• 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!

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

I know I'm jumping into this late, but measurements at 5 watts/1kHz seem much more relevant to typical home listening conditions than measuring 200+watts at 10-20kHz, which, if actually delivered to a typical loudspeaker, would instantly blow out the tweeter, not to mention your eardrums.

These tests are interesting and show a potential pitfall of Class D designs for professional amplification where you might actually call on an amplifier to deliver that kind of power at those kinds of frequencies but these tests seem totally irrelevant to the typical use cases home audio Class D amps are designed for.

Surely you read the parts of thread where measurements were made at various levels and frequencies?


Most interesting to me is the rather noisy spectrum with no input as seen in the THD+N vs level sweep. Maybe it isn't there with an actual electrostatic speaker but would love to see the measurement.

@AdamG247 You said you have ESLs with a NC502MP, any interest in making some acoustic measurements?

Michael
 
May be I am completely wrong, but it looks like pma using such hype threads as promotion for his custom amplifiers compared with some popular amp like Hypex that could not handle some specific tests. Just an impression from not technical person... Sorry.
 
Hi Michael,

Yes I have ESL-X’s but not interested in getting into the middle of this debate. I prefer to remain neutral and just observe. Hope you understand my decision. Enjoy your weekend Sir.
 
May be I am completely wrong, but it looks like pma using such hype threads as promotion for his custom amplifiers compared with some popular amp like Hypex that could not handle some specific tests. Just an impression from not technical person... Sorry.
If you read it all - as I certainly didn't on the first pass - it's pretty clear he's being both positive and useful here

to be fair, the top post could do with an update and aggregation though to save half the thread being pointers to stuff on page 2, 10, 14 etc
 
Last edited:
We get it, you have no intellectual curiosity, just move on.

Michael

I must confess that I don't quite get why, of all the bases upon which someone could challenge or disagree with my prior comment (and there are many!), someone would chose something that so clearly makes no sense.

It's ironic as well, since dismissing someone's detailed comment and argument as you have is a great example of what it actually looks like when someone displays a lack of intellectual curiosity. :)
 
It would.be nice to start seing amplifier tests with an equivalent speaker circuit being implemented as part of the test side by side with the THD and the other ones we know already.

And maybe, given time, a revisit of some of the already tested amps.

That's the hope. And the next step.

Makes sense, thanks. So then what would qualify as a proper "equivalent speaker circuit"? In other words, much of this thread has been concerned with discussion of speaker loads that are atypical or particular to certain types of speakers. So logically it would seem that one would need to test amps with multiple simulated speaker loads to cover multiple different types of use cases and speaker pairings, yes? Not necessarily 10 or 15 different loads/circuits, but based on the discussion here, surely more than one, yes?

I honestly am not trying to be argumentative or disagreeable. Rather, I am asking about this because if @amirm were to add a simulated speaker load of some kind to his testing regimen, what would stop folks from making the same objection they have been making, on the grounds that this simulated load does not tell us how the amp will behave with a particular kind of speaker?
 
@tmtomh I am not going in details hire but you don't want bad implementations of Hypex modules from small manufacturers deliberately skipping to implement proper protection because it would hurt performance a bit and when they charge you a lot for not great implementation of OPAMP's on such (which again will degrade performance a bit) and is still a bare naked design which now costs significantly more with filmsy warenty my answer is hell no. Otherwise it's entirely your own choice. If you want a solid Hypex implementation with good warranty, long lasting support and part's availability you can but it won't be cheap. If you prefer to gamble well I don't. PMA isn't doing a wrong thing verifying the design nor insisting on more rigorous testing. May be standard he proposed in doing so isn't best nor absolutely suitable but be my guest in improving it or proposing a better one. Who ever menage to really improve methodology even a little bit has my aplauz.

Sigh. This is precisely the problem with this discussion. "I am not going into details" but "you need a good implementation" is precisely a less specific version of what pma does (albeit I certainly agree that he does it in more detail and with his own, original measurements, which I do respect and admire - have to give him his due for that). We know that a good implementation is important, but it has not been established that the amps pma tests and criticizes are not in fact good implementations. And your contribution to this question is, "I am not going in[to] details here." That's the definition of FUD.
 
Hi Michael,

Yes I have ESL-X’s but not interested in getting into the middle of this debate. I prefer to remain neutral and just observe. Hope you understand my decision. Enjoy your weekend Sir.

I understand, it is a big ask. Have a good weekend yourself!

Michael
 
Sigh. This is precisely the problem with this discussion. "I am not going into details" but "you need a good implementation" is precisely a less specific version of what pma does (albeit I certainly agree that he does it in more detail and with his own, original measurements, which I do respect and admire - have to give him his due for that). We know that a good implementation is important, but it has not been established that the amps pma tests and criticizes are not in fact good implementations. And your contribution to this question is, "I am not going in[to] details here." That's the definition of FUD.
I stated that I am not interested but then again I am not criticising anything you are. So please go ahead into details and present facts proving your point and not colored in any way because that whose your point to begin with (as we all are biased after all). Until you are able to do so beyond reasonable doubt and based only on facts (experimental methodical scientific) at least don't make subjective speculative claims publicly. Do more talk less. End of a discussion for me (it never actually even whose).
 
It would.be nice to start seing amplifier tests with an equivalent speaker circuit being implemented as part of the test side by side with the THD and the other ones we know already.

And maybe, given time, a revisit of some of the already tested amps.

That's the hope. And the next step.
Which equivalent speaker. They are all different. Different ones may work well with one amp and not with another. The work well / badly might be reversed with another speaker.

It will not tell us very much.
 
May be I am completely wrong, but it looks like pma using such hype threads as promotion for his custom amplifiers compared with some popular amp like Hypex that could not handle some specific tests. Just an impression from not technical person... Sorry.
AFAIK, he doesn't sell his amps. If that's not the case, I'm sure he'll correct me.
 
If someone able is to make an equivalent circuit it is absolutely possible.
Equiv. circuit only works for small signals/lower power. Speaker impedance is power sensitive as induction and such changes with that. In my speaker impedance measurements, if I change the drive voltage, the output distinctly changes.
 
AFAIK, he doesn't sell his amps. If that's not the case, I'm sure he'll correct me.
My understanding also. Perhaps @Leva should apologise for a fairly nasty insinuation.
 
The AudioGraph Active LoadBox offers 20 different load impedance settings (8, 4, 2, 1 ohms each with 5 different phase angles).

ap_powercube.png


Note that in the specifications, the complex impedance phase angles are specified only at 1 kHz. As far as I understand, the phase angle and impedance magnitude (as least when the phase angle is not zero) are frequency dependent. Which means, if I m not wrong, these complex impedance values are only valid at 1 kHz.

This raises a number of questions. As we saw in this thread, the impedance of the load at frequencies far away from that of the signal, can have a significant effect. I was unable to find the impedance curves of the Active LoadBox with frequency (there will be 20 of them, for each of the 20 different settings). Since it seems the impedance values are only applicable to a 1 kHz test signal, will this Active LoadBox be able to answer enough of our questions?

@amirm May I ask a couple of questions:
  1. Can this box be run at frequencies other than 1 kHz, if the impedance setting is not purely resistive? If it can, is it 20 - 20k Hz and is there any de-rating required?
  2. Can you run a few impedance sweeps to measure the effect of frequency on the load impedance? As it was shown in this thread that a capacitive load which gives very low impedance at ultrasonic frequencies can impact the amplifier performance in the audible band.
 
I stated that I am not interested but then again I am not criticising anything you are. So please go ahead into details and present facts proving your point and not colored in any way because that whose your point to begin with (as we all are biased after all). Until you are able to do so beyond reasonable doubt and based only on facts (experimental methodical scientific) at least don't make subjective speculative claims publicly. Do more talk less. End of a discussion for me (it never actually even whose).

You did state you're not interested - which is evidently untrue based on any reasonable reading of the number and content of your posts about this.

My prior posts are full of detail, including specific examples of precisely what factors he does and does not take into account. I make no subjective claims except when clearly stated in the context of putting forth an argument and usually clearly marked as IMHO. So whomever's mode of discourse you're disagreeing with, it's not mine.
 
I agree, but sadly I’d say @pma has a lot more than “a little” knowledge. He’s just so intent on pursuing a single-minded argument about Class D that he doesn’t care what key information he has to ignore or how many parameters he has to strip away in order to come up with the graphs he posts.

Designing tests that are not representative of a real world scenario, and then claiming those tests have important real world implications, is unfortunately not a result of a lack of knowledge; it’s his entire purpose. And he doesn’t seem to care how misleading his posts end up being as a result.

Not so sure. Designing tests that don't represent real world scenarios betrays lack of understanding to my mind. It's also a pretty basic error to not think about class d operation and the effect of a contrived load that's effectively a short circuit at the switching frequency.
 
Last edited:
Equiv. circuit only works for small signals/lower power. Speaker impedance is power sensitive as induction and such changes with that. In my speaker impedance measurements, if I change the drive voltage, the output distinctly changes.
I’ve never measured loudspeakers, but have designed/measured many passive filters. Equivalent circuits hold up well until there is enough nonlinearity in the system to force an accommodation in the model. Looking at a few of your speaker reviews, I see THD measurements around 1W and 10W (? Guessing based on 86 dB and 96 dB SPL - please correct me). The THD numbers are only a few percent. Even at 10%, the impact to an equivalent circuit would be tiny. So am not following the small-signal argument here unless you’re referring to many tens of watts as the comparison point? Feels like I’missing something, apologies in advance!

Re: equivalent circuit questions, idea is to generate an LRC circuit that mimics the speaker’s behavior over the full audio range, and maybe somewhat beyond if useful. At a single frequency, these two circuits are equivalent: a single capacitor with Xc = -j50, and a series LC with Xc = -j90 and Xl = +j40. At any other frequency, these two different circuit topologies diverge. That’s why describing an equivalent circuit at only one frequency is incomplete/insufficient.
 
I've done THD vs frequency sweeps with loudspeaker loads (dynamics, not yet electrostatics) and just have not seen much difference with engineered amps between that and resistive measurements, at least at tolerable power levels from a neighbors-call-the-cops perspective.
Okay. That sounds like anecdotal, no?
But much much closer than either a large capacitor or a series resonant load.
I do know either isn't reflecting real world loads.
 
Back
Top Bottom