[SpHEAR-devel] latest musings (PCB, new Octathingy, calibrationand more)

[Fernando Lopez-Lezcano]

On 06/26/2018 07:23 PM, Marc Lavallée wrote:
> Hi Umashankar (and all).
...
> Because building an ambisonics microphone is a lot of work (even the
> ones from the SpHEAR project),

Yup, it is a LOT of work... almost not worth it, ha ha :-)

>I got a cheap Twirling720 Lite; it looks
> more a toy than a proper microphone, but I like its form factor, its
> built-in USB interface, and maybe its a good enough device to hack in
> order to learn the subtle art of measurement and calibration. I just
> hope there's no processing between the capsules and the USB interface
> (something to discover).

Keep us updated...

> Fernando, I was asking about the article I found because I like the idea
> of using a reference microphone to calibrate another one (with a
> transfer function). So i wondered if a set of individually calibrated
> capsules (using the same reference microphone) could be used without
> further measurements to make a calibration profile based on a
> theoretical ambisonics microphone, knowing the distance between the
> capsules and the center of the array; I guess it would fail because the
> capsules holder (and other capsules) are not acoustically transparent.

Hmmm, I guess it would be possible and better than nothing, but I'm not 
sure how practical. After all, if you can measuring the individual 
capsules you might as well measure the finished microphone and be done 
with it :-) And as you point out the acoustic shadows of the capsule 
holder, other capsules, the mount, etc, all affect the response of the 
microphone and would be difficult (but not impossible) to simulate. 
Aaron (Heller) has some matlab/octave code to simulate arbitrary capsule 
arrays, but the simulation of the Octathingy, for example, and the real 
thing as fabricated are different...

Senheisser does that (AFAIK) with their Ambeo microphone, they do not 
provide individualized calibrations, I presume they have the resources 
to match the capsules (and/or not having a perfect result is good enough)...

For a tetrahedral design you can get away with 8 measurements equally 
spaced around the microphone in the horizontal plane, which is not too bad.

> The devil is in the details (n'est-ce pas?).

Indeed! A big part of the calibration is getting good measurements, and 
things are not as easy as you would think (but not impossible). I'm 
slowly learning...

-- Fernando