads. When you sign up to buy a speaker like this, you need to shop carefully for amplifiers. Again, regardless of topology.

 

[amirm]

But shouldn't it oscillate at ca. 450kHz?

It does. That wasn't the point. The point was that class D amps are always oscillating in ultrasonic range. We don't hold that against them because the frequency is so high. And level very low relative to max power.

 

[ahofer]

This may be an ignorant question, but what parts of this are due to lack of independent power supplies as opposed to the 252 design? There is more than one design difference here.

 

[MaxwellsEq]

Perhaps a 60 kHz oscillation is interesting from a scientific or purist perspective. But it’s is irrelevant from an engineering (practical) standpoint. Since the typical speaker load cannot respond to a 60 kHz stimulus, it is not relevant to the analysis of an audio amplifier. So, why measure it and portray this as a “defect” compared to another amplifier topology?

You never want unplanned oscillations at any frequency, ever. The amplifier demonstrates some level of instability into certain realistic loads and may be sensitive to capacitive cabling.

This has got absolutely nothing to do with the presence of content, or whether we can hear it. It's a substantial area of research and @pma is doing a good job at identifying a possible domain limitation of the design.

 

[MaxwellsEq]

This may be an ignorant question, but what parts of this are due to lack of independent power supplies as opposed to the 252 design? There is more than one design difference here.

Absolutely. You can't rule out the difference between SMPS and linear PSUs, either.

 

[Fidji]

So? These loads are difficult for any amps, regarding of topology. Here is the note from the measurement page:

"Either way, that this speaker is a difficult load for the partnering amplifier to drive is compounded by the high electrical phase angle, and ameliorated only by the fact that music rarely has high levels of energy in the top octaves."

As noted though, in practice it is not an issue since music energy is so low at 20 kHz. But yes, amps of any topology can be forced into protection, and even damage with such loads. When you sign up to buy a speaker like this, you need to shop carefully for amplifiers. Again, regardless of topology.

I am driving electrostats with Purifi based amp w/o any issue. On 20dB gain.

 

[MaxwellsEq]

You also haven't got enough evidence here to matter. You have tested only one class D amplifier. That does not mean all class D amps are behaving this way. They may but you better do your homework before making sweeping statements.

You are correct. This is only one data point and does not necessarily prove that any other Class D amplifier behaves in this way. It doesn't prove the NC design or board is the culprit, since we would need to ensure it's not a PSU or wiring issue.

There was a thread a short while ago where @restorer-john reminded us of of the burst-recovery tests that objective magazines used to publish for power amps. What these tests do is scientifically and objectively allow someone to measure differences between two amplifiers which are otherwise largely similar. They show whether an amplifier rapidly gets back on its feet or does it take a while to get back up off the canvas. As an electronic engineer, I think it might be a useful test to have as part of objective measurements of speaker amplifiers.

 

[jhenderson0107]

You never want unplanned oscillations at any frequency, ever. The amplifier demonstrates some level of instability into certain realistic loads and may be sensitive to capacitive cabling.

This has got absolutely nothing to do with the presence of content, or whether we can hear it. It's a substantial area of research and @pma is doing a good job at identifying a possible domain limitation of the design.

Why is it a "substantial area of research" when virtually every loudspeaker is incapable of reproduction of such high (>60 kHz) ultrasonic frequencies? There is no evidence that such oscillation in class D amplifiers undermines the reliability of the amplifier nor the load to which it's connected and it is certainly inaudible.

Speakers within their rooms exhibit far more grotesque distortions. The difference between distortions induced by the speaker/room vs the electronics chain is (to paraphrase Twain) "the difference between lightning and the lightning bug".

The electronics chain preceding the speaker is simply not the tall pole in the tent.

 

[MaxwellsEq]

Why is it a "substantial area of research" when virtually every loudspeaker is incapable of reproduction of such high (>60 kHz) ultrasonic frequencies? There is no evidence that such oscillation in class D amplifiers undermines the reliability of the amplifier nor the load to which it's connected and it is certainly inaudible.

Speakers within their rooms exhibit far more grotesque distortions. The difference between distortions induced by the speaker/room vs the electronics chain is (to paraphrase Twain) "the difference between lightning and the lightning bug".

The electronics chain preceding the speaker is simply not the tall pole in the tent.

This
Is
Not
About
Whether
You
Can
Hear
The
Oscillation

 

[MaxwellsEq]

When you start out designing electronics, you discover an interesting law: all your amplifiers oscillate and your oscillators refuse to oscillate and only amplify.

As you gain experience, you realize that good designers do not make amplifiers that oscillate unexpectedly.

 

[ahofer]

good designers do not make amplifiers that oscillate unexpectedly.

What I'm wondering is whether this was 'unexpected' - given that it may be inaudible itself or in any knock-on effects.

 

[TimF]

An engine for an F35 is designed to be stable, reliable, functionary in specific performance parameters over a designated operating range. The engine will operate with a myriad of secondary and subsidiary harmonics--unavoidable but managed.. Everyone involved damned well knows that the forces under constraint and management in such an engine are such that operations outside of design parameters will be catastrophic. Put a 600 hp engine in a 2,400 pound fiberglass car and hit the gas hard and see what happens. In addition, the engine is designed for a specific mating to a specific airframe. That is the beauty of it: matching the planned and realized operating specs (performance) across its sweet spot (which may indeed be narrow and quite specific with hell to pay for joy rides outside of the sweet spot) with the well-fitted matching component (the airframe). There are only few good matches in terms of airframes, and few good uses for the engine that drives the F35; and there are alternatively a large number of ways to misuse, mismatch, mismanage, poorly operate that engine. It is a good engine, I imagine, for its intended use, and a very poor if not contraindicated engine for 99.5% of all other possible uses. Never will you get perfection. The more you get to perfection over the specific operating range, the more trouble you may get outside that range. Technology shifts are like political revolutions, perhaps. You can imagine the near impossibility of shifting human energy systems from carbon based to alternatives for carbon. The built legacy for carbon conversion is overwhelming both in terms of on the ground plants and equipment but also for the knowledge base.

 

[jhenderson0107]

This
Is
Not
About
Whether
You
Can
Hear
The
Oscillation

Thank you for that condescendingly-formatted clarification.

 

[jhenderson0107]

When you start out designing electronics, you discover an interesting law: all your amplifiers oscillate and your oscillators refuse to oscillate and only amplify.

As you gain experience, you realize that good designers do not make amplifiers that oscillate unexpectedly.

I've only been designing electronics for fourty-five years but have yet to be subject to that "law".

I have seen many unexpected phenomenon in mechanical and electrical systems over that time. I choose to allocate precious R&D time and dollars only to those which affect the required functionality and reliability of the product.

 

[valerianf]

Thank you @pma for this very interesting measurements. As I see that you are using a Picoscope, there is an interesting measurement that could be added. i hooe

 

[valerianf]

Thank you @pma for this very interesting measurements.
As I see that you are using a Picoscope, there is an interesting measurement that could be added. i hope that the Picoscope is able to do a FFT.
With the same test conditions as before, it it possible to inject a 6khz signal and make the FFT?
I would be interested to see what happens at 12 khz and 18 khz with the class D amp.
These are not ultrasonic frequencies!

 

[amirm]

With the same test conditions as before, it it possible to inject a 6khz signal and make the FFT?
I would be interested to see what happens at 12 khz and 18 khz with the class D amp.
These are not ultrasonic frequencies!

Such tests are part of my amplifier tests. See for example:

Black line is 5 kHz. As you see, prior to clipping, it stays below -95 dB. And that is sum total of harmonics and noise up to 45 kHz.

 

[valerianf]

@amirm "Such tests are part of my amplifier tests."
I did not find the same test for the Benchmark AHB2.
May be I missed it.
I am not sure that the AHB2 will represent the same shape of distortion, and once again the FR calculation is not the same for class D and class AB at the advantage of class D.

Why not measuring the 2 amps classes with the same bandwidth (the one for class D) and compare the results?
My point is comparing a modern Class AB and a modern Class D amp using exactly the same tests, if possible going closely to the maximum power output capability or around 25W where it seems to be worst..
Everybody knows that the weakness of class D is the distortion increasing with the frequency (ASR measurement shows it) and a possible hiss at high frequencies.
Regarding the hiss, it can be measured using a microphone: it would be nice to add this measurement for all equipements using class D amps (i.e. active speaker).

 

[pma]

The whole amp is oscillating at its switching frequency which is part of its functionality. So finding oscillations is like saying day is brighter than night. We don't worry about the switching frequency oscillations in exchange for superb performance elsewhere.

The small difference is if the amp is oscillating at 380 kHz with resistive load or if the 66kHz oscillations occur after connecting to capacitive load, and with much higher amplitude. And such oscillations are a result of change of operating conditions of the switching amplifier. One could learn from Bruno in the Hypex Ncore thread (diyaudio, about 10 years ago) that the limit for additional capacitance that does not affect the operating conditions of Ncore is 200nF. The reason is simple, output LC filter uses 2uF capacitance and 2.2uF added in my test completely change working conditions of the amplifier. Ncore is just not designed to work with such load and speakers with similar impedance should be avoided. This information is valuable for the potential customers and not something you should fight against. You cannot protect something that is unprotectable.