How Is a Mic's Freq. Response Tested?

[squibble94]

Hey guys. I'm just curious as to how a mic's frequency response is determined. Anyone know?

 

[sjjohnston]

Several possibilities:

- Right Way: Create known-flat acoustic signal by playing pink noise (or white noise, or sweeps) through a carefully calibrated speaker system in an anechoic chamber; put the mic in front of it; hook it up through a carefully calibrated preamp and AD converter to a spectrum analyzer (or, if you use sweeps, you could use a VU meter, which is simpler in some ways but perhaps less accurate).

- More Cludgy Way: If your acoustic source (or post-mic path) isn't quite so perfect but you do have another mic with flat (or at least known) frequency response, put each of them in front of a pretty-flat speaker in a pretty-uncolored room, analyze both of them, then calculate the frequency response of the mic you're measuring using the known response of the other mic and the measured differences.

- Marketing Way: Just copy the chart from somewhere else.

 

[moresound]

Several possibilities:

- Right Way: Create known-flat acoustic signal by playing pink noise (or white noise, or sweeps) through a carefully calibrated speaker system in an anechoic chamber; put the mic in front of it; hook it up through a carefully calibrated preamp and AD converter to a spectrum analyzer (or, if you use sweeps, you could use a VU meter, which is simpler in some ways but perhaps less accurate).

- More Cludgy Way: If your acoustic source (or post-mic path) isn't quite so perfect but you do have another mic with flat (or at least known) frequency response, put each of them in front of a pretty-flat speaker in a pretty-uncolored room, analyze both of them, then calculate the frequency response of the mic you're measuring using the known response of the other mic and the measured differences.

- Marketing Way: Just copy the chart from somewhere else.

This is the way I've had mine done at Earthworks (about 5 minutes away from me) every time I go a purchase microphones they shoot them for me and give me a copy of the readout.

 

[sjjohnston]

Not too surprising: if anyone does it the right way, it presumably should be Earthworks, since they make a fair number of the mics people use to measure the frequency response of everything else.

 

[squibble94]

Hmm interesting. That's how I thought they did it, but it seems like too many variables come into play. Big margin of error there.

 

[brassplyer]

Not too surprising: if anyone does it the right way, it presumably should be Earthworks, since they make a fair number of the mics people use to measure the frequency response of everything else.

Curious how close the chart that comes with the mic typically is to their measured response?

 

[Shadow_7]

Hmm interesting. That's how I thought they did it, but it seems like too many variables come into play. Big margin of error there.

Lots of opportunities to do it wrong. The speakers have EQ traits. The preamps have EQ traits. The room has EQ traits. And then there's the mic that you're actually testing. But there's probably a fair amount of calibration of each of those elements for the higher end brand / models (if only in theory). And then there's a fair number of charts that were obviously drawn, not measured. Or just copied and they never even bothered to white out the model of the "other" mic from the chart. And even measured, it's likely that only one in every couple thousand mics are actually tested before leaving the factory. Depending on the brand and model of course. Not that a measured mic is accurate by the time it crosses a continent, an ocean, and other environmentals of getting from point A to point B, where you'll actually use the mic.

 

[RRuskin]

Back in the day, one method a flattening out the printed specs was to speed up the pen speed of whatever was writing the graph.