Why can't a Sine Wave be used for testing high-power Class D amps?

[fpitas]

That's an old post, from the time I was an admin/head moderator at that forum. Our policy was to not censor posts that were technically stupid, but rather to allow smarter members to correct things in public.

Understood. But the floods of goofiness there got to me.

 

[bibio]

I'm starting to regret asking......
Does none of the info on the DIY-audio thread ring true?

 

[fpitas]

I'm starting to regret asking......
Does none of the info on the DIY-audio thread ring true?

There's no telling from here. You can set DSP to monitor average power and limit intelligently when it gets too great. Perhaps someone had that setting on a low power, or it was a default.

 

[DVDdoug]

The sinewave is reproduced by the amp but as the level is increased the amplifier heavily limits the signal. You don't get anywhere near the potential full output of the amp.

Two things -

1. MOST manufacturers "fudge" the specs. (This not just a "class-D" issue".) Crown is a "reputable" manufacturer but when Amir tested the "525W" Crown XLS-1502 it clipped at 336W (64% of it's rated power) and it got worse when the amp was pushed continuously and it heated-up". Many other manufacturer's are MORE dishonest. I've got a car amplifier rated for 5000W and I'd be surprised if it can put out 1000. (It goes "loud enough" with a pair of 12-inch subwoofers but I've never tried measuring it... And 5000W real-watts could fry the speakers.)

There is a Federal Regulation but it seems to be no longer enforced.

2. Many (most) amplifiers can put-out higher power on short-term peaks then they can put-out continuously. That's mostly OK because music and other normal program material has peaks about 10dB higher than the average. Of course it varies and modern more-compressed music has a lower peak-to-average level. There is no standard for "peak power", or maybe there is one but you can't count-on manufacturers and I'd ignore the manufacturer's peak-power claims.

The continuous power rating is "worst case" and "conservative". If the amp can put-out 500W continuously you can know it can put-out 500W on the peaks.

So it's not "sine waves" that "stress" the amplifier, but the fact that the testing is done with continuous waves (where the peak and average is the same).



BTW - Speaker ratings are tricky and often dishonest too. Speakers are NOT rated for continuous power (even when honestly following the standards), and the tweeter can't take as much power as the woofer.

 

[solderdude]

I guess it depends on the class-D circuity and above all the (internal) error detection.
It is easy to just lower the input when nearing the clipping level when an error detection flag in a circuit is active.
Some circuits want to ensure that when nearing the clipping limit the OP devices keep on switching and do not 'rail stick' during overload.

Also the power supplies used for speaker amps (with dynamic loads) are nothing like the average SMPS which are designed for constant or not much varying loads.
So short term 4kW in an amp (music or burst signal) is possible but continuous might not be a good idea with certain amps (and their AC/DC converters)

 

[voodooless]

I'm starting to regret asking......
Does none of the info on the DIY-audio thread ring true?

Not really…

… so they’ll us, what is your source regarding sine testing of the Labgruppen and Crown amps?

 

[kemmler3D]

If an amp is "clamping down" on sine waves there's probably a "feeback killer" function running in the DSP. This would be not-rare in the PA world, basically nonexistent in the home amp world.

If so, It has nothing to do with Class D or otherwise.

Other than that, I think @DVDdoug 's comment makes a ton of sense. Power specs are often lies or half-truths. A sine wave is a high RMS signal compared to music... you need to be sure you've got an honest quote for the continuous power rating (not just "max power" or "watts") before you conclude the amp isn't meeting spec... which it may not.

 

[ChrisG]

Hello,
As the title suggested, I don't understand why a sine wave can't be used to test modern, high-power, class d amplifiers.
Some of the large PA amps I've recently looked at have dsp build-in which appear to clamp down any waveform that looks like a continuous sine wave.
Would someone be able to explain the reasoning behind this? Is it the Class D amp or the SMPS that is the issue?
Considering the waveforms of some modern music, does this not affect the audio quality? Some synth music is very sinewave based. How does the DSP differentiate?
So if you can't use a sine wave, how could I perform a frequency sweep of an amplifier? I have one particular amp that has built-in dsp and speaker presets, I'd like to see the crossover point and such.
Thanks in advance for your help.

In times gone by, you could connect an amplifier to a resistive load bank, run up a sine wave with a frequency of your choosing, measure the output voltage while watching a 'scope for clipping, and work out the power output.

The amplifiers that did well under this test regime were often large and heavy, because they were designed to pass this sort of testing.


With the demands of:
- More power
- Lighter weight
- More efficiency
- Smaller size (ie, fitting more amps in a chassis or rack)

from the live sound world, some changes started to appear.


All electrical signals have a crest factor, typically measured in dB. This denotes the difference between the peak and average power level in a given signal.

A square wave has a crest factor of 0dB, since the signal spends all of its time at +/-100%.
A sine wave has a crest factor of 3dB.
Pink noise is 12dB
Music can be anything from 40dB to 6dB - the latter being for heavily-compressed program material.


In most cases, it makes very little sense to optimise an amplifier to be able to hit full power, continuously, for hours at a time: that does not happen with real music.
Instead, many manufacturers choose to produce amplifiers which will put out large voltage swings for a few cycles (which is often all that's necessary), but sacrifice the long-duration sine tests to get there.

I have a rack of Powersoft T-series amplifiers. The T604, in some ways, is disappointing: according to Production Partner's rather extensive testing, the 6KW-(total)-rated 4-channel amplifier will only produce 4x 380w into 4ohm, all channels driven.
That's with a 1kHz sine wave, after a minute of testing.

However, that same amplifier will also put out 4x 2634w unclipped peaks, when driven with pink noise.

It's worth noting that, after considering crest factors etc, that the sine wave would actually present more heating power to a resistive load: 4x 380w RMS, vs 4x 324w RMS.

Other tests yield results that are in-between: https://www-production--partner-de...._sl=de&_x_tr_tl=en&_x_tr_hl=en&_x_tr_pto=wapp


Now, it is possible to make a class D amplifier that will produce large amounts of sine-wave-power. It's expensive, and almost always unnecessary. I'd much rather have an amplifier capable of 500w average and 5kw peaks, instead of an amplifier that'll put out 1kw sine waves for hours at a time.


It's also worth briefly noting that some genres of music will often present relatively low crest factor signals, particularly in the bass, which can be troublesome for these "squishy" amps. The operators that run those PA systems will often choose the old-school amplifiers, like the Crown MA5002VZ.

Finally, to respond to DVDdoug's comments about speaker ratings: there are standards in place for this stuff, but you have to be careful: to use a tweeter-example, a 100w test signal with a 3.5kHz 12dB/octave highpass filter does not mean your tweeter can take 100w. It means it can only take the highpassed portion of that signal, which is probably about 5w.

Chris

 

[bibio]

In times gone by, you could connect an amplifier to a resistive load bank, run up a sine wave with a frequency of your choosing, measure the output voltage while watching a 'scope for clipping, and work out the power output.

The amplifiers that did well under this test regime were often large and heavy, because they were designed to pass this sort of testing.


With the demands of:
- More power
- Lighter weight
- More efficiency
- Smaller size (ie, fitting more amps in a chassis or rack)

from the live sound world, some changes started to appear.


All electrical signals have a crest factor, typically measured in dB. This denotes the difference between the peak and average power level in a given signal.

A square wave has a crest factor of 0dB, since the signal spends all of its time at +/-100%.
A sine wave has a crest factor of 3dB.
Pink noise is 12dB
Music can be anything from 40dB to 6dB - the latter being for heavily-compressed program material.


In most cases, it makes very little sense to optimise an amplifier to be able to hit full power, continuously, for hours at a time: that does not happen with real music.
Instead, many manufacturers choose to produce amplifiers which will put out large voltage swings for a few cycles (which is often all that's necessary), but sacrifice the long-duration sine tests to get there.

I have a rack of Powersoft T-series amplifiers. The T604, in some ways, is disappointing: according to Production Partner's rather extensive testing, the 6KW-(total)-rated 4-channel amplifier will only produce 4x 380w into 4ohm, all channels driven.
That's with a 1kHz sine wave, after a minute of testing.

However, that same amplifier will also put out 4x 2634w unclipped peaks, when driven with pink noise.

It's worth noting that, after considering crest factors etc, that the sine wave would actually present more heating power to a resistive load: 4x 380w RMS, vs 4x 324w RMS.

Other tests yield results that are in-between: https://www-production--partner-de...._sl=de&_x_tr_tl=en&_x_tr_hl=en&_x_tr_pto=wapp


Now, it is possible to make a class D amplifier that will produce large amounts of sine-wave-power. It's expensive, and almost always unnecessary. I'd much rather have an amplifier capable of 500w average and 5kw peaks, instead of an amplifier that'll put out 1kw sine waves for hours at a time.


It's also worth briefly noting that some genres of music will often present relatively low crest factor signals, particularly in the bass, which can be troublesome for these "squishy" amps. The operators that run those PA systems will often choose the old-school amplifiers, like the Crown MA5002VZ.

Finally, to respond to DVDdoug's comments about speaker ratings: there are standards in place for this stuff, but you have to be careful: to use a tweeter-example, a 100w test signal with a 3.5kHz 12dB/octave highpass filter does not mean your tweeter can take 100w. It means it can only take the highpassed portion of that signal, which is probably about 5w.

Chris

Thanks very much for the explanation ChrisG.

 

[Steve81]

My apologies to @fpitas, I mistook him for the OP, and thought he was not understanding a basic explanation of how limiters work, and calling me a troll. Mea culpa.

 

[radix]

Hi, amplifers like the lab gruppen PLM's:
https://www.labgruppen.com/product.html?modelCode=P0F66
Or the Crown I-tech range:
https://www.crownaudio.com/en/product_families/i-tech-hd-series

Try sticking a sine wave into an amp like this, it's not happy. I'm just trying to understand the technical reason why.

That crown has a lot of DSP features, including a cross over and LevelMax. If you have not turned off those things, it could account for the odd results you see. I'm not an expert on that line, just gone through the manual before.

Marc