Complex Load for Power Amplifier torture testing

[restorer-john]

Quad amplifiers always were deliberately bandwith limited, and the effect on square waves was easily visible.

Yes, not a pretty sight. Pretty much everything in the 1960s and 70s that came out of good old England was bandwidth limited. Darned Philips transistors.

Then came Japanese high speed silicon.

 

[Deleted member 65]

From a laymans perspective @pma amp test proposal seems to align with how Audiograph (PowerCube, LoadBox) is testing power amp stability.

”Our latest system lets you test all your new audio amplifier constructions, conventional or Class D. Do you think Class D is the solution for stability? You’d better check it before you go into production!”

 

[restorer-john]

Yes, so the idea that square waves should come out square and flat is quite misleading.

No, it isn't and you know it. With respect (and I have a lot of it for you), you are making excuses before the tests are done and presented because we all know what a Class D amplifier does to square waves.

Anyway, it's not the square wave depiction I care about and not really the focus of PMA's main post content, it's the behaviour (THD+N) into reactive phase angles at all frequencies and power levels.

 

[March Audio]

No, it isn't and you know it. With respect (and I have a lot of it for you), you are making excuses before the tests are done and presented because we all know what a Class D amplifier does to square waves.

Anyway, it's not the square wave depiction I care about and not really the focus of PMA's main post content, it's the behaviour (THD+N) into reactive phase angles at all frequencies and power levels.

Yes it is. These are not excuses, these are technical explanations of what can be observed. I have been specific about what I am referring to and not talked about load effects.

There are people that don't understand this, many readers don't have the technical knowledge, and I have frequently seen it misrepresented on forums as if a square wave is representative of quality. You do so yourself.

So John, do you think a square wave that shows Gibbs effect is an issue and indicative of poor audio quality?

Yes or no?

 

[March Audio]

A secondary question. What will you see on a scope connected to a dac playing a 13kHz square wave signal that was recorded at cd standard 44.1kHz sample rate?

 

[March Audio]

Yes, not a pretty sight. Pretty much everything in the 1960s and 70s that came out of good old England was bandwidth limited. Darned Philips transistors.

Then came Japanese high speed silicon.

When you say pretty sight, are you implying a relation to audio quality?

Dare I mention how digital audio is bandwidth limited?

 

[March Audio]

From a laymans perspective @pma amp test proposal seems to align with how Audiograph (PowerCube, LoadBox) is testing power amp stability.

”Our latest system lets you test all your new audio amplifier constructions, conventional or Class D. Do you think Class D is the solution for stability? You’d better check it before you go into production!”

Oh it is indeed relevant and any designer would be checking for peaking and stability issues.

 

[solderdude]

below the 10kHz squarewave of the 1ET400A

I see peaking but no Gibbs stuff. Mostly both in class-D and AB / A I only see either peaking or smoothing of squarewaves or oscillations and not the Gibbs stuff as amplifiers usually do not filter as steep.

 

[restorer-john]

Yes it is. These are not excuses, these are technical explanations of what can be observed. I have been specific about what I am referring to and not talked about load effects

The main gist of this thread is about difficult loads and distortion in amplifiers. Pavel likes square wave testing. So do I, but I primarily use an analogue signal generator with whatever LPF I want. When testing amplifiers, you don't want the source's bandwidth to be the limiting factor, you need to be able to control it.

You're quite happy to dismiss wide bandwidth testing altogether in one category (Class D amplifiers) and yet you are also selling a D/A converter (source) that will take 768KHz sampled PCM and DSD256... Can you see the irony?

So John, do you think a square wave that shows Gibbs effect is an issue and indicative of poor audio quality?

Yes or no?

So Alan, do you believe Purifi make the best amplifier in the world? Yes or No? See what a stupid question that is?

An amplifier (remember this is what we are talking about in this thread) that reproduces square waves within the audible bandwidth that have significantly different characteristics to the signal that was injected are, in my opinion, compromised designs. Compromises are made with Class D to obtain a heap of power in a lightweight and cheap to manufacture product. Some people are very happy with those compromises. For those who don't compromise, there's always Accuphase I guess.

Why are you using these pics?:

They are FIR style square waves. What comes out of your Class Ds looks more like an IIR 1st generation CD player signal with less ringing.

 

[Thomas savage]

You have kindly omitted the fact that capacitive reactive impedance adds complex imaginary impedance part and this may radically change stability of the amplifier. This is something you do not test and you seem to refuse to test it. You would see the result after addition of capacitance and applying square wave. I thought you were Electrical Engineer, were not you? Or you came from IT industry? That would explain it.

If your going to be putting measurements up like you have here ( very much appreciated) you will have to learn to take the subsequent questions and deal with differing opinions.

You broke under a complex load here and went unstable. Please make some design adjustments.

 

[restorer-john]

I see peaking but no Gibbs stuff.

The bandwidth of the AP isn't anything like a scope.

 

[solderdude]

The bandwidth of the AP isn't anything like a scope.

Is the gibbs stuff above 1MHz ?
The amp itself runs at 500kHz so expect the post filter to limit the frequency below that.

 

[JohnYang1997]

Care enough to consider complex load but have you measured speaker impedance over 100khz? Your model is simply not correctly emulating a speaker load.
If you want to test stability just use direct capacitor at output. 1nF 10nF 100nF.
If you want to measure distortion just use pure resistive load 8ohm 4ohm 2ohm.
Don't mix these together

 

[solderdude]

The proposed load is a challenging load for most amps.
Not a realistic load.

I think it would be a good thing to have some 'standard' loads in a certain complexity one can measure with.
Looking at reality all loads differ from those (unless ful range magnetostat ?) so even when testing with 1, 2 or even 4 different types of loads this still does not guarantee proper operation with all speakers.
But with measurements it would be a good thing to measure limits and to this is a standardized way.
One can dream up their own loads though.

 

[restorer-john]

If you want to test stability just use direct capacitor at output. 1nF 10nF 100nF.
If you want to measure distortion just use pure resistive load 8ohm 4ohm 2ohm.
Don't mix these together

Right. Tell that to the "real world" loads inside your speakers- you know the ones that have L, C and R.

 

[restorer-john]

Let's face it, resistive testing is easy, quick and gives some numbers. Pretty much the best numbers an amplifier will get. That's why they are used for marketing purposes. But tell me, who has a purely resistive speaker?

We aren't selling amplifiers (well March is). We are trying to characterize good, average and bad amplifiers and to be perfectly honest, it was done with much greater rigor and care in the distant past. Far too many valid and important tests have been chucked out the window in high fidelity and it does nothing for the reputation of the industry.

 

[Blumlein 88]

I think this simulation shows what you guys are talking about. Upper is 10 khz squarewave and lower is 1 khz squarewave. Done at 384 khz sample rate in Audacity. Both had an 18 db/octave filter down -3 db at 50 khz applied. If measured with a scope having real wide mhz bandwidth it would be even more like a slightly rounded squarewave.

 

[March Audio]

The main gist of this thread is about difficult loads and distortion in amplifiers. Pavel likes square wave testing. So do I, but I primarily use an analogue signal generator with whatever LPF I want. When testing amplifiers, you don't want the source's bandwidth to be the limiting factor, you need to be able to control it.

You're quite happy to dismiss wide bandwidth testing altogether in one category (Class D amplifiers) and yet you are also selling a D/A converter (source) that will take 768KHz sampled PCM and DSD256... Can you see the irony?



So Alan, do you believe Purifi make the best amplifier in the world? Yes or No? See what a stupid question that is?

An amplifier (remember this is what we are talking about in this thread) that reproduces square waves within the audible bandwidth that have significantly different characteristics to the signal that was injected are, in my opinion, compromised designs. Compromises are made with Class D to obtain a heap of power in a lightweight and cheap to manufacture product. Some people are very happy with those compromises. For those who don't compromise, there's always Accuphase I guess.

Why are you using these pics?:

View attachment 42026

They are FIR style square waves. What comes out of your Class Ds looks more like an IIR 1st generation CD player signal with less ringing.

Yes, whats your point?

Square waves are useful for looking for stability granted, however if you bandwidth limit, be it due to the amps inherent frequency response, or if you filter to remove switching frequency with class d, then you will inevitably remove the harmonics which make a square wave square. You will also suffer Gibbs effect.

There is no irony because I make no claim about the usefulness of high bandwidth sources. I have repeatedly stated, as you fully well know, that I believe high bandwidth recordings to not only be of no use, but a potential liability.

The Purifi amp is extremely good, both technically and subjectively. I have no idea if its the best amp in the world. Its a rather redundant question as its not possible to answer without measuring and hearing all amplifiers.

John, with respect, I dont think you have grasped what constitutes a square wave. It is a fundamental frequency and an infinite number of odd harmonics. Its shape is therefore fundamentally defined by system bandwidth.

Why do you think I asked you the question about the 13kHz square wave being played by a dac from a 44kHz recording?

So you are going to have to justify your statement.

"An amplifier (remember this is what we are talking about in this thread) that reproduces square waves within the audible bandwidth that have significantly different characteristics to the signal that was injected are, in my opinion, compromised designs"

So referencing the above, how about a 10kHz square wave? How will that appear after being recorded at a 44kHz sample rate (22 kHz bandwidth)?
It will be a sine wave because none of the odd harmonics will have been captured. The first odd harmonic will be at 30kHz. See below.

So by definition you must think that the digital audio we have been using for the past 30 years is "compromised".

By your standards it is far more compromised than a far wider bandwidth class d amp and yet not a murmer from you about that. You are simply contradicting yourself by accepting CD and criticising class d.

I am using those pictures to demonstrate what happens when you bandwidth limit a square wave. Its nothing to do with FIR or IIR or digital audio.

 

[March Audio]

Let's face it, resistive testing is easy, quick and gives some numbers. Pretty much the best numbers an amplifier will get. That's why they are used for marketing purposes. But tell me, who has a purely resistive speaker?

We aren't selling amplifiers (well March is). We are trying to characterize good, average and bad amplifiers and to be perfectly honest, it was done with much greater rigor and care in the distant past. Far too many valid and important tests have been chucked out the window in high fidelity and it does nothing for the reputation of the industry.

Problem is you havent demonstrated why some of these things are "important" or justify why it denotes good, bad or otherwise.

For example, looking like a perfect square wave clearly isnt important for audio quality because bandwidth limited CD would sound crap as a result. It simply doesnt.

So you do a high power test at 20 kHz (which will never happen in real life) and harmonic distortion appears at 40, 60khz etc. Well you cant actually hear those harmonics and they are not reproduced by your speakers, so what is the detriment to sound quality? Thats precisely why you see Hypex test with 19+20Khz IMD to see what happens below 20 kHz which you can hear.

You are just holding on to these "old standards" without applying critical thought.

When was the last time you had an amp go unstable?

Pavel wants to test with a 0.5 ohm load which is just not representative of any quality design of speaker.

 

[rha61]

From a laymans perspective @pma amp test proposal seems to align with how Audiograph (PowerCube, LoadBox) is testing power amp stability.

”Our latest system lets you test all your new audio amplifier constructions, conventional or Class D. Do you think Class D is the solution for stability? You’d better check it before you go into production!”

Agree
that was done in the excellent The Audio Critic