Measurement of power handling and dynamic distortion of speakers with sine bursts

[pma]

Speakers are usually able to accept quite high impulse power and measuring of speaker distortion by sine bursts would show if we are still in relatively linear region. The pain of sine burst distortion measurement is in length of the burst, which may not be long enough to get sufficient resolution for spectrum analysis. I tried to make a compromise – test signal is a sine burst at 100Hz frequency, 20 periods, amplitude shaped by Blackmann window. The first results look promising.

The test burst had voltage amplitude of 30.3Vpeak, that means 21.4Vrms of equivalent sine wave. This would be 76W at 6ohm load impedance. The woofer measured was 6.5” Seas W18NX001 and 21.4Vrms of steady state 100Hz sine would be impossible to apply. Speaker input voltage and acoustical output were recorded to left and right channel of the wav file. Distortion was estimated by post-processing of the recorded file. The result is interesting – voltage distortion at speaker terminals is about 0.02% and distortion on acoustical side is about 2%, which is much less than I was expecting from measurements with steady state sines.

Measured burst shapes are attached as well as distortion measured.

Burst voltage at speaker terminals

Burst measured at acoustical side

Distortion at electrical and acoustical side

Distortion plots with levels equalized. Blue - electrical distortion, red - acoustical distortion. Electrical recording was lower in level, it had to be amplified of 11dB to equalize levels, which can be seen in higher noise floor above 3kHz.

 

[miero]

Is a distortion of the AD path smaller than the distortion of the loudspeaker?

 

[pma]

Is a distortion of the AD path smaller than the distortion of the loudspeaker?

Definitely.

 

[LTig]

And now the same with 1 kW please.

 

[pma]

And now the same with 1 kW please.

I need some time to prepare bridged amp with enough output swing. This will take some time, I guess by the end of the next week, as I would be too busy till Thursday. However, I would not drive the speaker to destruction.

 

[Rincewind]

What you can do is drive the loudspeaker until you hit 3% distortion, then stop. Risk of damage increases dramatically as you go beyond that. This is how those S&R graphs work. Only for larger drivers at low frequencies is a 10% limit used and it would be horrible to listen to 10% distortion at 1 kHz anyway so there is no point in measuring higher distortion at higher frequency.

 

[LTig]

I need some time to prepare bridged amp with enough output swing. This will take some time, I guess by the end of the next week, as I would be too busy till Thursday. However, I would not drive the speaker to destruction.

I know. Neither would I. It was a joke targeted at @restorer-john .

 

[pma]

I know. Neither would I. It was a joke targeted at @restorer-john .

I was quicker than expected. I prepared a bridged amp yesterday with output capability of 70Vpeak. Made tests this morning with Seas W18NX001 and stopped at 54Vp (38Vrms, 243W), as I did not like the waveform coming from speaker and did not want to destroy it. Factory specs says 250W short term power handling and this seems to be the real usable limit. Below are the plots with a burst which has at 54Vp at peak. Measured at 10cm distance. Yes I checked that microphone is not saturated and distorted going to 1m - it only added noise and messed the result. Distortion was not decreased from larger distance, in fact it was higher due to reflections.

Speaker - acoustical distortion

Blue - speaker terminals, Red - acoustical

Time record

 

[pma]

Amp distortion into 8ohm resistive load with the same burst (54Vp). Please note that higher electrical distortion in the previous post is due to 2x6m speaker cable resistance + nonlinear speaker load, which makes a voltage divider with cable resistance and amp output impedance.

So, what prevents @amirm from doing impulse burst measurements of speakers?

 

[RayDunzl]

So, what prevents @amirm from doing impulse burst measurements of speakers?

Mark Twain on "Work versus Play"
August 30, 2008 at 17:52 Tags Quotes

There's a lovely quote by Mark Twain in "The adventures of Tom Sawyer" about the difference between working and playing:

Tom said to himself that it was not such a hollow world, after all. He had discovered a great law of human action, without knowing it--namely, that in order to make a man or a boy covet a thing, it is only necessary to make the thing difficult to attain. If he had been a great and wise philosopher, like the writer of this book, he would now have comprehended that Work consists of whatever a body is OBLIGED to do, and that Play consists of whatever a body is not obliged to do.​

 

[miero]

@pma would you try this test on your other speakers? It could show us how much is this issue relevant & testable . . .

 

[pma]

@pma would you try this test on your other speakers? It could show us how much is this issue relevant & testable . . .

Yes, at the end of the week.

BTW, it is not an "issue", it is the method to find impulse power capability of the speaker. With a sine or sweep of the same amplitude, the speaker would have been blown now. And, distortion in the burst impulse is much lower than with sine or sine sweep.

 

[miero]

I wonder if the measured distortion during the burst would be different if the burst would start not from a silence, but from a continuous sine tone at level for example -20dB under the burst peak.

 

[QMuse]

So, what prevents @amirm from doing impulse burst measurements of speakers?

Maybe because that method is on one hand very aggressive toward speakers and on the other hand not particularly accurate. Remember what you said yourself:

"The pain of sine burst distortion measurement is in length of the burst, which may not be long enough to get sufficient resolution for spectrum analysis."

 

[pma]

I wonder if the measured distortion during the burst would be different if the burst would start not from a silence, but from a continuous sine tone at level for example -20dB under the burst peak.

Like B&K?
https://www.bksv.com/media/doc/bo0385.pdf
It is a good paper, have a look.

Maybe because that method is on one hand very aggressive toward speakers and on the other hand not particularly accurate. Remember what you said yourself:

"The pain of sine burst distortion measurement is in length of the burst, which may not be long enough to get sufficient resolution for spectrum analysis."

Yes but here it was optimized to both burst and FFT length. Distortion of the burst itself is below 0.001% H2, which is in fact a window decay. This is, to me, accurate enough. And, burst testing has been used for decades to find power capabilities of speakers. All we can see is that it is almost pointless to speak about amp or DAC distortion. It is all within speakers.
And the method is not aggressive to speakers, you are wrong in this assumption. It is very gentle, because a sine or sweep of the same amplitude would destroy the speaker. The magic is in low energy and thus low heating with the burst. And as B&K AN properly states, it is closer to music signal than the sweep or sine.

 

[QMuse]

And, burst testing has been used for decades to find power capabilities of speakers.

Sure, but to accurately measure distortion?

It is all within speakers.

Indeed it is.

And the method is not aggressive to speakers, you are wrong in this assumption. It is very gentle, because a sine or sweep of the same amplitude would destroy the speaker.

Sure it would as they would last longer than burst. Btw, wouldn't it make more sense to have speakers acoustic response y-axis in SPL dB?

 

[miero]

So, what prevents @amirm from doing impulse burst measurements of speakers?

I guess another investment is required for this type of measurement:

 

[Hayabusa]

@pma

Did you verify that the acoustic output levels scaled with input power?
(What was the compression?)

 

[pma]

@pma would you try this test on your other speakers?

@miero , I made the measurements for you, on Quadral Ascent 90. The test burst peak voltages were 35Vp and 54Vp. Same burst as before. The Quadral woofers performed significantly worse than the Seas W18NX001 tested previously, even though Quadral woofers are bigger. Yes Seas price is $220 each (Amazon), though Quadral uses consumer noname plastic body drivers.
https://www.amazon.com/Seas-W18NX-001-Nextel-Woofer-E0042-08S/dp/B00ZYSJXQU
Measurements below, I think explanations are not necessary.

 

[restorer-john]

I know. Neither would I. It was a joke targeted at @restorer-john

It's not a joke, you will need well over 1.5kW to do this properly. And no, you won't burn speakers out.