Speaker Max SPL Criteria?

[hardisj]

2) Max SPL for loudspeakers.


I perform max SPL testing using Klippel's MTON module. It uses multitone signal with user-set thresholds of distortion AND compression.

I made a video demonstrating how this is done. TURN YOUR VOLUME DOWN.

Rather than regurgitate everything I do, here is a direct quote from my Selah Audio Purezza review:
https://www.erinsaudiocorner.com/loudspeakers/selah_audio_purezza/

Maximum Long Term SPL:
The below data provides the metrics for how Maximum Long Term SPL is determined. This measurement follows the IEC 60268-21 Long Term SPL protocol, per Klippel’s template, as such:

• Rated maximum sound pressure according IEC 60268-21 §18.4
• Using broadband multi-tone stimulus according §8.4
• Stimulus time = 60 s Excitation time + Preloops according §18.4.1

Each voltage test is 1 minute long (hence, the “Long Term” nomenclature).

The thresholds to determine the maximum SPL are:

• -20dB Distortion relative to the fundamental
• -3dB compression relative to the reference (1V) measurement

When the speaker has reached either or both of the above thresholds, the test is terminated and the SPL of the last test is the maximum SPL. In the below results I provide the summarized table as well as the data showing how/why this SPL was deemed to be the maximum.


This measurement is conducted twice:

• First with a 20Hz to 20kHz multitone signal
• Second with a limited 80Hz to 20kHz signal

The reason for the two measurements is because it is unfair to expect a small bookshelf speaker to extend low in frequency. Applying both will provide a good idea of the limitations if you were to want to run a speaker full range vs using one with a typical 80Hz HPF. And you will have a way to compare various speakers’ SPL limitations with each other. However, note: the 80Hz signal is a “brick wall” and does not emulate a typical 80Hz HPF slope of 24dB/octave. But… it’s close enough.

You can watch a demonstration of this testing via my YouTube channel:

Test 1: 20Hz to 20kHz

Multitone compression testing. The red line shows the final measurement where either distortion and/or compression failed. The voltage just before this is used to help determine the maximum SPL.

Multitone distortion testing. The dashed blue line represents the -20dB (10% distortion) threshold for failure. The dashed red line is for reference and shows the 1% distortion mark (but has no bearing on pass/fail). The green line shows the final measurement where either distortion and/or compression failed. The voltage just before this is used to help determine the maximum SPL.

Test 2: 80Hz to 20kHz

Multitone compression testing. The red line shows the final measurement where either distortion and/or compression failed. The voltage just before this is used to help determine the maximum SPL.

Multitone distortion testing. The dashed blue line represents the -20dB (10% distortion) threshold for failure. The dashed red line is for reference and shows the 1% distortion mark (but has no bearing on pass/fail). The green line shows the final measurement where either distortion and/or compression failed. The voltage just before this is used to help determine the maximum SPL.

The above data can be summed up by looking at the tables above but is provided here again:

• Max SPL for 20Hz to 20kHz is approximately 103dB @ 1 meter. The compression threshold was exceeded above this SPL.
• Max SPL for 80Hz to 20kHz is approximately 102dB @ 1 meter. The compression threshold was exceeded above this SPL.

Surprisingly, the max SPL test from 20Hz to 20kHz is higher than that of the 80Hz to 20kHz test. However, only 1dB different. A re-test yielded the same results.

 

[hardisj]

TL;DR

For max SPL testing, Klippel recommends using multitone. They have the MTON module. The MTON module has tons of built-in templates per various IEC standards. Or, you can tweak them to your liking (if you wish).

MTON, in combination with KLIPPEL's ISC module (which is used to generate a room correction curve based on anechoic data; in this case, the NFS for Amir) will allow you to test a speaker for max spl in your room (huge benefit) based on how you want to define thresholds for compression and distortion.

To me, this is the most logical way to make this approach. Rather than worry about specific numbers for various HD profiles, pick some thresholds that make sense and then just stick with it. It isn't "perfect" but it's better than nothing. And it's fully repeatable. FWIW, the Revel F226Be was the highest output speaker I had tested... until I tested the Klipsch Heresy IV which clocks in at 110dB @ 1m 20Hz - 20kHz. The Revel was 106dB. I didn't test either using the 80Hz cutoff method because I feared damaging them (tweeter is the concern here) because those levels are already quite high and since the limiting factor is typically attributed to the woofer's output in full-range mode (in one way or another), if the woofer were cut-off at 80Hz, you could very well push the speaker to a level that would result in damage.


IMHO, there is no reason to re-invent the wheel here. I'm not trying to sound like I'm the only qualified person to speak on this but I have consulted with Klippel about my methods and they agreed that it was logical. It's real-world applicable, it's a hard number that is directly relatable and, if people don't like that, they always have the graphics. You want to see the compression? The graphics have it. You want to see the distortion profile? The graphics have that.

Use Klippel's ISC module to build the room-correction from the NFS measurements, then use MTON to set up your max spl testing. Done.

 

[hardisj]

Oh, note that the Klippel modules' graphics don't look like mine. I use MATLAB to generate my own graphics to look the way I want them to. That's why I have the little "registered" sign after "KLIPPEL" in the graphic; to denote that this isn't (currently) part of KLIPPEL's graphic. For example, KLIPPEL doesn't provide the SPL in the legend as I have shown. They just provide the voltage. They also don't provide the 1% Distortion line (red) that I do. A few other things here and there (like the title automatically including the speaker so if things are taken out of context it is always evident what the graphic belongs to). Also, it guarantees my axes are always the same (because KLIPPEL auto-scales and it's easy to forget to change that).

I take the data, import it in to MATLAB and then generate my own legends that have the (calculated mean) SPL so people can see directly the voltage/SPL without having to do their own calculations. I had to incorporate a few if/else statements to match their color format for a pass/fail test. I am working with KLIPPEL to incorporate this idea in to their next release. Not sure if they will do that but they have at least expressed interest in what I am doing here and why. Below is an example from the Klipsch Heresy IV testing (which hasn't been posted yet):

The KLIPPEL graphic:

Mine (with mean SPL in the legend):

 

[hardisj]

And that's it. That's my $0.02. I put a LOT of time in to coming up with this... a lot of testing and experimenting with what makes sense and figuring out how best to test without damaging a speaker. Been doing this for a while. I know some may prefer a parameter change here or there but I won't change my methods. They are in stone and have proved helpful thus far. As long as you are very clear about what exact thresholds you set and the methods you use, that is what matters almost more than why you chose them. It would be nice if you replicated my efforts so we could have more directly-comparable data. But my feelings won't be hurt if you don't. Hope that helps!

 

[andreasmaaan]

And that's it. That's my $0.02. I put a LOT of time in to coming up with this... a lot of testing and experimenting with what makes sense and figuring out how best to test without damaging a speaker. Been doing this for a while. I know some may prefer a parameter change here or there but I won't change my methods. They are in stone and have proved helpful thus far. As long as you are very clear about what exact thresholds you set and the methods you use, that is what matters almost more than why you chose them. It would be nice if you replicated my efforts so we could have more directly-comparable data. But my feelings won't be hurt if you don't. Hope that helps!

100% agree with you @hardisj.

I think there are actually two parallel topics being discussed here, and the thread has got a bit confused as a result.

The first topic is max. SPL testing. When it comes to this, your approach is on-point IMO.

The other topic is the question of how to undertake distortion testing in a way that correlates to audibility. This is a very different topic IMHO.

I think it makes sense to treat them independently, as you have here.

Testing for max. SPL and testing for distortion in way that correlates to audibility are (or should be) two separate things.

 

[bennybbbx]

A real world test is how loud can a speaker get at 50 hz without hearable distortion on a base drum. I notice that 6-10 db before heavy speaker or amp clipping occur a rock basedrum(are around 50 hz) change in sound and sound ugly. I did not know how much thd this is. but you can try it out. play a rock basedrum until it sound diffrent and look which level the 50 hz signal in basedrum have. now use measure generator how much THD it is.

this video is a good test for rock basedrum.

 

[pkane]

We just need to get @pkane to incorporate it into Deltawave. He can provide distortion upon request using his Distort software, and if Deltawave could compute the DS metric value it would make all this easier to do. Of course easy for me to say since Paul would do all the work. I think all the needed formulas for the DS are in that paper.

Adding clipping distortion to DISTORT is easy -- in fact there are already a few of the clipped transfer functions supported by the app. I'll take a look at the DS metric, but at a quick glance it looks very similar to the DF metric proposed by our friend @Serge Smirnoff, except with some psychoacoustic weights.

 

[fredoamigo]

amirm Have there been formal studies/papers that we/I should be reading on this?

it is the most complete document and study in my possession but it is unfortunately in French

https://tel.archives-ouvertes.fr/tel-00806288v1/document

 

[hardisj]

Why 1% THD at 10 kHz [@hardisj]?

Not sure why you're asking me this. I have nothing to do with that metric.

 

[pozz]

@amirm I guess you've already read/watched Klippel's own material on this topic? This is the most pertinent one here.

You may also want to consider using m-noise instead of IEC-prescribed band-limited pink noise. I'm not sure it's clear yet whether the industry will pick on on m-noise as the new de facto standard for SPL testing, but it does seem to be getting talked about a lot lately.

Good article about M-noise in Production Partner: https://www.production-partner.de/test/meyer-sound-m-noise-schallpegel-messen/

 

[pozz]

Another good piece about Max SPL testing using M-Noise and the coherence measurement: https://www.aes.org/e-lib/online/browse.cfm?elib=20582

Accompanying article: https://www.merlijnvanveen.nl/en/study-hall/154-coherence-and-reverberation