• 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!

Handling a complex load by power amplifiers

I agree with that sentiment and we are in good company.
It would complicate testing, make much more measurements, much more reporting and would require repeating tests under various loads, maybe even in circuit measurements and what loads should be used. That would have to be documented and agreed upon (standardized even?)
Given the amount of reviews churned out by Amir I wonder if that would be possible for all amps.

I would like to see some measurements added but here too... it takes time and reporting.
 
I agree with that sentiment and we are in good company.
It would complicate testing, make much more measurements, much more reporting and would require repeating tests under various loads, maybe even in circuit measurements and what loads should be used. That would have to be documented and agreed upon (standardized even?)
Given the amount of reviews churned out by Amir I wonder if that would be possible for all amps.

I would like to see some measurements added but here too... it takes time and reporting.
I aknowledge the challenges inherent. And I also think Amir does a spectacular job already despite what others want to twist my words into.
When that is said, I do believe that we should have more in focus what the current tests have of limitations and doesn't show. -That there is more to it than a glance at a 4R ohm sweep tells.
And I do think that there should be room to discuss this in orderly manner. It does feel like the execution squad is ready whenever someone dares to question..

I still hope that we eventually get tests with equivalent loads. I think it would end a lot of discussions.
 
Amplifier design and testing is about facts, not about emotionally coloured opinions. The amplifier circuit that has a LC undamped resonant circuit at the output with resonance not much above the audio band and that relies only on damping from a loudspeaker should never put on the market and should never be recommended to potential customers. The risk hazard for the speaker and amplifier circuit itself is real.

In the TPA32XX based amplifier with LC filter tuned at 50kHz with wave impedance of 6.32 ohm (a compromise between frequency response and damping), and it is the case of all similar designs, not only AIYIMA A07 that was a DUT here, the resonance with a typical speaker with inductive impedance at higher frequencies occurs even with a typical band limited music programme any time the amplifier goes to clipping. See the example below.

A07_wide_dummyload_amywinehouse-clipping_sm.png


The resonance is poorly damped and oscillating response lasts quite a long time.
The amplitude of resonant oscillations depends how deep is the amplifier sent into clipping and also on the steepness of the signal below clipping. The amplitude of oscillations may become huge

A07_wide_dummyload_10k-clipping_sm.png


If we push the amplifier even in the deeper clipping, the peak output voltage from A07 exceeds 100V. The amplifier losses any control above the signal, the current that flows through the output L, C and switching transistors is huge, the amplifier collapses and shuts down itself.
 
As for recommendations or lack thereof @amirm gives, he can't win, whatever he does. If he considers the price, the scales change. But when the price is astronomical or ridiculously cheap, that has to weigh in to the recommendation too. Fractions of a dB separate many devices and we all know how dubious rankings become in that situation. He's between a rock and a hard place and does his best. :)
Well put!
 
I'm perhaps in danger of repeating myself. As far as I'm aware, there is only limited information about real-world speakers' ultrasonic impedance characteristics. We can extrapolate from data below 20kHz, and it's probably save to assume that manufactures are not normally adding poles or zeros above 20kHz, but we are all forced into working with an assumptions that most real world speakers represent a benign load to amplifiers which depend on ultrasonic oscillations for their operation.
 
I'm perhaps in danger of repeating myself. As far as I'm aware, there is only limited information about real-world speakers' ultrasonic impedance characteristics. We can extrapolate from data below 20kHz, and it's probably save to assume that manufactures are not normally adding poles or zeros above 20kHz, but we are all forced into working with an assumptions that most real world speakers represent a benign load to amplifiers which depend on ultrasonic oscillations for their operation.
I don't know. I may be old school but doesn't some speakers show large phase degrees and low impedance in lower frequencies as well?
The Iron Law dictates something like that.
 
The PowerCube presents a complex load to the amplifier. The phase angle is calibrated at 1 kHz but you could use it at other frequencies (guessing most testers just use 1 kHz). It allows you to sweep resistance and reactance (phase angle) and thus impedance. Much like frequency sweeps that exercise a single frequency at a time to produce a frequency response plot, the PowerCube allows you to plot the response over a complex (resistance, inductance, capacitance, RLC) impedance holding frequency constant and sweeping impedance. You can see how the amplifier responds to a complex (in the mathematical/engineering sense of resistance and reactance) load.


It'd be interesting to run the test at say 20 kHz (at lower power level) where feedback (loop gain) is reduced. The counter argument is you can achieve the same effective results by sweeping to larger reactances at lower frequencies and that is what the PowerCube is designed to do.

FWIWFM - Don
 
Last edited:
Amplifier design and testing is about facts, not about emotionally coloured opinions.

Agree. But as usual, you tend to discount important factors as allegedly "emotional opinions" and and rather blind to the role of your own "emotional opinions" in driving what you focus on, what you don't focus on, and how you communicate about it.
 
Agree. But as usual, you tend to discount important factors as allegedly "emotional opinions" and and rather blind to the role of your own "emotional opinions" in driving what you focus on, what you don't focus on, and how you communicate about it.
And as usual you go fishing without contributing.
As you were.
 
The resonance is poorly damped and oscillating response lasts quite a long time.
The amplitude of resonant oscillations depends how deep is the amplifier sent into clipping and also on the steepness of the signal below clipping. The amplitude of oscillations may become huge

If we push the amplifier even in the deeper clipping, the peak output voltage from A07 exceeds 100V. The amplifier losses any control above the signal, the current that flows through the output L, C and switching transistors is huge, the amplifier collapses and shuts down itself.
I’m a EE and I understand what you are saying in theory. But does it happen in reality when using a garden variety speaker that will be used with an $80 amplifier fed with a DAC, or a media player.

If you are confident that your theory will hold how about devising a test and show us those huge amplitude oscillations and the speaker receiving in excess of 100V peak?

If you don’t want to buy the amplifier, I will buy and drop ship to you.
 
I’m content to let others read our respective comments and come to their own conclusion.
On a serious note though, in whitch way are you contributing other than shallow critique of more knowledgeable people?

Please quote.
 
On a serious note though, in whitch way are you contributing other than shallow critique of more knowledgeable people?

Please quote.

I’ll be happy to refer you to several comments, as soon as you supply examples of Amir “praising bad equipment.” :)
 
I’ll be happy to refer you to several comments, as soon as you supply examples of Amir “praising bad equipment.” :)
I've stated no such thing ever. Please provide evidence of your claims.
In the mean time please list three of the key concrete things that makes PMA's tests invalid or irrelevant and emotional. Thank you.
 
Note the parts in the red ellipses. Fit them and the resonance peak goes away. Specifically the 3R3 + 1uF.

View attachment 280544

Alternatively include a note that the Amplifier should not be used with speakers, unless they are purely resistive in nature in the HF fegion and the amplifier ideally should not be used with speakers but only dummy load resistors.

Thor
I am guessing that you have gotten yourself a thread ban for your comments. I'd have to say your argument was utterly unconvincing when the TI evaluation board has DNP (do not populate) for the components which you claimed (without proof) were so critical. So blame TI when they don't even bother to install them in their own eval/demo boards.

Below are the simulated responses after adding the Zobel. The resonance peak was lowered, but didn't go away. But it shifted the resonance frequency lower and made the audible band (20 - 20k Hz) frequency response worse. So, which is the better engineering compromise? The one the Aiyima chose, or the one you claimed should make this amp totally useless to drive speakers when without? I'd go with the Aiyima one as the better choice. And so the Aiyima folks are, at least in this case, more competent than you as engineers.

Also shown are the responses using a different load (dashed curves) to make the amp look a lot worse. (The load is a 3 Ω resistor in series with a 2 μF capacitor, similar to a Zobel, and never goes below 3 Ω.) It is to show that it will to be very difficult to find the "universal complex load (or loads)" for amplifier testing, as different amps will probably has their worst behaviors with different complex loads.

zobel_response.png
 
I've stated no such thing ever. Please provide evidence of your claims.
In the mean time please list three of the key concrete things that makes PMA's tests invalid or irrelevant and emotional. Thank you.

Sigh. Okay, let's try one more time. Here's where you talked about the praising of bad equipment:

Yeah, or stop praising bad stuff because it's cheap.
Either the reviews should be 100 % objective and the price just stated at the time of review. And the reviewer should not be recommending. -How could he anyway, as he does not know either budget or usecase for the individual??

You clearly argue that "the reviewer" should "stop praising bad stuff because it's cheap," and that "the reviewer should not be recommending."

If you really want to, you can try to claim that you did not mean Amir when you said that - but given that you posted this in a thread and discussion that's questioning Amir's use of simple loads (or in the case of the Powercube, complex loads but not the right kind of complex loads), and given that Amir quite notably uses "recommended" and "not recommended" as his main review conclusions, it is obvious to any reasonable reader that you are referring to Amir here. Perhaps you were not referring only to Amir, but it is not plausible to try to claim you were not referring to Amir.

Hopefully we can put that particular point to bed, because in my view it's not the main point anyway. The main point is that you claim that "bad stuff" is being "praised because it's cheap." The reason I asked you to provide examples or evidence of that being done in reviews here was not to play "gotcha." Rather, it was for precisely the reason I originally stated: to allow others to gauge the validity of your claim. If you were to provide one or more examples of reviews that led you to make that claim, then other members here could look at those reviews and think about the following key issues:
  • What measurements, performance, build characteristics, or other aspects make it bad?
  • Do they agree with you that the gear is in fact bad?
  • Was the gear praised because of its price, or because of its performance regardless of price, or for some combination thereof?
  • Is other, similar gear that performs better available at similar prices? Conversely, does the gear in question perform as well as other gear costing a good deal more?
By providing one or more examples, you allow people to learn about the criteria you are using to make your claim, and therefore you give them the ability to make an informed decision about whether or not they agree with you. In the absence of that, we're left only with your repeated complaint that cheap, "throwaway," Chinese gear "rapidly shoved into aluminum enclosures" bothers you and "grinds your gears." That's a lot of heat and very little light.

As for @pma 's posts, I would first say that your request that I provide 3 examples is made in bad faith, simply to be disagreeable and to try to "top" my request that you provide a single example to back up your claim. But with that said, sure, no problem. Here are a few:

1. pma has far more engineering expertise than I do, and he absolutely devises specific tests and shares clearly documented results here. Good on him for that, 100%. However, he also has demonstrated a clear pattern of ignoring key factors that are important to take into account. For example, he's run unfiltered 20kHz square waves into Class D amps and on that basis claimed that they are fundamentally flawed or that "Class D has a long way to go" (note the sweeping claim that it's the whole topology, not just this or that amp or implementation).

2. Class D has a switching frequency by design, and of course you can exploit that by running certain kinds of signals into such an amp. He has steadfastly ignored or dodged the questions of (a) whether such signals represent what a musical source will actually send to an amp, and (b) whether it is reasonable or legitimate to design an amp to perform at its best with typical musical signals. By ignoring or dodging these questions, he often claims that basic characteristics of Class D topology are not characteristics but rather flaws. It is well-known - and pma himself has noted many times in other comments - that the best Class D designs use filters to control most of these issues, are load-invariant, and are capable of driving 2-ohm loads (and in some cases sub-2 ohm loads) without issue. But he does not generally acknowledge these facts when he writes about his amp torture tests.

3. He has also repeatedly asserted that a "20 year old design" - the Hypex UcD180 module - is superior to today's NCore and Purifi designs because the UcD's distortion levels are more consistent across all audible frequencies, while the NCore and Purifi's distortion levels start to rise from around 5kHz upwards. What he ignores when he makes this argument is that the NCore and Purifi designs have lower distortion at all frequencies, even at their worst-performing frequencies, than the UcD has at any frequency, even its best-performing frequency. So in this instance, pma's attachment to consistency of distortion, even if that distortion is always higher, is in my view an excellent example of a subjective or aesthetic preference, which has nothing to do with the actual comparative performance or quality of the amps in question.

Personally I don't like calling other members "emotional" - I think it's dismissive. But by pma's use of the term, I would say his preference for the UcD based on how pretty its distortion curve looks is precisely an example of what he refers to as "emotional." For goodness' sake, one could simply reduce the feedback in an NCore or Purifi design and the distortion would rise at sub-5kHz frequencies - that would degrade the overall measured distortion performance of those modules, but by pma's standard it would improve those modules' performance because it would make their distortion more consistent across frequencies - and that distortion would still be lower than that of his beloved UcD180.

So there are three examples.

But let me be clear - my issue is not with pma running these tests. My issue is with the conclusions he draws from them and how he writes about the issues. In my opinion, his posts create confusion and spread FUD. I'm not saying that's his intention; but it does seem to have that effect. In this thread, a member has posted that they are feeling that Class D is not a great choice for higher frequencies (I forget the exact words but it's easily findable above). IMHO that represents a step backwards in our collective clarity and understanding about Class D amps, particularly the many implementations that are load-invariant. And IMHO this backwards step, this increase in confusion and overarching fear of using Class D, is a direct result of pma's posts.
 
I don't know. I may be old school but doesn't some speakers show large phase degrees and low impedance in lower frequencies as well?
The Iron Law dictates something like that.
That's certainly often the case. The reason I raised ultrasonics is that they may interact with the switcher in class D amps, but we don't have enough data about speaker ultrasonic impedance behaviour to know.

Meanwhile, all classes of amp have to deal with 0 to 20Hz. There's an argument that SMPS can recover from low frequency impulses better than classic baseband transformer solutions.
 
Sigh. Okay, let's try one more time. Here's where you talked about the praising of bad equipment:



You clearly argue that "the reviewer" should "stop praising bad stuff because it's cheap," and that "the reviewer should not be recommending."

If you really want to, you can try to claim that you did not mean Amir when you said that - but given that you posted this in a thread and discussion that's questioning Amir's use of simple loads (or in the case of the Powercube, complex loads but not the right kind of complex loads), and given that Amir quite notably uses "recommended" and "not recommended" as his main review conclusions, it is obvious to any reasonable reader that you are referring to Amir here. Perhaps you were not referring only to Amir, but it is not plausible to try to claim you were not referring to Amir.

Hopefully we can put that particular point to bed, because in my view it's not the main point anyway. The main point is that you claim that "bad stuff" is being "praised because it's cheap." The reason I asked you to provide examples or evidence of that being done in reviews here was not to play "gotcha." Rather, it was for precisely the reason I originally stated: to allow others to gauge the validity of your claim. If you were to provide one or more examples of reviews that led you to make that claim, then other members here could look at those reviews and think about the following key issues:
  • What measurements, performance, build characteristics, or other aspects make it bad?
  • Do they agree with you that the gear is in fact bad?
  • Was the gear praised because of its price, or because of its performance regardless of price, or for some combination thereof?
  • Is other, similar gear that performs better available at similar prices? Conversely, does the gear in question perform as well as other gear costing a good deal more?
By providing one or more examples, you allow people to learn about the criteria you are using to make your claim, and therefore you give them the ability to make an informed decision about whether or not they agree with you. In the absence of that, we're left only with your repeated complaint that cheap, "throwaway," Chinese gear "rapidly shoved into aluminum enclosures" bothers you and "grinds your gears." That's a lot of heat and very little light.

As for @pma 's posts, I would first say that your request that I provide 3 examples is made in bad faith, simply to be disagreeable and to try to "top" my request that you provide a single example to back up your claim. But with that said, sure, no problem. Here are a few:

1. pma has far more engineering expertise than I do, and he absolutely devises specific tests and shares clearly documented results here. Good on him for that, 100%. However, he also has demonstrated a clear pattern of ignoring key factors that are important to take into account. For example, he's run unfiltered 20kHz square waves into Class D amps and on that basis claimed that they are fundamentally flawed or that "Class D has a long way to go" (note the sweeping claim that it's the whole topology, not just this or that amp or implementation).

2. Class D has a switching frequency by design, and of course you can exploit that by running certain kinds of signals into such an amp. He has steadfastly ignored or dodged the questions of (a) whether such signals represent what a musical source will actually send to an amp, and (b) whether it is reasonable or legitimate to design an amp to perform at its best with typical musical signals. By ignoring or dodging these questions, he often claims that basic characteristics of Class D topology are not characteristics but rather flaws. It is well-known - and pma himself has noted many times in other comments - that the best Class D designs use filters to control most of these issues, are load-invariant, and are capable of driving 2-ohm loads (and in some cases sub-2 ohm loads) without issue. But he does not generally acknowledge these facts when he writes about his amp torture tests.

3. He has also repeatedly asserted that a "20 year old design" - the Hypex UcD180 module - is superior to today's NCore and Purifi designs because the UcD's distortion levels are more consistent across all audible frequencies, while the NCore and Purifi's distortion levels start to rise from around 5kHz upwards. What he ignores when he makes this argument is that the NCore and Purifi designs have lower distortion at all frequencies, even at their worst-performing frequencies, than the UcD has at any frequency, even its best-performing frequency. So in this instance, pma's attachment to consistency of distortion, even if that distortion is always higher, is in my view an excellent example of a subjective or aesthetic preference, which has nothing to do with the actual comparative performance or quality of the amps in question.

Personally I don't like calling other members "emotional" - I think it's dismissive. But by pma's use of the term, I would say his preference for the UcD based on how pretty its distortion curve looks is precisely an example of what he refers to as "emotional." For goodness' sake, one could simply reduce the feedback in an NCore or Purifi design and the distortion would rise at sub-5kHz frequencies - that would degrade the overall measured distortion performance of those modules, but by pma's standard it would improve those modules' performance because it would make their distortion more consistent across frequencies - and that distortion would still be lower than that of his beloved UcD180.

So there are three examples.

But let me be clear - my issue is not with pma running these tests. My issue is with the conclusions he draws from them and how he writes about the issues. In my opinion, his posts create confusion and spread FUD. I'm not saying that's his intention; but it does seem to have that effect. In this thread, a member has posted that they are feeling that Class D is not a great choice for higher frequencies (I forget the exact words but it's easily findable above). IMHO that represents a step backwards in our collective clarity and understanding about Class D amps, particularly the many implementations that are load-invariant. And IMHO this backwards step, this increase in confusion and overarching fear of using Class D, is a direct result of pma's posts.
I didn't mean Amir so that whole argument is wasted energy from your side. Sincerely apologize.

Reg. the three points..I disagree. Please elaborate.
You seem confused about the point PMA is trying to make. Maybe you should make an effort in understanding the theory.. I shall refrain from making a gnarly remark about a high hat at this point.

We should be thanking a user for showing stuff that has directly audible consequences in some cases and isn't shown in regular testing. Not crucify him. That's all backwards. -Meaning persons like yourself has no interest in revealing anything new. Imagine science in general with that state of mind. Scary..
 
Last edited by a moderator:
That's certainly often the case. The reason I raised ultrasonics is that they may interact with the switcher in class D amps, but we don't have enough data about speaker ultrasonic impedance behaviour to know.

Meanwhile, all classes of amp have to deal with 0 to 20Hz. There's an argument that SMPS can recover from low frequency impulses better than classic baseband transformer solutions.
Okay.
That last point reg. SMPS. -Can you point me to where more info about this can be had?
Is it because of lower response time, higher switching speed?
 
I didn't mean Amir so that whole argument is wasted energy from your side. Sincerely apologize.

Reg. the three points..I disagree. Please elaborate.
You seem confused about the point PMA is trying to make. Maybe you should make an effort in understanding the theory.. I shall refrain from making a gnarly remark about a high hat at this point.

We should be thanking a user for showing stuff that has directly audible consequences in some cases and isn't shown in regular testing. Not crucify him. That's all backwards. -Meaning persons like yourself has no interest in revealing anything new. Imagine science in general with that state of mind. Scary..

Sorry, no sale. You don’t get to ask someone to elaborate when you yourself are unwilling (or perhaps unable, I don’t know) to do so. If you think I’m misunderstanding pma’s posts then you can explain what exactly you think I’m misunderstanding, at a level of detail similar to the detail I provided in response to your last request/demand. As for whether you were referring to Amir, I’ll leave it to others to assess for themselves what a plausible reading of your statement is, and I’ll reiterate that the main point was not Amir but rather the claim that “bad gear” is being “praised because it’s cheap” - a claim for which you’ve still provided no examples or evidence.

Finally, I would like it if we could find a way to wind down this particular exchange. It’s about to start going in circles (if it hasn’t already), and if @AdamG247 were to tell us to knock it off already, imho he’d be well within his rights.
 
Back
Top Bottom