You've got the right idea.

Unless you have access to a genuine anechoic chamber, don't perform the tests by moving the sound source round the microphone, as the non-perfect characteristics of the room will have an effect. Instead, use as dead a room as you can find. Mount the microphone on a spindle rotatable about a vertical axis and take the measurements at different microphone angles with the sound source in a fixed position. Then repeat the test at several different sound source positions at a constant radial distance from the microphone, adjusting the zero degrees setting at the microphone appropriately. Average the results for each incident angle.

To display the results, either plot them by hand on polar graph paper or use a computer spreadsheet that will plot columns of data in polar format.

Of course, the whole exercise assumes that your sound source has a known frequency characteristic that you can use to correct the response curve you get from the microphone under test. The usual thing is to perform an initial frequency calibration pass using a quality measurement microphone, as this type of microphone is adjusted to be flat to (say) 0.5dB over the audio frequency range. Again, this calibration test needs to be repeated with the sound source at the various points around the circle centred on the microphone and the results for each angular position averaged. Make sure you are either out of the room or else positioned at a constant point for each measurement, as your body will have an effect on the room reflection characteristics.

Let us know how you get on.

Thanks for the great response. I don't speak audioese (yet) so let me see if I can rephrase something you wrote to make sure I am getting it:

>>Of course, the whole exercise assumes that your sound source has a known frequency characteristic that you can use to correct the response curve you get from the microphone under test. The usual thing is to perform an initial frequency calibration pass using a quality measurement microphone, as this type of microphone is adjusted to be flat to (say) 0.5dB over the audio frequency range.<<

=The tested mic may not respond accurately to the test frequency because of its response curve, so you should measure the test frequency with a different mic that has a flat response curve. Would this problem be avoided by using a pure sine wave as a test frequency, and testing multiple frequencies and averaging the results? (I don't have a flat response mic, or an anechoic chamber for that matter.... :tongue:)

It would be nice if something like that would give acceptable and repeatable results, but unfortunately, it's not the case. The point of the calibration exercise is that any loudspeaker that you use as a sound source for testing your microphone is going to have a frequency response curve that could well be even more bumpy than that of the microphone under test. In addition, the received sound at the microphone position will also be affected by the characteristics of the room in which you are doing the testing. The result of this is that you will have no way of knowing whether the ups and downs in the curve that you measure is due to the frequency response of the test microphone, the sound source or the room.

What you would be doing by running a calibration using a measurement microphone and then recording your test microphone is largely taking the characteristics of both the sound source and the room out of the equation. You would effectively be determining the difference between the response curves of the two microphones, and if one of those is nearly flat, the result could be taken as the response of the test microphone. By way of example, the [[url=http://[/URL]="http://www.cross-sp…"]Dayton EMM6[/]="http://www.cross-sp…"]Dayton EMM6[/] is a reasonably low-cost measurement microphone that could be used for these tests.

The standard test procedure is indeed to use sinewaves at a series of frequencies (e.g. 3 steps per octave), and in an anechoic chamber, simply rotating the test microphone would give very repeatable results. However, when you are looking at differences between microphones, even small effects of room reflections can make a big difference. These can be largely compensated for by (a) the calibration pre-pass and (b) averaging the measurements from many different positions of the sound source in the room while keeping the incident angle on the microphone constant. In procedural terms, it's much easier to run through the range of frequencies and angles of incidence at one sound source position and then change that position and repeat the runs than it is to be constantly moving the sound source.

Hmmm... I have a few more questions, if you have the patience:

Would you mind detailing the procedure for using the response of the measurement microphone to alter the data from the test microphone?

The polar pattern charts I've looked at show a range of values from 0dB to -25dB. I'm assuming that 0dB means freescale--does that mean that when running your test from 0 degrees, you increase the volume of the test frequency to the point where the input is almost clipping and then go from there? Or is the 0dB just a relative value?

(from Wikipedia) Polar patterns..."represent the [="http://en.wikipedia.org/wiki/Locus_(mathematics)"]locus[/]="http://en.wikipedia…"]locus[/] of points that produce the same signal level output in the microphone if a given [[url=http://="http://en.wikipedia…"]sound pressure level[/]="http://en.wikipedia…"]sound pressure level[/] (SPL) is generated from that point." Is the polar pattern a representation of physical points, or of dB levels?

Thank you so much for your helpful replies!

[[url=http://[/URL]="http://www.youtube…"]Microphone Test - Figure-8 Polar Pattern Test - YouTube[/]="http://www.youtube…"]Microphone Test - Figure-8 Polar Pattern Test - YouTube[/]

This should help with ideas.

Thanks, what was actually more helpful was [[url=http://[/URL]="T=56">http://www.youtube…"]this video[/]="T=56">http://www.youtube…"]this video[/] that yours led me to. Check out how the mic is mounted on rails to facilitate different testing positions! (and the super-cool anechoic chamber!!)

ridermic, post: 407874 wrote: Thanks, what was actually more helpful was [="http://www.youtube.com/watch?v=rYWnSuAxato&feature=player_detailpageT=56"]this video[/]="T=56">http://www.youtube…"]this video[/] that yours led me to. Check out how the mic is mounted on rails to facilitate different testing positions! (and the super-cool anechoic chamber!!)

Now you have found that Shure video, go back and re-read the responses you have had so far in this thread.

ridermic, post: 407874 wrote: Hmmm... I have a few more questions, if you have the patience:

Would you mind detailing the procedure for using the response of the measurement microphone to alter the data from the test microphone?

The polar pattern charts I've looked at show a range of values from 0dB to -25dB. I'm assuming that 0dB means freescale--does that mean that when running your test from 0 degrees, you increase the volume of the test frequency to the point where the input is almost clipping and then go from there? Or is the 0dB just a relative value?

(from Wikipedia) Polar patterns..."represent the [[url=http://="http://en.wikipedia…"]locus[/]="http://en.wikipedia…"]locus[/] of points that produce the same signal level output in the microphone if a given [[url=http://[/URL]="http://en.wikipedia…"]sound pressure level[/]="http://en.wikipedia…"]sound pressure level[/] (SPL) is generated from that point." Is the polar pattern a representation of physical points, or of dB levels?

There's not a lot more to say about running a system calibration pass before doing a test pass - it's standard engineering procedure to reduce uncertainties in the measurement process.

The polar pattern represents attenuation plotted against physical angle. The radial scale used is simply zeroed (0dB) to whatever output you get from the test microphone at 0 degrees incidence and (say) 1KHz at an SPL that does not overload the microphone but gives you enough signal to measure at the lowest response directions. These may be 60-70dB below the maximum, and you don't want that signal lost in the noise of the microphone and pre-amp.

We can't write your assignment for you.

Thanks again, I'll stop pestering you now!