testing and measuring a microphone?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

pucho812

Well-known member
Joined
Oct 4, 2004
Messages
14,831
Location
third stone from the sun
what's a good way to test and measure a microphone?  I don't have access to an anechoic chamber. I thought of using one of those silent guitar cabs,  you know the isolation cab where it's a speaker inside a box that has a mic mount and is designed to keep the sound in and be as dead as possible.
 
cyrano said:
Do it outside, as high as possible?

Bill putnam sr. used to do that when testing speakers at his house in Reseda which is an area of L.A.  I doubt I could do that in the current living environment of a complex.  Plus I would assume things like wind would change measurements.
 
I don't think you can use an isolation cab because it would be too small. I think you would get standing waves (null spots in the response). You need "dead" space around the mic which is usually placed 1 meter away from the speaker. So unfortunately there's no really good way to do this. And because no speaker is perfectly flat there is no really good way to do this at all AFAIK. There are probably reference mics and speakers that are used as just that - references.

For DIY, I would just use the quietest place that has the least echo. So a bedroom with lots of drapes and thick carpeting and maybe you can hang some thick blankets from the crown molding to minimize reflections.

You might also use an "impulse response" style stimulus which sounds like a repeating pop but with all frequencies blended (so it has a thud to it but also an airy "ess" to it as opposed to a continuous discrete sweeping tone. That will be much less offensive to your neighbors and that is what the pros would use. I don't know if that can be done entirely in software or if the software is designed to work with certain hardware. I use a Quant Asylum QA400 which is hardware but the software is specifically designed to decode the impulse response that it generates.
 
Hi Pucho
Make a medium sized, well-padded (unreflective) box, place the mic @ one end and a spark gap across from the mic, take an FFT and you can get a Frequency response. If the box is large enough you could get some limited polar information. If you have a Calibrator you can get the sensitivity of the mic. This is the way to get the basic data of the DUT.
Duke
 
  I wrote about this recently, you can use a tube long enough so you can measure before the reflection from the bottom comes back, the speaker would be placed at one end and the mic close to it. You need like 10m to be able to measure down to 20Hz. You need the with of the tube to be small to measure high frequencies so they apear as planar wave fronts but you don't need to be much long, so you might want to have more than one different sized tube, a big one for LF and a small one for HF. If you match the with of the tube with the size of the driver it might be easy to mount them in.

Note than in a small tube is kind of hard to measure polar patterns as it's tricky to spin the mic around.

JS
 
Speaking as a former long time B&K guy and current AP employee, I would measure in as large a room as you have access to and use a software that makes use of a time selective method.  The basic idea is that the software removes room reflections so you are calculating the direct sound from the speaker only.  This isn't as good as measurements in a good anechoic chamber but i've gotten good results in bedroom sized rooms.  It is helpful to place the reference speaker and microphone as far away from reflections in the space as possible.  One way to improve this is to place absorptive material on the floor half way between the speaker and microphone to remove the most common first reflection.    Depending on your test setup, strategically placed absorption panels placed in other locations can improve the room performance as well.    Measuring low frequencies well will be dependent on your room and is generally difficult. 

There are a number of systems that will do this.  APx systems can do this (apologies for the obvious biases on my part) but there are other inexpensive (or free?) measurement softwares that may be able to do this as well.   

Squarewave's suggestion of using an impulse is a good idea and would be a direct method of the above calculation if you have a FFT analyzer with basic triggering capability and can use a rectangular window or no window (Hanning won't work well unless you are very lucky with where the impulse falls in the FFT window... and even then, not great).  The caveat here is that it isn't the easiest thing to create an ideal impulse.  Low frequency in particular - a popped balloon doesn't work well at low frequency (~<250Hz or so, in my experience).   

All of the above assumes that you are interested in frequency response primarily.  If you want to get distortion information, that is a whole other can of worms and difficult with direct speaker measurements.  The speaker is likely to have more distortion than the mic you are measuring. 

Something that hasn't been brought up is making a reference measurement with a (*good*) measurement microphone.  This is a relatively important step as it shouldn't be assumed that the reference speaker is flat.  It may be kinda flat but a better measurement would measure the speaker first to account for its response.
 
Excellent post by Tony, which finally admits to the world that real LF measurements are difficult, and there are no shortcuts.
Remember that your speaker cannot have a crossover.
 
I wonder if recording a small part of a sweep so the sound does not reach any close surface before the mic, silence to let the reflections to die and next part of the sweep and so on.

There could be problems with the signal going on and off so some window could be necessary  and some care putting the parts together.

if it work then it could be measured anywhere.
 
12afael said:
I wonder if recording a small part of a sweep so the sound does not reach any close surface before the mic, silence to let the reflections to die and next part of the sweep and so on.

There could be problems with the signal going on and off so some window could be necessary  and some care putting the parts together.

if it work then it could be measured anywhere.
That's exactly how windowed measurements are made; however there is a low frequency limit that is fixed by the length of the 1st reflection path. Complete frequency response assessment is done by grafting together several instances and different methods. For low frequencies, close miking (pressure mode) is preferred.
 
I remember going to Gotham USA to get Neumann microphones checked out. They had a padded tube about 3ft long that they put the microphone in to test. 
 
Gold said:
I remember going to Gotham USA to get Neumann microphones checked out. They had a padded tube about 3ft long that they put the microphone in to test.
Tubes and all sorts of acoustically damped enclosures have been used for quality control, putting the DUT in comparison with a reference mic or golden sample. Significant differences indicate possible malfunction; however, the resulting responses of such set-ups are not usable as reference graphs.
 
abbey road d enfer said:
putting the DUT in comparison with a reference mic or golden sample. Significant differences indicate possible malfunction; however, the resulting responses of such set-ups are not usable as reference graphs.

I do the same thing,
I use pink noise, reference microphone (same brand and model) close to the speaker take the Freq response chart , I use Smaart.
Do the same thing at the same exact position for the microphone under test.

Then compare the Freq Charts of both
 
I was just looking at the mics presented on REW website and to get a feel for if microphone SNR has an influence on typical (non-anechoic chamber) measurements.  A quick look at the specs for Dayton EMM-6 indicated it provided 70dB SNR, and the UMIK-1 was just a bit better.

Has anyone noticed mic SNR as a concern for testing ?

I recently got two ACO electret mics, each with a preamp, and a Cirrus CRL511E calibrator that allows each mic to be isolated from ambient, especially above a few hundred Hz, where the noise floor around the calibrator 1kHz fundamental test tone is about -110dB.  Those mic's supposedly have ruler flat response (the worst being 10-20k to 0.1dB), and I'm hoping their ~25yr age hasn't degraded response.
 
One of the REW reference papers on  impulse measurement schemes was looking at how low the noise floor needed to be to achieve optimum performance.  Some schemes needed at least 80dB.
 
trobbins said:
One of the REW reference papers on  impulse measurement schemes was looking at how low the noise floor needed to be to achieve optimum performance.  Some schemes needed at least 80dB.
That is true for architectural acoustics, where RT60 and waterfalls imply meaduring down to 80dB below stimulus for less than 1dB error, but for frequency response measurements, 40dB S/N is enough.
 

Latest posts

Back
Top