Measuring speaker output at multiple power levels

[CustomSpeakerPods]

Hello,
My name is Brandon, I am the owner of Custom Speaker Pods. My company deals mainly with higher powered pro-audio speakers and amplifiers for car audio. I am interested in testing speakers that I sell or plan to sell in the future.
The end goal of the testing will be to provide speaker frequency response data at multiple power output levels, and provide this data to customers interested in the speaker. Power output levels could include the standard 1 Watt measurement level, as well as a median power level and up to the manufacturers RMS power level. I am hoping to prove or disprove speaker performance at the RMS rating.
My current testing setup includes a car stereo head unit connected to my PC with a 2.2kW single channel, full range class D amplifier with rated output of 800 watts at 4 ohm. Audio measurement includes a Dayton Audio UMM-6 microphone and REW for signal generation and data collection.
If anyone can provide insight on methods/equipment for performing this test I would greatly appreciate it.

Thank you, Brandon

 

[fpitas]

The elephant in your room is one you probably know well; a lot of those car drivers can't really take the rated power. Or maybe for a millisecond.

 

[HarmonicTHD]

REW is good. Alternatively there is also Multitone analyzer which is similar and also offers test scripts. It is programmed by @pkane who will have more insights than I do.

I presume you have a calibration file for the Dayton and can calibrate for SPL too?

 

[CustomSpeakerPods]

The elephant in your room is one you probably know well; a lot of those car drivers can't really take the rated power. Or maybe for a millisecond.

Yes, and that is what I'm looking to test. Currently the industry matches amplifiers and speakers based on their respective RMS ratings, but I want to make sure that is a safe thing to do mainly to avoid people blowing stuff up based off of possibly false information. There is available data/reviews/testing done on amplifier output, but I don't know of anyone really testing the ability of the speakers, and with new speakers coming out nearly every day, I want to be able to give people product suggestions that I know work, and avoid warranty returns etc based off of bad information.

The example I am working with now, and my planned build for my personal vehicle, is a combination of the following:
Kicker ST7 4ohm midrange: 4 per door, each door powered with a DS18 GFX-2.2k1. I was going to wire the speakers in parallel to provide an approximate 1-ohm load to the amplifier which is specified to output 2.2kW at 1-ohm. This would give the potential ability to give each speaker a maximum wattage of 2.2/4 kW (550 W). The speakers are rated for 250W RMS and 500W Peak.

I am wanting a testing procedure to allow me to understand how a system like this will behave at elevated power levels; what happens to the frequency response, how much power can be applied before failure, how long will the components last at the applied power, etc.

All of this testing will mainly be for the speakers since amplifier output is somewhat understood and can be measured easily using scopes etc.

I feel like I am recreating the wheel at this point and would love to have a conversation about this subject... or be pointed in the right direction.
Thanks, Brandon

 

[CustomSpeakerPods]

I presume you have a calibration file for the Dayton and can calibrate for SPL too?

I have the calibration file for the UMM-6. The issue I see with this microphone will be the clipping level at higher wattage testing. How can I overcome this? Multiple microphones at varying distances? Physical covers/filters? I have no idea.

 

[fpitas]

REW is good. Alternatively there is also Multitone analyzer which is similar and also offers test scripts. It is programmed by @pkane who will have more insights than I do.

I presume you have a calibration file for the Dayton and can calibrate for SPL too?

But, be a little careful. Inexpensive measurement mics may not respond linearly at high levels.

 

[HarmonicTHD]

But, be a little careful. Inexpensive measurement mics may not respond linearly at high levels.

Good point. Yes and I don’t know or own the Dayton, so can’t really say.

 

[fpitas]

The "power to failure" test is straightforward, at least. From what I've heard, you can tell by the smell and quantity of smoke whether you're getting close to failure

 

[fpitas]

One thing to consider for your power test. The voice coil is cooled by the cone movement pumping air. But you can only get so much Xmax, so you can't go too low in frequency. I'm thinking the max power will depend on frequency a lot because of that.

 

[Deleted member 48726]

The "power to failure" test is straightforward, at least. From what I've heard, you can tell by the smell and quantity of smoke whether you're getting close to failure

"Getting close to failure" ?? -Smelling the varnish is "close to failure". -SEEING the varnish as vapor is "failure".

 

[fpitas]

"Getting close to failure" ?? -Smelling the varnish is "close to failure". -SEEING the varnish as vapor is "failure".

I predict that Brandon will become quite expert at this technique.

 

[fpitas]

Another observation on power testing is that the pole piece and magnet assembly serve as a (crappy) heat sink until you get them hot. So, you have to decide on the duration you want for your power test.

 

[DVDdoug]

Speaker ratings are "statistical". A speaker is supposed to handle regular program material hitting the rated power on the peaks with a much lower average level. And a 100W tweeter only has to handle the weaker high-frequency part of that program. You can fry speakers with continuous test tones, especially tweeters!

And when you drive an amplifier into clipping the average power goes up even though the peaks are clipped-limited, and clipping generates harmonics that go to the tweeter. There is a popular myth that a clipped amplifier is worse than a higher-rated amplifier but either way you can over-power the speaker and the higher power amp (if it's actually putting-out more power) is even worse.

Here is an old JBL paper about speaker ratings. But I don't think too many consumer speaker manufacturers stick to these standards.



P.S.
As you may know, there are car stereo loudness competitions and I'd guess these guys have some good information about reality vs. published specifications.

 

[fpitas]

Speaker ratings are "statistical". A speaker is supposed to handle regular program material hitting the rated power on the peaks with a much lower average level. And a 100W tweeter only has to handle the weaker high-frequency part of that program. You can fry speakers with continuous test tones, especially tweeters!

And when you drive an amplifier into clipping the average power goes up even though the peaks are clipped-limited, and clipping generates harmonics that go to the tweeter. There is a popular myth that a clipped amplifier is worse than a higher-rated amplifier but either way you can over-power the speaker and the higher power amp is even worse.

Here is an old JBL paper about speaker ratings. But I don't think too many consumer speaker manufacturers stick to these standards.

Everything you said is true. But the car crowd's expectations are a little different.

 

[CustomSpeakerPods]

The "power to failure" test is straightforward, at least.

Yes and no. What do I test with? Sweep, white noise, pink noise?
Do I limit the frequency range as per your statement about the Xmax? (which in itself is an interesting mode of the test; how does the frequency range change with power?) I noticed during a quick bench test at higher wattage (125W) the sound starts to 'clip' (completely observational, just sounds static-y), is this due to reaching the mechanical limits at lower frequencies?

As you may know, there are car stereo loudness competitions and I'd guess these guys have some good information about reality vs. published specifications.

Doug,
Great to have you as part of this conversation as well. Your insights from the Arduino conversation were great.
To your point about the SPL competitions: I know these guys very well as I live in this world, but their focus is more towards loud sub-stages. My products (and business focus) compliment the loud sub-stage by allowing people to install multiple speakers to keep up with big-power subs. I have used the Term-lab system a couple times. There are a couple limitations of this system. 1. It is designed to measure low frequencies at high SPL, it uses an atmospheric pressure sensor to measure the SPL (I took the sensor apart for analysis). 2. It has no data logging functionality, peak only. I spoke with the product creator briefly about this (and maybe I should have further discussions with him?).

 

[No. 5]

Repeatability is key here. Have you considered a tetrahedral test chamber? There's DIY plans for them available, as well as directions on how to use them. If you go that route, the SPL capabilities of your microphone are going to be even more important, but I'd suggest finding a testing method that works dependability before getting a new mic.

It's probably also beneficial to have some electronic test gear as well so you can verify the actual voltage and current going into the drivers under test.

Cool project!

 

[CustomSpeakerPods]

Repeatability is key here. Have you considered a tetrahedral test chamber?

Interesting. I don't know how far into this I want to get as there is a point of diminishing returns on a project like this. I agree that it needs to be repeatable (as should testing of any kind). Before I start to think about chambers, I would need to understand what 'microphone' I would need to measure frequency responses that range up to/over 130dB.

 

[No. 5]

I think that's a very reasonable place to start. I'm very skeptical that the mic you have will be able to take that kind of SPLs, it is testable, but there are microphones on the market that most certainly can. Off the top of my head for reasonably adorable options, the iSEMcon EMX-7150 can supposedly do 145dB with less than 3% THD, the Audix TM1 is spec'd to do up to 140dB (no THD given). For almost double what those cost, an Earthworks M23 is spec'd to 140dB also (and no THD there either).

There's probably cheaper options too if you look, the Behringer B-5 is supposed to be able to do 140dB, but you'd have to calibrate it (not that had to do). If you have a large budget, the sky's the limit, I think Brüel & Kjaer has microphones that'll do over 170dB.

 

[CustomSpeakerPods]

Let me back up a bit: A speaker with a 95dB sensitivity running at 250W will only produce ~118dB. The UMM-6 is rated for up to 127dB, so should work.. for now. Is it possible I have an incorrect setting in REW? My issue is that it warns of signal clipping during a sweep test at higher volumes. Or is it a factor of the way the mic handles the higher frequencies?

 

[RayDunzl]

Is it possible I have an incorrect setting in REW? My issue is that it warns of signal clipping during a sweep test at higher volumes.

On Windows with REW and a UMIK-1:

Reduce the recording level.

This gives the UMIK-1 in REW more headroom.

REW figures out the re-calibration necessary on its own.

Turn it down for more headroom.

Previous test:

UMIK-1 High Ambient SPL Measurements in REW?

I recently purchased a UMIK-1 to use with REW for calibrating speakers, and I ran into a question when I was performing initial checks in REW: When I open the SPL meter and let it log sound without playing anything (just recording background ambient sound, no pink noise or anything of that...

www.audiosciencereview.com