Gating a pseudo-anechoic measurement

[tuga]

I've plonked one speaker in the middle of the room and made a few 100ms sweep measurements to determine the mic height for flattest response.
Fuzzmeasure gives me the option to define the windowing parameters (start and duration).
How do I go about it?

 

[jhaider]

What are you trying to measure? LF or MF/HF?

If the latter, given your data start with going to the impulse response and set your curtains just before the first downwards spike (3.5ms or so) and before the first spike after the impulse (about 5.3ms).

If you're trying to measure LF, skip the windowing, put the speaker in its normal spot, move the microphone to 5 or 6 points around your seat, and average the measurements.

 

[tuga]

What are you trying to measure? LF or MF/HF?

If the latter, given your data start with going to the impulse response and set your curtains just before the first downwards spike (3.5ms or so) and before the first spike after the impulse (about 5.3ms).

If you're trying to measure LF, skip the windowing, put the speaker in its normal spot, move the microphone to 5 or 6 points around your seat, and average the measurements.

I am trying to measure the speaker response before the room kicks in, above 300 or 400 or 500Hz (?).

 

[Head_Unit]

I am trying to measure the speaker response before the room kicks in, above 300 or 400 or 500Hz (?).

Indoors, I did this stuff for years and don't trust looking at the spikes. Ijust don't feel there's a substitute for a measuring tape and a calculator.

I used a custom ceiling tile to mount coaxes etc into, microphone at 1m, the ceiling was let's call it 3m. This meant the direct sound hit after 1/343=0.00292S, and the floor bounce hit the mic at (3+(3-1))/343=5/343=0.0146S, so a clear window of 0.0117S. That meant a valid data point down to 1/0.117=85 Hz.
NOTE: that does NOT necessarily mean your data is good down to 85 Hz per MLSSA's Doug Rife. Maybe with a gated sine, but something with an FFT means a good data point every 85 Hz: 85, 170, 255, 340...not a lot of resolution down low.

If you're trying to do this in a house with 8'=2.43m ceilings, with the speaker let's call it 1.2m up and the mic at 1m, the direct sound is still 0.00292S. The floor and ceiling bounces oh I forgot, you need geometry class. The first bounce is along a path which is two hypotenuses of two triangles of sides 0.5 and 1.2 meters = 1.3x2 =2.6m, or 0.00671S. Subtract the direct 0.00292 for 1/0.00378 = valid data points only at 265, 530, 795, 1060...you can see this is not really too good.
AND there is an assumption that you have totally cleared out a kind of ellipsoid all around the measurement area.
--> If you're trying to measure just the speaker, better outside on a large ladder (maybe facing sideways to try and minimize reflections off the ladder.

Anyway I hope you get the idea...if you want help with the calculations, quote me and reply.

P.S. You also have to consider measuring distance versus size of the speaker. If for instance the speaker is a multiwoofer tower, 1 meter is not far enough for all those drivers to really integrate. 1m is really only OK for coaxes and maybe very compact two-ways.

 

[DDF]

I wrote an excel utility you can try
https://www.diyaudio.com/forums/multi-way/284567-measurement-utility-micheight.html

 

[tuga]

Indoors, I did this stuff for years and don't trust looking at the spikes. Ijust don't feel there's a substitute for a measuring tape and a calculator.

I used a custom ceiling tile to mount coaxes etc into, microphone at 1m, the ceiling was let's call it 3m. This meant the direct sound hit after 1/343=0.00292S, and the floor bounce hit the mic at (3+(3-1))/343=5/343=0.0146S, so a clear window of 0.0117S. That meant a valid data point down to 1/0.117=85 Hz.
NOTE: that does NOT necessarily mean your data is good down to 85 Hz per MLSSA's Doug Rife. Maybe with a gated sine, but something with an FFT means a good data point every 85 Hz: 85, 170, 255, 340...not a lot of resolution down low.

If you're trying to do this in a house with 8'=2.43m ceilings, with the speaker let's call it 1.2m up and the mic at 1m, the direct sound is still 0.00292S. The floor and ceiling bounces oh I forgot, you need geometry class. The first bounce is along a path which is two hypotenuses of two triangles of sides 0.5 and 1.2 meters = 1.3x2 =2.6m, or 0.00671S. Subtract the direct 0.00292 for 1/0.00378 = valid data points only at 265, 530, 795, 1060...you can see this is not really too good.
AND there is an assumption that you have totally cleared out a kind of ellipsoid all around the measurement area.
--> If you're trying to measure just the speaker, better outside on a large ladder (maybe facing sideways to try and minimize reflections off the ladder.

Anyway I hope you get the idea...if you want help with the calculations, quote me and reply.

P.S. You also have to consider measuring distance versus size of the speaker. If for instance the speaker is a multiwoofer tower, 1 meter is not far enough for all those drivers to really integrate. 1m is really only OK for coaxes and maybe very compact two-ways.

Thanks for detailed reply.
I should have asked the questions before I made the measurements.

The goal is to measure the speaker response in above the transition region in order to determine if there are any peaks and dips worth addressing and if the response is being affected by reflections above 400-500Hz (room response shows a broad peak at around 600Hz).

I am using the evaluation copy of the software and have no access to the manual.

I have used the shortest possible sweep of 100ms, perhaps mistakenly, in the hope that this would keep reflections at bay.
I wonder if this is the cause for the unexpected roll-off below 100Hz.

The mic and optimal axis are at around 0.88cm from the floor. Room height is around 2.45m and the sofa was the nearest object at around 1.5m from the axis source point, closer to the woofer.
The software mentions a distance of 1.26m which I presume to be that of the mic to source.
I piled a sofa cushion and a soft pillow on the floor between the speaker and the mic to reduce the effect of floor bounce which according to this calculator is at around 181Hz and can be verified by the measurement.