Microphone placement for measurements

[dima1stg]

I'm wondering if there is a common opinion on microphone placement for measurements.
For simplicity, let assume we're measuring at one point only being the main listening position and frequency range > 150hz.
Obvious choices I see are:
1. Next to the ear measuring speaker on the same side - the downside I see is that in reality the head prevents reflections from the other side to reach the mic thus altering the real picture. There is indeed a measurement difference between when I sit there and when I don't sit there.
2. At the nose point - didn't measure, but theoretically there should be the same effect from the head blocking reflections from the rear of the sphere
3. At the center of the head

And another thing - in the last 2 situations above, should the mic point to the corresponding speaker or should it point straight?

 

[skid00]

My suggestion is that you try all variations, and compare the graphs.

I think the largest consensus is to slowly sweep the 'head space' or the entire row of listener's head spaces while using pink noise.

I run a sweep for each ear, at the ear's location in space, with the mic pointing at the speaker under test. I can get distortion and impulse data from that.

I have a 1 foot square of memory foam that I pushed my mic thru, to block some of the reflections from the room, furniture, etc. This greatly diminished my distortion readings.

 

[dima1stg]

Let say I did all 3, all graphs are different, which one should I take as a base for correction?

 

[Sonnie Parker]

I've always used center of head when doing single measurements for my listening position. Varying the mic has never yielded any noticeable differences for me, but if you wanted to move it around, you can average the responses with REW. https://www.roomeqwizard.com/help/help_en-GB/html/graph_allspl.html

 

[dima1stg]

Did you point mic to speakers?

 

[Sonnie Parker]

Straight up at the ceiling, and I use the 90-degree calibration file for my UMIK-1.

I believe if you measure at the centermost location of your head at ear level and correct only for that response, it should be about as good as if you measure several locations and correct for the average. There will be some variation and imperfection either way you measure if you think about it... half a dozen to one, six to another. It's like measuring and correcting the seats on either side of you. Okay... well... you may have improved the other two seats ever so slightly, but you made the primary center seat ever so slightly worse or maybe even far worse than you'd prefer. You compromised to try to get the other two seats possibly better. Whereas if you just measured and corrected only the primary center seat, you make it near perfect, but the other two seats are still not so bad even though you didn't correct them.

 

[Anatoliy Gavrilov]

To measure, I place a microphone in front of the chair (listening point), the microphone is at the height of the tweeter, aimed at one speaker. Then I turn the microphone towards the second speaker - the measurements of the left and right speakers should match as much as possible. And measurements with a microphone near the left-right ear can be considered insignificant. Yes, and you can check this simply by moving the microphone stand left and right 10-12 cm from the listening point.

 

[DanDan]

My recommendations vary between Centre of the Head and the Two Ear positions.
As well as Measuring for it's own sake, I reckon it best to try to find the optimum (least bad) mic position.
It is very important to use the Mic on a Boom stand, NOT the table stand which daftly comes with various Measurement Mics. Also crucially important. Move the body away from the mic.

 

[Midiman]

In this context I have a REW beginner's question. In other measurement software it is quite common to measure several times at slighlty different positions and then average those measurements. This greatly cancels out (micro-) local effects for the benefit of making the general response more visible and relevant.
I found the option in REW to make several averaged measurements in a straight sequence but not to average different measurements taken independently from each other. Some time is at least needed to slightly move the mic some cm.
If this option was available I would consider using it also for the different mic locations described in this thread.
Am I missing something, did I oversee the feature?

 

[John Mulcahy]

There are a variety of averaging options on the All SPL graph.

 

[Midiman]

Ok, thanks, I'll check it out.