Measuring a speaker

[hisbluness]

Hey guys,

I know this is not exactly about room treatment but I think this topic fits best here:

Im currently searching for a software/solution to measure the frequency response of different speakers.

I know the MLS (Maximum Length sequence) is probably the best way to do this but I have a hard time finding a good software to do this. There are several standalone products for pros which are just not affordable for me. The best thing would be a free software (I own an interface and a appropriate mic for measuring) that is easy to use but has all the functions you need. I wouldnt mind to pay either, but only if that is the only option.

Can any of you recommend something? Im looking for either a Mac or Windows XP software but I would prefer mac. Linux is an option too, but the least preferable

Thanks

chris

[nucelar]

Hi, is this for personal research/curiosity or for serious application (like publishishing a product specification)?
if the former: commercial MLS software is expensive afaik, I used it in university and our department developed its own system based on matlab. Maybe you should look into swept sine analysis, like FuzzMeasure (Mac).
Why not try to setup the speaker outdoors (in a quiet place) and just feed it pink noise and RTA the microphone signal? It may be sufficient for your needs...
Or set up a small anechoic chamber.
If you need a "certified" measurement, you not only need certified, calibrated equipment, but also should follow the standarized procedures (AES) to make it. I suggest you visit your local technical university (acoustics department or similar) where they may be able to help you, because it's not trivial.
Cheers

[hisbluness]

Hi,

thanks for the quick response.

Its not for a product specification but for a study I am doing. Its nothing too big and I nee to measure quite a few speakers which makes it difficult to do this in a remote place like university. I have a room I can do this in which is treated quite well so that should do, I guess.

I just found ARTA and it looks quite promising. Looks like it can do most of the stuff.

ARTA home

I will try to reach a university as well but I dont know if thats the most practical solution right now

[Ethan Winer]

Originally Posted by hisbluness

Im currently searching for a software/solution to measure the frequency response of different speakers.

I agree with the suggestion for FuzzMeasure, but that will not let you separate the speaker from the room. The only ways to do this correctly, for cheap, are 1) hoist each speaker up a tall flagpole and measure with a very tall microphone stand, or 2) find an auditorium and hoist the speakers up so they're at least 30 to 50 feet from any surface. Then you can gate the measurement to reject the reflections that arrive much later. The farther away the nearest room boundary, the lower in frequency your tests will be valid to.

--Ethan

[foxfyr]

...No need for such elaborate steps!

By placing the measurement object on a rigid, unobstructed ground surface such as a driveway or parking lot away from any walls or boundary surfaces, and placing the measurement microphone flush with the ground, an accurate approximation of the anechoic response of the object together with its image source may be obtained.

But there is much more. Without realizing it, you are opening a much larger can of worms than your realize. And with further examination, I might suggest that the 'frequency response' may actually be one of the last things you really want to know about the speaker.

I would suggest you look at the impulse response and then convolve the various alternative views. The frequency response will tell you simply that 'lots of stuff' is going on, none of which can be reliably discerned from the frequency response view.

What do I mean by this?

First, are you measuring in the near or far field? This is important as in the near field you are seeing more of the individual driver interaction and the superposition of the differing source signals offset in time - due to the differing acoustic centers of the drivers due to the physical offset inherent in separate non-coincident drivers, as well as the acoustic astigmatism of the acoustic origin among the various drivers themselves. (If you are not familiar with this, the apparent source of the signal is not necessarily the voice coil, but is often shifted forward - complicated by the origin physically shifting with frequency.)

The importance of this? As the various sources (not to mention the virtual diffractive edge sources) combine (superpose), the result is comb filtering in the frequency domain and polar lobing in the spatial planes.

Speakers with larger driver offsets using passive crossovers will suffer even more (and this is a primary reason for active crossovers with delay settings used to bring the measured driver signal offsets into alignment - at least with respect to one plane).

These differences will become less critical in the far field as the difference in signal driver offset become small relative tothe distance/time of travel to the far field listening position. But the effects of spatial polar lobing will be amplified as the spacing between lobes increases with radial distance.

Additionally, impulse based measurements can often be windlwed to remove the interference enabling a more atomistic examination of the isolated response minus the destructive interaction.

Thus, there is much more beneficial information to be gained by first measuring the system in the time domain and then convolving the frequency domain from this base information, The frequency response simply provides a partial snapshot of the resultant complex interaction of all of the various factors, without providing any insight into what the cvarious factors are or how they are interacting.

ARTA is an excellent place to begin. Just get it!!!!

There is lots more to this, and I realize that I have provided a very abbreviated 100,000 foot view, but trust me, thime based measurements are the way to go, and ARTA is definitely a worthwhile investment easily capable of packages costing literally 10x as much!

Below is an illustration generated in EASE that shows an example of the comb filtering and polar lobing, in this case based upon two single offset drivers at various fractions of a wavelength. Note: This occurs with separate drivers in their common passbands in the crossover region as well as with separate speakers, repeating with each increase in the order of magnitude (driver-driver, speaker speaker, array-array...and all of the combinations and permutations available as well! And then you get to look at this relationship in terms of the superposition of signals between the direct source and all of the virtual reflected sources inherent in a bounded space - eg. room!!!). And once you realize that, then the polar response and Q of a speaker become an even more critical response as one comes to understand the importance of controlled Q in minimizing destructive interference between sources and within rooms that causes so many of the anomalies that contribute to the frequency response that simply indicate that many more critical factors are involved in the actual response than what is normally assumed...and more often than not, these factors are also mixed up in what we assume is a strict measurement of the speaker itself in the frequency response. As, how many speakers are used in an anechoic chamber where multiple sources (at each order of magnitude) and boundaries are not a factor?

[Ethan Winer]

Originally Posted by foxfyr

I would suggest you look at the impulse response and then convolve the various alternative views.

Yes, this is why I suggested room measuring software.

--Ethan

[hisbluness]

Thank you for your input so far. As always, foxy is delivering fantastic information, although I cant claim to understand everything But I will work it out.

I might have been not 100% clear, and I must agree its probably the impulse response I want to measure. After that I can still use FFT to see the spectrum etc etc.

The only thing Im insecure about is this: Measuring the impulse response is usually done by, well, playing an impulse, right? I thought that this bring 2 problems along:

1) Usually a bad S/N ratio compared to different methods
2) This is nothing very healthy for a speaker when done at higher levels

And I thought that measuring with an impulse is always more likely to be affected by room acoustics.

What I know about MLS is that its a better approach since its more indepented from room acoustics and that the impulse response is obtainable by using cross-corelation etc etc.

My approach, and I know I havent read enough on it yet, would be to measure with MLS and obtain the impulse response from that (ARTA can do this afaik). Using FFT will present me the spectrum view which is basically what I want.

You are invited to correct me or tell me I know nothing

Im always amazed how quick this board is and how much valuable informatin is presented within hours. This subforum, along with Mastering, is just a pleasure to read. Education all along e.

[Ethan Winer]

Originally Posted by hisbluness

Measuring the impulse response is usually done by, well, playing an impulse, right?

It was in the past, sometimes using a starter's pistol or popping a big balloon. These days it's done with a sine wave sweep, and software code converts that into an equivalent impulse.

--Ethan