REW-Moving Microphone Method Help

[bigjacko]

Below are some nice videos which show it

So I would assume they point the microphone towards the speaker at a range around the listening position and a circle a bit bigger than head? If I don't have the foam tip in microphone can I just cut a bit of foam and put it on or it does not matter much?

 

[Pio2001]

When doing the measurement how do we move the microphone? Do we point it towards speaker and draw circle towards speaker? How far to the speaker and how big should the circle be? I saw the video posted here but cant really tell how they did it. Thanks for the help.

Hi bigjacko,
The microphone should always aim at the ceiling.
It should move inside all the listening zone. For a one-seat listening position, this zone is 60 x 60 cm, minimum horizontally, and at least 30 cm vertically, centered at the listening position.
The microphone should move inside the whole zone, and spend an equal amount of time everywhere in the zone. You can for example zig-zag horizontally, then zig-zag vertically.

 

[Pio2001]

The more interesting difference I see is in the results when changing the length.

Pink periodic noise (pink PN) is a pink noise that has been tweaked so as to have a flat frequency response (from an RTA point of view) for every period.
The "sequence length" of the pink PN generator must be the same as the "FFT length" of the analyzer. It allows the analyzer to get a flat signal for every acquisition (granted that the analyser works at the same sample rate as the generator).
At a sample rate of 48 kHz, a period of 64k samples is a bit more than 1 second.

The longer the length, the more resolution you get in the frequency response measurement. 64k is enough.

This has the consequence of adding more peak transient energy in the lower half of the response.

It doesn't. The pink noise just takes longer to repeat itself.

 

[Lorenzo74]

Hello. I have used REW some but saw that some people use a moving microphone method to get a quick spacial average. The video I watched made it look quick and easy. I was able to to fumble my way through partially, but the graph doesn’t look like frequency response. Anyone know the steps to get it to show frequency response?

just ask @hardisj !

 

[ernestcarl]

It doesn't. The pink noise just takes longer to repeat itself.

Thanks! I've read that info from the manual and was wondering about it again. You are right. The transient peaks will not rise much above a certain level.

But by increasing the resolution, it is essentially adding and spreading out more of that transient peak energy in the lower half of the response. This will register in the microphone and show on the graphs clearer as well -- and surely this is going to work or "stress" the driver harder esp. the higher SPL you go.

When one is doing slow sine sweeps, the driver will noticeably show compression earlier. When playing periodic pink noise at very high resolution, more of the peaks get through fine without obviously showing any compression.

Pink periodic is not really a good "stress test" anyway -- and I would probably go with M-Noise instead as that one has been intently designed to be used as a performance test. I haven't played with it myself, though.

You can read more about it here:

https://m-noise.org/

You can also download the wav file with instructions if anyone's interested.

M-Noise

 

[ernestcarl]

The microphone should always aim at the ceiling.

I usually just use the 0 deg. calibration file myself and point it in front. Doesn't have to be directly in front of the speaker. Charles Sprinkle mentioned that he really doesn't care that much where it's pointed at -- as he seems to be only interested in correcting up to 700Hz max. So it depends on what you're trying to do with the measurement you're going to be taking.

 

[jlo]

Pink periodic is not really a good "stress test" anyway -- and I would probably go with M-Noise instead as that one has been intently designed to be used as a performance test.

M-noise was developped to check power capabilities of loudspeakers not to measure frequency response !

 

[jlo]

I usually just use the 0 deg. calibration file myself and point it in front. Doesn't have to be directly in front of the speaker. Charles Sprinkle mentioned that he really doesn't care that much where it's pointed at -- as he seems to be only interested in correcting up to 700Hz max. So it depends on what you're trying to do with the measurement you're going to be taking.

I really prefer to point mic to ceiling and use 90° calibration file : the sound field is nearer random incidence than direct field. When using 90° cal, you can point your mic to any direction while you move it, it is easier than keeping it directed to the loudspeaker, especially when you do multichannel measurements. And also when you do ie L and R loudspeakers at the same time (to check low frequencies mono summation) or multiple surrounds at the same time.
In fact, the ideal would be to use a random incidence calibration, instead of a 90° : random incidence correction is somewhere between 0 and 90° calibration (but nearer to 90°).

 

[thewas]

The microphone should always aim at the ceiling.

This imho isn't a fixed rule, some can point it to the floor and some to the sides.
I personally prefer the later (same like Charles Sprinkle in the video) for my stereo pair measurements as this way I can stand further away from the mic on the side and also consider more the side reflections (I do 2 measurements from both the left and right side of the LP), although the differences are rather tiny and more academic and for the internal satisfaction.

 

[vavan]

some can point it to the floor and some to the sides.
I personally prefer the later (same like Charles Sprinkle in the video)

with 90° calibration file?

 

[thewas]

with 90° calibration file?

Of course.

 

[dasdoing]

it makes no sense to me to point at the ceiling. my mic came with on axis calibration file so I aim directly at the tweeter. but my room is reflection treated, it might make more sense to aim in between the reflection point and the speaker in a reflective room

 

[Pio2001]

At the listening position, we measure both the contribution of the direct sound coming from the speakers, and of the diffuse sound coming from the room.
If we point the microphone towards the speakers and use the 0° calibration, the direct sound is properly calibrated, but not the diffuse sound, that is measured with less high frequencies than it actually has.
If we point the microphone sideways and use the 90° calibration, the direct sound is still correct, but this time the diffuse sound is measured with more high frequencies than it actually has.
We can't measure both the direct and diffuse sound with a complete microphone calibration.

 

[dasdoing]

If we point the microphone towards the speakers and use the 0° calibration, the direct sound is properly calibrated, but not the diffuse sound, that is measured with less high frequencies than it actually has.

where does the loudest and least delayed reflection come from though? from pretty near the speaker (reflection of the nearest side-wall)

 

[ernestcarl]

M-noise was developped to check power capabilities of loudspeakers not to measure frequency response !

Nothing to disagree there... But whoever said we should use it to measure the frequency response? I don't remember saying that.

 

[ernestcarl]

I really prefer to point mic to ceiling and use 90° calibration file : the sound field is nearer random incidence than direct field. When using 90° cal, you can point your mic to any direction while you move it, it is easier than keeping it directed to the loudspeaker, especially when you do multichannel measurements. And also when you do ie L and R loudspeakers at the same time (to check low frequencies mono summation) or multiple surrounds at the same time.

I personally find it easier to just point it facing the front -- so like I said, I don't point it directly 0 deg. on-axis the speakers -- as long as it's generally along the front horizontal field is good enough for my own needs. I understand that you want to be as accurate as possible. I don't use MMM primarily for speaker EQ, but rather just room EQ -- for speaker EQ above 1 kHz, I look at and apply EQ to the whole family of horizontal and vertical curves which I have measured in the pseudo-anechoic nearfield (for the coaxial S8, at least).

I have opted not to apply any 'fine' speaker correction to the rest of my speakers but just simple HF shelving.

 

[dasdoing]

I personally find it easier to just point it facing the front -- so like I said, I don't point it directly 0 deg. on-axis the speakers -- as long as it's generally along the front horizontal field is good enough for my own needs. I understand that you want to be as accurate as possible. I don't use MMM primarily for speaker EQ, but rather just room EQ -- for speaker EQ above 1 kHz, I look at and apply EQ to the whole family of horizontal and vertical curves which I have measured in the pseudo-anechoic nearfield (for the coaxial S8, at least).

I have opted not to apply any 'fine' speaker correction to the rest of my speakers but just simple HF shelving.

you eq your room below 1k, and speaker only above 1k?

 

[ernestcarl]

you eq your room below 1k, and speaker only above 1k?

Depends on which speaker system it is we are refering to... for my KH120s' there's almost zero EQ above 500Hz and a -1dB HF shelving at 2 kHz.

*-1dB HFS from monitor's own internal parametric switches.

Left-Right symmetry is good with only some minor differences.

Shared Output EQ:

If you mean to say with the Sceptre S8s, my 'fine' speaker EQ starts above 1 kHz. I do have EQ applied below and around the transition zone -- so 1 kHz below, yes. However, my EQ around the transition is unnecessary, IMO. My primary concern is in the regions below 400 or 500Hz and above 1 kHz.

 

[ppataki]

I have recently tried the moving mic method in my room and the results look totally different compared to a sweep measurement at the listening position:

It seems that below 200Hz the resolution of the measurement is like 1/1 octave

I used the below configuration:

I tried 64k also, the same result

I can hear the dips shown by the sweep method so I am pretty sure the MMM results are not correct

Am I doing anything wrong or the MMM is not really to be used for measuring below 200Hz?

 

[markus]

I have recently tried the moving mic method in my room and the results look totally different compared to a sweep measurement at the listening position:

View attachment 101665

It seems that below 200Hz the resolution of the measurement is like 1/1 octave

I used the below configuration:

View attachment 101666

I tried 64k also, the same result

I can hear the dips shown by the sweep method so I am pretty sure the MMM results are not correct

Am I doing anything wrong or the MMM is not really to be used for measuring below 200Hz?

Are the results repeatable? How large is the area you're covering with the mic? Might just be low frequency noise.