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

[amirm]

@miero , I made the measurements for you, on Quadral Ascent 90. The test burst peak voltages were 35Vp and 54Vp.

So your test was pumping 135 and 365 watts respectively into that 7 inch driver, correct?

 

[pma]

So your test was pumping 135 and 365 watts respectively into that 7 inch driver, correct?

Definitely incorrect. 35Vp stands for 35Vpeak (of the windowed burst). 35Vpeak would make 24.75Vrms of equivalent sine wave. 24.75Vrms makes 76.56W on 8ohm load (two 8ohm woofers in the box). So I am deep within the specs of the speaker. The second burst would make 182W of equivalent sine on 8ohm. And the windowed burst duration is 0.2s. It was all explained in the post #1 of the thread.

 

[restorer-john]

Here's a brief summary of how Audio used to test speakers for peak capabilities, this was a Paradigm 7se review, 8" 2 way @ $469 pair:

 

[pma]

@restorer-john John, and I even do not use multiple bursts, just 1 burst two times. 1st run is to set the level (it is captured into DSO), the 2nd run, after at least 30s pause, is for measurements + recording of the wav file, that contains both electrical and acoustical output. As our capabilities to analyze recorded files have improved grossly in last decades, we do not need to send the sequence of bursts to the speaker. Yes the burst is a bit longer, 10 periods of 100Hz Blackman windowed, but speaker stress (other than cone excursion) is minimal, as only just 1 burst is sent after a long time interval.

 

[restorer-john]

and I even do not use multiple bursts, just 1 burst two times.

We used to do burst testing (many years ago now) using an externally bridged Perreaux PMF-3150B* (~1KW at 8R bridged) and listened for audible noises. At low frequencies THD on bursts (low cycle count, not shaped, but zero-cross toneburst source) was an obvious "rattle" when limits were reached. Non-destructive.

*conservatively rated at 300W@8R both channels driven. It achieved 360 and 600@4R and around 1KW BTL (ext) @8R. Classic Hitachi triple MOSFETs from the 1980s for each rail/ch.

 

[pma]

We used to do burst testing (many years ago now) using an externally bridged Perreaux PMF-3150B* (~1KW at 8R bridged) and listened for audible noises. At low frequencies THD on bursts (low cycle count, not shaped, but zero-cross toneburst source) was an obvious "rattle" when limits were reached. Non-destructive.

*conservatively rated at 300W@8R both channels driven. It achieved 360 and 600@4R and around 1KW BTL (ext) @8R. Classic Hitachi triple MOSFETs from the 1980s for each rail/ch.

Here I can hear the loss of cleanliness and the sound (even that of the single burst) starts to be contaminated by mechanical noise. In spectrum, you can see a loss of resolution and noise-like decay. Best seen in this measurement
https://www.audiosciencereview.com/forum/index.php?attachments/quadral_54vp_spectrum-png.53034/
From the time record, we can see the amplitude point where the windowed burst starts to loose sine shape and becomes distorted - same is observed in spectrum when the amplitude is set to such value.

P.S. Re the amplifier used, I can also get about 600W/4ohm, it is my CFA in the bridged mode.

 

[miero]

Another idea to try. Make a difference audio of normalized acoustic/electric recordings. Maybe it will tell us something...

 

[restorer-john]

cleanliness and the sound (even that of the single burst) starts to be contaminated by mechanical noise. In spectrum, you can see a loss of resolution and noise-like decay. Best seen in this measurement

Yes, the sine loses its symmetry as the amplitude increases within a cycle or two and the recovery is affected from cycle to cycle. It's easy to hear and I agree, we couldn't do analysis due to the low cycle count and acquisition gear back in the day.

The capture and post process method you have employed is an excellent way to see this outside of real time.

 

[Hayabusa]

@miero

View attachment 53032

Would be nice to see actual acoustic output levels at different input levels to estimate how compressed the output signal gets at higher power levels. Now it looks like you manually overlay the input and output without any reference?

 

[KSTR]

Another idea to try. Make a difference audio of normalized acoustic/electric recordings. Maybe it will tell us something...

You cannot directly subtract the input from the output unless you correct the input to a modelled output having the same frequency response (magnitude and phase). This can be done but it is a bit complicated.

Much easier to do, and equally revealing, is subtraction of output signals taken at different power levels. For example, doubling the voltage (==4x power) from step to step, then you only need to scale down the recording of louder signal by a factor of two (6.02...dB) and in theory you'd have a residual that is uncorrelated noise only. In practise you will see a lot of difference, at lower levels mostly simple harmonic distortion. At higher levels we'll start to see large differences when the driver is reaching its motor and mechanical limits of linear operation, and power compression kicking in (unless current drive is used).

Many thanks to @pma and @restorer-john for this, it reflects my personal experience as well that with good drivers you can easily apply 10x...30x the thermally rated power without any problems and moderate distortions, even for tweeters, unless you hit excursion limits (including the suspension, or even the cone itself, folding over). 30x power still only is 5.4x the voltage, a headroom of 15dB.
Electrically, the limit is when you melt the wire or have windings of the coil arcing over (with a two-layer VC wound on the same side of the former).

 

[pma]

Would be nice to see actual acoustic output levels at different input levels to estimate how compressed the output signal gets at higher power levels. Now it looks like you manually overlay the input and output without any reference?

Yes the acoustical levels are not calibrated in these measurements, one can get a rough estimation from speaker's sensitivity (CNO/Seas 85dB/2.83Vrms/m, Quadral 89dB/2.83Vrms/m). The measurements are made close to membrane, to get enough S/N and spectral resolution, but always checked for possible microphone overloading and distortion as explained in post #8
https://www.audiosciencereview.com/...f-speakers-with-sine-bursts.11794/post-341384
Measurements from 1m are noisy and messed up with room reflections and the show higher distortion for the reason of reflections. As explained by Keele,
[1] Keele, D.B.: Low-Frequency Loudspeaker Assessment by Nearfield Sound-Pressure Measurement, JAES, 45th Convention AES, LA

harmonic distortion measurement can be made in in the nearfield and these measurements should correlate well with an identical set of measurements in the farfield if all distortion components are within the specified frequency limit (which is related to membrane radius).

Regarding "compression", this is reflected in harmonic distortion.

 

[pma]

Significant improvement of resolution of burst spectral distortion analysis was achieved by peak-hold method.

 

[pma]

Keele, D.B.:
Assessment of Loudspeaker System Short-Term Maximum Output

https://docplayer.net/145907189-Assessment-of-loudspeaker-system-short-term-maximum-output.html

However, who cares, except for @restorer-john ?

See the huge difference between NHT-M00

and PSB Stratus Gold

(from the same source)

 

[ScottG]

-most useful for tweeters and their applied high-pass filter (..which SL isn't doing here).

http://www.linkwitzlab.com/frontiers_6.htm#Y -

 

[restorer-john]

Here's the PDF @pma referenced if anyone is interested.

 

[restorer-john]

The only part I don't agree with is this:

The nominal rated impedance of the speaker may be wildly different to the actual impedance at the spot frequencies he used to calculate peak input power.

 

[pma]

I regret to see you in a "Sabbatical Leave", John, you were one of the very few here with whom I shared similar view points. Hope you will take a good rest .

 

[Wombat]

I regret to see you in a "Sabbatical Leave", John, you were one of the very few here with whom I shared similar view points. Hope you will take a good rest .

I am happy to see differing intelligent/practical contributions. I don't feel anxious about them, rather, something to ponder and discuss. That is what a forum is for.