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

[Deleted member 48726]

I seriously doubt that this speaker presents a capacitive load to the amplifier above 50 kHz.

It can't be reactive if it's a short..

 

[fpitas]

On the positive side, pma's amp puts up with it quite well. So if you just insist on owning these things, there's that.

 

[Hapo]

To be fair, Hypex is very clear in its specs about minimum design load. And this corner case is outside of spec. I could also drop line voltage to something below the rated minimum and show that the amplifier doesn't perform properly under those conditions, either.

...this...

...between reading pma's "unrealistic" test and the clearly specified parameters of the Chinese amp modules (even) I was able to determine that one cheap class D amp might be better for my particular conditions than another...

 

[Hapo]

On the positive side, pma's amp puts up with it quite well. So if you just insist on owning these things, there's that.

...I'd like to see him get off his high horse, drop the apparent agenda, and repeat his test with a Purifie 1ET4070SA amplifier...

...I would like to buy amps that pass this "unrealistic" test...he might be onto something here...

...I'd like to see him do the test with an AB amp on the order of, say, a Dayton APA 150... .!..!.

I can only imagine a lot of garden variety A/B amps get upset driving those things.

 

[fpitas]

...I'd like to see him get off his high horse, drop the apparent agenda, and repeat his test with a Purifie 1ET4070SA amplifier...

...I'd like to see him do the test with an AB amp on the order of, say, a Dayton APA 150... .!..!.

Yes, I'm happy enough seeing the testing. Like all testing though, especially with fringe performance, it's open to interpretation.

 

[Deleted member 48726]

You earned a two week timeout for this one. Maybe that will give us enough time to study up on how to read and interpret an impedance plot.

I'm curious to whether warnings are given before this kind of timed exclusion? I mean it seems to me no one other really are doing these critical tests and seeks to reveal data not apparent in current test protocols. It would IMO be a shame to drive people like pma away, frustrations or not. -Of course the language used should be nicer than this last insult of his.

 

[Hapo]

...I can't read plot charts so I didna take offence...

...can't we all just get a bong...???...

 

[fpitas]

...I can't read plot charts so I didna take offence...

I can, but sometimes a snide comment is just a snide comment.

 

[Deleted member 48726]

...I can't read plot charts so I didna take offence...

You're hilarious with your remarks and cavemen..

 

[Hapo]

...I amuse myself...that is what matters... (grin)

...I wouldn't ban someone like pma over a snide comment...I just ignore them when they get like that...

...people were bagging on his test and calling his work irrelevant...that's sort of snide, eh...???...

...I want him to post more tests and upset every one with facts...

...IMHO that is what is really going on here...

...did y'all see where I posted a main line brand name speaker that I am looking at that claims a nominal 8 ohm impedance that just might actually be 2.5 ohms sometimes...???...

...I recall when 16 ohms was common, and 2 ohms was a short...what do you think will happen if I hook these speakers up to any amp that is rated @ 4 ohms...???...

...this last is an actual question...

 

[PeterOo]

Regarding this plot:

I will reiterate what I mentioned in an earlier post, that the designer of this loudspeaker should be shot. The impedance trends to a dead short above 20kHz. This is extremely negligent design practice. Besides this one speaker, what else you got? You know there have been tens of thousands of speakers designed with "normal" loads.... why focus on this one?

From the Stereophile review of this loudspeaker:

But this impedance graph corresponds to the Brilliance Control(BC) being set to maximum which results in an augmentation of 25dB at 23 kHz relative to the middle position of BC. This middle position “seemed to give the the flattest measured response“.
The minimum setting of BC resulted in 4 Ohm at 20 kHz and -7 dB at 23 kHz. What impedance at 20 kHz in the middle position was I could not find. Somewhere between 4 and 0 Ohm probably.

So even this one in a million electrostatic LS does not even show this almost zero impedance pma simulated unless you completely FU your frequency response. What is the relevance of that?

With the middle position of BC giving an almost flat FR it seems the designer apparently knew how to design a proper speaker. He just gave the user the opportunity to artificially augment frequencies which a large part of their user cannot even hear anymore, and destroy their amps in the process.
I think having him shot is a bit much, but a thorough spanking might be called for.

Attached the FR with BC at max (corresponding to impedance graph used by pma) and BC at minimum, both normalised to BC at middle position.

Link to full stereophile measurememts

Backing off the BC to the Minimum position raised the impedance at 20kHz to a more reasonable 4 ohms. However, this drastically reduced the sonic level of the treble, as can be inferred from fig.2, which shows the effect of the BC referenced to the response with the control set at 12 o'clock (middle curve, which therefore appears as a straight line), which seemed to give the flattest measured response. The control gives a massive maximum response change of 32dB at 23kHz, which is quite excessive, in my opinion.

 

[PeterOo]

...people were bagging on his test and calling his work irrelevant...that's sort of snide,

If you do measurements on a simplified model and you get unexpected results that have not been reported on actual speakers, the first thing a scientist or engineer should do is
1) try to understand what is happening (shorting of switching frequency? Impedance out of spec? Do ask your peers for input!)
and 2) verify your measurement is not an artefact of your simplified model.

If you instead start a separate thread on a different amp using this unvalidated model, start an ABX thread to see if other people can hear how terrible it sounds and not engage in a critical discussion with your peers (which is what ASR is for) on the validity of your model, I think you can expect (deserve) some criticism.

 

[atmasphere]

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.

FWIW, This is a weakness of class D amplifiers, when presented with a capacitive load. The output inductor can resonate with the capacitance of the speaker (perhaps an ESL). This is one argument for self-oscillating class D amps, as they can have enough feedback to minimize or prevent excitation of the resonance. I know of at least one class D amplifier that apparently will fail if run on an ESL for extended periods. I don't know if the amp went into oscillation and failed or if the HF resonance was causing it to overheat so until more is known, this bit remains anecdotal.

This has been a problem in a number of traditional class A/AB amplifiers as well. I remember some of them from the 1980s that would oscillate when presented with Polk Audio speaker cable of the period, which was highly capacitive. IMO, this indicates a problem in the feedback loop design.

IIRC the Hypex is a self-oscillating design with quite a lot of feedback. It is surprising to see that 60KHz sine.

Edit: I just figured out this thread has 29 pages...

Somewhere between 4 and 0 Ohm probably.

So even this one in a million electrostatic LS does not even show this almost zero impedance pma simulated unless you completely FU your frequency response. What is the relevance of that?

The Sound Lab is an ESL and ESLs don't play by the same rules that box speakers do. One of the more obvious reasons is there's no box so no resonance. The electrostatic principle produces an impedance curve that is not a map of the speaker efficiency in the way that it is for a box speaker. So in the bass where the impedance is high you need the same power (not voltage) as you do at 10KHz where the impedance is much lower in order to get the same sound pressure. The ideal amp would make constant power into any load rather than constant voltage.

This speaker is often used with tube amplifiers, which can have a high output impedance. Depending on the position of the Brilliance control, the impedance at 20KHz is 1.5 to 3 Ohms. Many tube amps will not make power into lower impedances, hence the control. Its meant to correct the speaker so as to match the voltage response of the amplifier, in much the same way that vintage horn speakers have a midrange and tweeter level control. They aren't there to adjust the speaker to the room (a common misconception)!

(historical trivia: level adjustment controls on speakers went away when the voltage rules became accepted in the industry. ElectroVoice and MacIntosh led the way on this account in the late 1950s.)

The Sound Lab also is set up to adjust the bass levels, as amps with a higher output impedance might make too much bass, while many solid state amps won't make any (like most ESLs, the Sound Lab impedance varies by about 10:1 from the bass to the highs). You can see that if the speaker does not obey the voltage rules that are commonly accepted in box speakers that adjustment ability is needed. I don't think Stereophile really understood that.

 

[PeterOo]

FWIW, This is a weakness of class D amplifiers, when presented with a capacitive load. The output inductor can resonate with the capacitance of the speaker (perhaps an ESL). This is one argument for self-oscillating class D amps, as they can have enough feedback to minimize or prevent excitation of the resonance. I know of at least one class D amplifier that apparently will fail if run on an ESL for extended periods. I don't know if the amp went into oscillation and failed or if the HF resonance was causing it to overheat so until more is known, this bit remains anecdotal.

This has been a problem in a number of traditional class A/AB amplifiers as well. I remember some of them from the 1980s that would oscillate when presented with Polk Audio speaker cable of the period, which was highly capacitive. IMO, this indicates a problem in the feedback loop design.

IIRC the Hypex is a self-oscillating design with quite a lot of feedback. It is surprising to see that 60KHz sine.

Edit: I just figured out this thread has 29 pages...

The Sound Lab is an ESL and ESLs don't play by the same rules that box speakers do. One of the more obvious reasons is there's no box so no resonance. The electrostatic principle produces an impedance curve that is not a map of the speaker efficiency in the way that it is for a box speaker. So in the bass where the impedance is high you need the same power (not voltage) as you do at 10KHz where the impedance is much lower in order to get the same sound pressure. The ideal amp would make constant power into any load rather than constant voltage.

This speaker is often used with tube amplifiers, which can have a high output impedance. Depending on the position of the Brilliance control, the impedance at 20KHz is 1.5 to 3 Ohms. Many tube amps will not make power into lower impedances, hence the control. Its meant to correct the speaker so as to match the voltage response of the amplifier, in much the same way that vintage horn speakers have a midrange and tweeter level control. They aren't there to adjust the speaker to the room (a common misconception)!

(historical trivia: level adjustment controls on speakers went away when the voltage rules became accepted in the industry. ElectroVoice and MacIntosh led the way on this account in the late 1950s.)

The Sound Lab also is set up to adjust the bass levels, as amps with a higher output impedance might make too much bass, while many solid state amps won't make any (like most ESLs, the Sound Lab impedance varies by about 10:1 from the bass to the highs). You can see that if the speaker does not obey the voltage rules that are commonly accepted in box speakers that adjustment ability is needed. I don't think Stereophile really understood that.

Thanks for the background info. So if the high BC is to correct for tube amps with high output impedance, then why use that setting for an amp with low output impedance? (Like pma is doing in his measurements).

 

[MaxwellsEq]

FWIW, This is a weakness of class D amplifiers, when presented with a capacitive load. The output inductor can resonate with the capacitance of the speaker (perhaps an ESL). This is one argument for self-oscillating class D amps, as they can have enough feedback to minimize or prevent excitation of the resonance. I know of at least one class D amplifier that apparently will fail if run on an ESL for extended periods. I don't know if the amp went into oscillation and failed or if the HF resonance was causing it to overheat so until more is known, this bit remains anecdotal.

This has been a problem in a number of traditional class A/AB amplifiers as well. I remember some of them from the 1980s that would oscillate when presented with Polk Audio speaker cable of the period, which was highly capacitive. IMO, this indicates a problem in the feedback loop design.

IIRC the Hypex is a self-oscillating design with quite a lot of feedback. It is surprising to see that 60KHz sine.

Edit: I just figured out this thread has 29 pages...

The Sound Lab is an ESL and ESLs don't play by the same rules that box speakers do. One of the more obvious reasons is there's no box so no resonance. The electrostatic principle produces an impedance curve that is not a map of the speaker efficiency in the way that it is for a box speaker. So in the bass where the impedance is high you need the same power (not voltage) as you do at 10KHz where the impedance is much lower in order to get the same sound pressure. The ideal amp would make constant power into any load rather than constant voltage.

This speaker is often used with tube amplifiers, which can have a high output impedance. Depending on the position of the Brilliance control, the impedance at 20KHz is 1.5 to 3 Ohms. Many tube amps will not make power into lower impedances, hence the control. Its meant to correct the speaker so as to match the voltage response of the amplifier, in much the same way that vintage horn speakers have a midrange and tweeter level control. They aren't there to adjust the speaker to the room (a common misconception)!

(historical trivia: level adjustment controls on speakers went away when the voltage rules became accepted in the industry. ElectroVoice and MacIntosh led the way on this account in the late 1950s.)

The Sound Lab also is set up to adjust the bass levels, as amps with a higher output impedance might make too much bass, while many solid state amps won't make any (like most ESLs, the Sound Lab impedance varies by about 10:1 from the bass to the highs). You can see that if the speaker does not obey the voltage rules that are commonly accepted in box speakers that adjustment ability is needed. I don't think Stereophile really understood that.

Fascinating! Thank you for this contribution, I learned some things I didn't know, which always makes me happy

 

[theREALdotnet]

I guess pma found this "1 in a million" speaker that is representative of his load, and indeed it breaks the NCore amp.

Does it though? I think apart from raised THD and ultrasonic noise the NCore was coping fine.

Unlike some Naim amp I’ve known, which would have spilled its guts if I had waved a 2.2µF capacitor within 10cm of its output terminals…

 

[Rottmannash]

...if only Bruno would chime in and settle this once and for all.

 

[Hapo]

...no one is going to answer my question, are they...

...methinks something bad would happen...

 

[jimk1963]

Just dawned on me, this "extreme" test load of 4 ohms || 2.2uF is not as low of an impedance in the audio band as I had guessed, based on all the comments.

Freq (kHz) Zmag (Ohms) Phase(Z) (Degrees)
1 4.0 -3.1
10 3.5 -28.9
20 2.7 -47.9
31 2.0 -59.7
70 1.0 -75.5

At or below 20 kHz, this extreme load still has an impedance of 2.7 ohms. That's low, yes, but believe I've read on here that plenty of speakers have 2 ohm minimums - is that correct?

Seems it's above the audio band where this dummy load is wreaking all the havoc with the NC252MP - is that a fair assumption? I guess these amps are pretty wideband, looking at @amirm 's plots.

Do speaker manufacturers usually specify impedance much beyond 20 kHz? Wondering if there could be more speakers out there like that A1 that someone talked about earlier, that dipped down to nearly zero ohms above the audio range. If it isn't spec'd, how would one know this? Apologies in advance if dumb question.

Or is it not so much the impedance magnitude, but rather the large phase shift at, say 10 kHz and above, that causes the amp to degrade?

It was this comment from @atmasphere that made me wonder about the "root cause" that explains OP's findings:

"FWIW, This is a weakness of class D amplifiers, when presented with a capacitive load. The output inductor can resonate with the capacitance of the speaker (perhaps an ESL). This is one argument for self-oscillating class D amps, as they can have enough feedback to minimize or prevent excitation of the resonance. I know of at least one class D amplifier that apparently will fail if run on an ESL for extended periods. I don't know if the amp went into oscillation and failed or if the HF resonance was causing it to overheat so until more is known, this bit remains anecdotal."

Just academic curiosity... not in any way trying to defend this dummy load as representing any speaker's equivalent circuit.

 

[BenjaminB]

I'm curious to whether warnings are given before this kind of timed exclusion? I mean it seems to me no one other really are doing these critical tests and seeks to reveal data not apparent in current test protocols. It would IMO be a shame to drive people like pma away, frustrations or not. -Of course the language used should be nicer than this last insult of his.

The communication between mods and pma is mainly between them.

Mods have here reacted to a statement of pma, mods decided to exclude pma from responding in this thread for a limited time. Very mild response from the mods - many forums are much more strict. Mods job is to keep some order at the forum, some kind of action must be taken when discussions get too heated (insults are unnecessary).

If pma really have an urgent need to respond he can easily create a // thread. He is in no way driven away from the forum.

I don't think anyone questions pma measurement, if I remember right also Amir has stated that.