Decent speaker measurement setup

[HighTower]

I've searched this website for about two hours, not the right way probably. Anyway, I run a company and we are now in the process of having our own drivers custom built to our specs. In the past we just purchased off the shelf products from established manufacturers so there was no real need for testing these drivers as we would like to do now.

I've seen a lot of setups that use the typical soundcard, jig and mic but nobody seems to be using an amplifier. Now I understand that's not necesarry for T/S parameters but we would like to go a bit deeper:

- Freq. response
- THD down to < 0,1% and THD at high SPL levels (up to drivers thermal or mechanical limit)
- Waterfall / spectral decay

We've done these measurements in the past but only in the "quick & dirty" way, with expected results. The FR was OK off course but accurately measuring THD was a problem and for the spectral decay we had to get extremely close to the driver to get decent results.

The problems where probably in the lack of an anechoic chamber and the fact that the amplifier we used had quite a bit of distortion, especially at low levels (as is with crossover distortion).

So now we are asking ourselves if we should build our own test setup from or just buy an off the shelf "kit"? I don't mind spending a few grand on a decent setup but I would very much like it to be straightforward and somewhat idiot proof. I don't want my staff to be testing speakers at high levels only to be wondering if the distortion is coming from the speaker or from the sound card.

What we used up untill now:

- UMIK 1
- Arta
- M-Audio sound card (I'd have to check the exact model)

Besides that we have an HP 8903B audio analyzer running Peter Millet's SW.

So what's your best bet? Please note that we run a business so time is money, both in setting up the equipment as in performing the measurements themselves (faster procedures = saving money).

And I am curiosu to know which all-in-one solutions like the AP APx1701 are available and what they cost? Thanks in advance!

 

[ergo]

The lowest end 'pro' setup would probably be CLIO 12 + CLIO amplifier and potentially the QC add on for the software that would allow creating test scripts that allow very easy push button testing and pass/fail
CLIO 12 | Audiomatica Srl

Step up in price and in my opionion the functionality and ease of use is
Functions of FX100 Audio Analyzer I NTi Audio
but Flexus software dows not do waterfall afaik.

I have used both and had hands on experiences with both

Then there is the new player
Avermetrics | AverLAB | Avermetrics
It has attractive price, but needs some looking into how much loudspeaker specific measurement support they have already. Price/performance looks good.

Then there is Loudsoft in Denmark
FINE R+D-Anechoic measurement to 100kHz
offers also both hardware and software

Also Listen Inc
https://www.listeninc.com/products/soundcheck-software/measurements/
Who offer both software and hardware

And Audio Precision that will be most expensive of these I believe, but they keep adding good loudspeaker oriented tests and getting better in that field too.

 

[1audio]

Actually Listen will prove to be the most expensive.

The quick way to get decent speaker measurements is Dayton Audio OmniMic V2 Computer Based Precision Room Measurement System Follow the directions and you will get pretty decent results with little hassle. If none of your staff want to be measurement gurus this is the best value.

Arta should be perfectly adequate if you dive deeper. You do need a good amp and some decent space to work in. I have used open warehouse space with machinery running in another part to get anechoic measurements to 20 Hz (the ceiling was 30' up) using windowed ground plane measurements. You need an adequate amp of course (.05% at full output would be fine, check thrift stores for older AV receivers) but your speaker will be the bigger issue regardless. Use a chirp for response. Move closer for distortion to get the snr up. The trick is a windowed measurement where the nearest surface is a good distance except the floor. A hard concrete floor works well with 20+feet (3M) between the speaker and mike and the nearest surface. if the micrphone is on the floor in an expansion crack the results are even better. putting the mike on an absorbtive pad will ruin the measurement (this from experience). Here is more on ground plane measurements Ground Plane Measurements for optimiziing an loudspeaker

Don't trust the speakers to all be the same. The unit to unit variation can be significant depending on your supplier. Building a test fixture for drivers may be important and can be simple since you are looking for consistency, not absolute values. A box with a hole for the driver and a mike inside can do it.

 

[HighTower]

The lowest end 'pro' setup would probably be CLIO 12 + CLIO amplifier and potentially the QC add on for the software that would allow creating test scripts that allow very easy push button testing and pass/fail
CLIO 12 | Audiomatica Srl

Step up in price and in my opionion the functionality and ease of use is
Functions of FX100 Audio Analyzer I NTi Audio
but Flexus software dows not do waterfall afaik.

I have used both and had hands on experiences with both

Then there is the new player
Avermetrics | AverLAB | Avermetrics
It has attractive price, but needs some looking into how much loudspeaker specific measurement support they have already. Price/performance looks good.

Then there is Loudsoft in Denmark
FINE R+D-Anechoic measurement to 100kHz
offers also both hardware and software

Also Listen Inc
Measurements – Listen, Inc.
Who offer both software and hardware

And Audio Precision that will be most expensive of these I believe, but they keep adding good loudspeaker oriented tests and getting better in that field too.

Wow thanks, that's a lot more info than I'd hoped for! I'll be digging in to those options but for some reason CLIO QC with the Amp/Switch Box seams a fair option.
Waterfall is a fixed requirement, even free software like ARTA can do this (and we like it).

There is one thing I can't find for instance for the CLIO system and that's the Noise Floor if the combined system. Maybe it's a non-issue (because the speakers distortion is always bigger than that of the measurement system?) but I'd like to see that confirmed before buying anything. Also considering that we don't have an anechoic chamber so we will be doing nearfield tests mostly and I can imagine that could be a problem when doing measurements at higher SPL levels.

Actually Listen will prove to be the most expensive.

I know, which is why I don't want to have to rely on that more than necessary. When comparing samples by listening it's easy to make the mistake of focussing on a particular aspect of a loudspeaker and neglect other important factors. One sample may sound "better" while in fact it's just louder due to a higher fs, at the expense of low frequency extension.

The quick way to get decent speaker measurements is Dayton Audio OmniMic V2 Computer Based Precision Room Measurement System Follow the directions and you will get pretty decent results with little hassle. If none of your staff want to be measurement gurus this is the best value.

Thanks for providing the option but I'm affraid this will be a bit problematic in our case. In our company where the concrete floor is there is also a lot of noise. Besides we do have some capable people in our company to setup the measurements but in the end I like the idea that less educated people can do standardized measurements with repeatable results with the push of a button as ergo mentioned.

Arta should be perfectly adequate if you dive deeper. You do need a good amp and some decent space to work in. I have used open warehouse space with machinery running in another part to get anechoic measurements to 20 Hz (the ceiling was 30' up) using windowed ground plane measurements. You need an adequate amp of course (.05% at full output would be fine, check thrift stores for older AV receivers) but your speaker will be the bigger issue regardless. Use a chirp for response. Move closer for distortion to get the snr up. The trick is a windowed measurement where the nearest surface is a good distance except the floor. A hard concrete floor works well with 20+feet (3M) between the speaker and mike and the nearest surface. if the micrphone is on the floor in an expansion crack the results are even better. putting the mike on an absorbtive pad will ruin the measurement (this from experience). Here is more on ground plane measurements Ground Plane Measurements for optimiziing an loudspeaker

Thanks for the tip. We've done windowed measurements with ARTA before in a similar way (just with a smaller room, hence smaller window) but found the results not to be repeatable; we expected the noise in the room to be the main culprit but to be honest the entire test setup was made on the fly. The Spectral decay plots came out quite nice though as long as we used nearfield measurements.

Don't trust the speakers to all be the same. The unit to unit variation can be significant depending on your supplier. Building a test fixture for drivers may be important and can be simple since you are looking for consistency, not absolute values. A box with a hole for the driver and a mike inside can do it.

Yes that would in fact be the next step. For now I'm focussing on the measurement equipment that can do all this AND the requirements for the room in which we do the measurements.

Given an average office building (with a ventilation system running in the background, traffic in the street, etc.), do we need to build an acoustic room in order to be able to measure THD reliably down to say 0,1% THD and decent waterfall plots? And if so:

How big does this room need to be for practical measurements? We're not interested in spectral decay < 200 Hz but accurately measuring THD would be nice. I understand the concept of gated measurements; the difference in distance between DUT-to-MIC and DUT-to-first-wall-and-MIC determines the lowest frequency you can accurately measure. What I don't know is if the measurement system needs at least a full wavelength (or more) for accurate measurement or that 1/2 or even 1/4th will be sufficient? I suspect for THD you would need at least 1 full cycle for an accurate measurement, but maybe for THD the reflections are not particularly harmfull for the measurement (as it would be for spectral decay)?

And last question: IF we were to build this kind of room, is the key factor in;

1. Eliminating noise from outside (since we will be using gated measurements), or;
2. Eliminate reflections within the room using acoustic wedgies 😀, or;
3. Both!

I tried finding examples of measurement setups on this forum and beyond but failed mostly. Therefore links to examples of low-pro or DIY setups are highly appreciated.

 

[ErikdeBest]

I am curious 🙂 which company do you own?

 

[HighTower]

Haha I tend to not disclose such information online, for the simple reason that I don't want to disclose our intentions to -for instance- competitors. We can't avoid people copying our products or ideas but at least we can make sure we get the biggest head start possible 😀
Therefore neither my name nor that of my company is visible on this website but you can DM me if you persist on knowing...

 

[ErikdeBest]

Hi! I am originally Dutch and interested in Dutch companies manufacturing loudspeakers - but it is really just curiosity and not worth going into PM.

The content of the thread is very interesting, thanks for starting!, as I am looking into improving my loudspeaker measurement techniques as well.

 

[HighTower]

Hi! I am originally Dutch

Obviously, since you are "de Best" 😎

and interested in Dutch companies manufacturing loudspeakers

Wel technically we are not actually manufacturing them but it will be the first time that we will have drivers tailor made to our requirements.

- but it is really just curiosity and not worth going into PM.

The content of the thread is very interesting, thanks for starting!, as I am looking into improving my loudspeaker measurement techniques as well.

You're welcome! I was a bit surprised that I couldn't find more topics like this but maybe that wasn't just due to my poor investigative skills..

 

[1audio]

How big does this room need to be for practical measurements? We're not interested in spectral decay < 200 Hz but accurately measuring THD would be nice. I understand the concept of gated measurements; the difference in distance between DUT-to-MIC and DUT-to-first-wall-and-MIC determines the lowest frequency you can accurately measure. What I don't know is if the measurement system needs at least a full wavelength (or more) for accurate measurement or that 1/2 or even 1/4th will be sufficient? I suspect for THD you would need at least 1 full cycle for an accurate measurement, but maybe for THD the reflections are not particularly harmfull for the measurement (as it would be for spectral decay)?

And last question: IF we were to build this kind of room, is the key factor in;

1. Eliminating noise from outside (since we will be using gated measurements), or;
2. Eliminate reflections within the room using acoustic wedgies 😀, or;
3. Both!

I tried finding examples of measurement setups on this forum and beyond but failed mostly. Therefore links to examples of low-pro or DIY setups are highly appreciated.

Having been there and done that- first, the electrical noise of the measurement system will not be a real issue unless you are way out in the countryside. There are so many acoustic and mechanical noise sources that you cannot control. If you need to get low acoustic noise then you could suspend a heavy concrete floor on air bags with similar walls etc. Nothing beats mass for isolation. However that's really a NASA level project. I find hard walls far enough away to work well. It makes windowing really clear. If you can average measurements (especially coherent averaging) you can remove many noise sources.

I have used double wall construction with two layers of sheetrock glued together with slightly compliant glue (constrained layer) and staggered metal studs. It worked quite well for both isolation and keeping bass in the room and made for a good sound room as well.

If you want to measure harmonics at .01% you need pretty special stuff. Microphones and mike preamps need to be validated and not many are really good to less than .01%. Ambient noise will also limit what you can measure. If you use 94 dB (1 pA) as a reference then an ambient noise floor of 34 dB would block a THD + N measurement below .1%. With FFT you can get lower. DiAna DiAna, a software Distortion Analyzer would be good for this because it can pull harmonics out of noise. The only issue on size is to pick frequencies where the room modes won't reinforce or null a fundamental or a harmonic. Measure near field and at various axis.

Meaningful waterfall measurements can be a challenge as well. The background noise will limit what you can do and won't be easy to chase. Not sure if there is a system that can coherent average impulse measurements for a waterfall. Maybe Praxis. Not sure of anything else. Environmental noise will limit LF usefulness. Using a very exotic sound chamber meant for electron microscopes I still had LF noise issues measuring headphones.

The lowest useful frequency is the reciprocal of the reflection time. 20 mS (20 feet here) would limit to 50 Hz and thats cutting it fine. So a space roughly 20 feet (6M) by 20 feet (6M) by 10 feet (3M) could get to 50 Hz. In truth room response departs radically from anechoic response down there anyway so "your judgement may vary" on what you decide is correct. I was using an improvised basketball court in a large warehouse for measurements. It worked great until someone realised what the area cost per month and they "reclaimed" it for product storage.

 

[HighTower]

Hi Demian, thanks for you input! For the record: we wanted to measure down to 0,1%, not 0,01%. And I don't think we need waterfall spectrum data below 200 Hz, so far I haven't found that kind of data very usefull; it was my idea that we could probably get more relevant information (step response, group delay) from a decent box calculation programm after carefully measuring TSP's?

But we would like to measure THD even at low frequencies; is this at all possible at low frequencies without a huge room?

 

[ergo]

CLIO FW-02 and CLIO amp combination is able to meet your criteria for <0.1% THD. I've verified that, just do not have the data saved.

One sure way to lessen the room impact to your THD measurement is moving the microphone close to the driver. But one should then consider that the sound pressure can raise very high. For example if you set the SPL to 100dB at 1m from driver and then move the microphone to 5cm from driver your SPL at microphone will increase to approx 127dB!
That is very loud and many cheaper microphones will totally overload or distort more than your set limit. It's worth to invest in a lab micropone like the ones from GRAS for example
Measurement microphone sets
that can handle upto 150dBSPL for example. CLIO's own microphones are not capable of this.

CLIO has a way to measure THD with sine sweep (chirp) or with stepped sine. Latter will be a better option for bit noisy environment. The above example with 127dB level at microphone would still give you 60dB+ headroom if your surrounding noise is <67dBSPL...

 

[1audio]

Hi Demian, thanks for you input! For the record: we wanted to measure down to 0,1%, not 0,01%. And I don't think we need waterfall spectrum data below 200 Hz, so far I haven't found that kind of data very usefull; it was my idea that we could probably get more relevant information (step response, group delay) from a decent box calculation programm after carefully measuring TSP's?

But we would like to measure THD even at low frequencies; is this at all possible at low frequencies without a huge room?

To measure .1% your test system residual needs to be at least 6 dB better. Usually lowering the level at the mike helps except for the noise floor.

 

[HighTower]

Allrighty then! I've gone ahead and requested pricing for Clio 12 incl. a cailbrated mic and the QC box (with the amplifier included). I know it's not the best piece of kit available but I think that it should meet our requirements for now and it allows us to get some hands-on experience. The system is compatible with 3rd party mics so we might buy a better one at some point (like GRAS mentioned above) if that should prove to be necessary. Thanks for all the input provided!