Influence of pressure-field, diffuse-field, and free-field microphones on microphone orientation

[Blumlein 88]

True. Almost no baffle step with this small size.

I don't think such generalizations are valid.
In the end its all just EQ what defines the class. Pressure field basically means flat response when flush mounted, no baffle step to compensate.

Like the pzm mics. I've used something like that for edge of the stage recordings. Separate with a 4x6 inch wood block. Strap SDCs on each side where it meets the floor. Lay it down and record. For unamplified acoustic groups works great.

 

[OCA]

But how does it create "significant" phase shifts with practical use?

View attachment 386869 View attachment 386868 View attachment 386867

View attachment 386872 View attachment 386871 View attachment 386870

With Earthworks M series 90 degrees calibration data and assuming it's minimum phase (probably not), there's -15 degrees shift at 7kHz, -26 degrees shift at 20kHz. It's quite significant if you are aligning a custom built speaker's tweeter with the mid-driver for example.

 

[ernestcarl]

With Earthworks M series 90 degrees calibration data and assuming it's minimum phase (probably not), there's -15 degrees shift at 7kHz, -26 degrees shift at 20kHz. It's quite significant if you are aligning a custom built speaker's tweeter with the mid-driver for example.

View attachment 386874

I honestly don't really get the significant problem/issue being solved here... sorry. Also haven't heard of any one else raising this missing phase information as a significant issue with supplied mic calibration files.

 

[NTK]

With Earthworks M series 90 degrees calibration data and assuming it's minimum phase (probably not), there's -15 degrees shift at 7kHz, -26 degrees shift at 20kHz. It's quite significant if you are aligning a custom built speaker's tweeter with the mid-driver for example.

View attachment 386874

At 7 kHz and 20 °C, the wavelength is 343.0/7000 = 0.049 m. A 15° phase shift equates to a displacement of the wave of 0.049 × 15/360 = 0.002 m = 2 mm.
At 20 kHz and 20 °C, the wavelength is 343.0/20000 = 0.017 m. A 26° phase shift equates to a displacement of the wave of 0.017 × 26/360 = 0.001 m = 1 mm.

 

[OCA]

We're talking about 90 degrees calibration file deviations and using a minimum phase version of the calibration file ensures everything lines up better without adding any extra hassle. In REW, you still upload one calibration file and only once. I don't understand the passion to chase inaccuracy.

Sure, for a USB mic with clock issues and other quirks, this kind of precision might not be as important, but with a mic like Earthworks, it’s something worth paying attention to. Ask a speaker designer to move the tweeter 2mm back because it's insignificant to you and you will get the idea.

 

[Keith_W]

I received this reply from Earthworks.

 

[KSTR]

Sure, for a USB mic with clock issues and other quirks, this kind of precision might not be as important, but with a mic like Earthworks, it’s something worth paying attention to. Ask a speaker designer to move the tweeter 2mm back because it's insignificant to you and you will get the idea.

With Earthworks M series 90 degrees calibration data and assuming it's minimum phase (probably not), there's -15 degrees shift at 7kHz, -26 degrees shift at 20kHz. It's quite significant if you are aligning a custom built speaker's tweeter with the mid-driver for example.

No, it's irrelevant for alignment as described in previous posts because both driver's responses see the exact same error. Only if you measured one driver with correction applied and the other without you'd obviously have problems... but even then only if the correction were minimum phase which typically it is not.

 

[Miley_B]

Odd that Earthworks website states that the M23/M23R is free-field but no mention of mic type for the M30.

All of my mic calibration files are basic .txt files, I believe specifically tab delimited ASCII text format. I can't see how the .txt file could be phase specific, that would surely be down to the implementation of the file within software?

I've never previously considered how programs such as Smaart or REW would implement a mic calibration file. Do they essentially apply an EQ to the measurement which would have a varying phase implication depending on the algorithm type, or is it maybe just shifting the line on the graphic display by the numbers within the file?

 

[GXAlan]

My understanding was that for surround sound, you need to point at 90 degrees (ceiling) so that the rear speakers are captured appropriately (and less likely to be blocked by the whole body of the microphone) and so that you have accurate distance calculators. Of course, does that alter the in-ceiling speakers?

For stereo, pointing toward the front was better because you don’t worry about the the impact of the body of the microphone.

 

[KSTR]

All of my mic calibration files are basic .txt files, I believe specifically tab delimited ASCII text format. I can't see how the .txt file could be phase specific, that would surely be down to the implementation of the file within software?

I've never previously considered how programs such as Smaart or REW would implement a mic calibration file. Do they essentially apply an EQ to the measurement which would have a varying phase implication depending on the algorithm type, or is it maybe just shifting the line on the graphic display by the numbers within the file?

Yes, for Q1 and Q3 (at least for REW, that is).

 

[Keith_W]

Odd that Earthworks website states that the M23/M23R is free-field but no mention of mic type for the M30.

All of my mic calibration files are basic .txt files, I believe specifically tab delimited ASCII text format. I can't see how the .txt file could be phase specific, that would surely be down to the implementation of the file within software?

I've never previously considered how programs such as Smaart or REW would implement a mic calibration file. Do they essentially apply an EQ to the measurement which would have a varying phase implication depending on the algorithm type, or is it maybe just shifting the line on the graphic display by the numbers within the file?

Most mic files are supplied in non-inverted format. REW does an automatic minimum phase inversion and applies that to all measurements.

If the mic files are supplied in inverted format, they need to be inverted before you can use it with REW. Some software (like Acourate) forces you to perform your own manual inversion so you have to load the mic file and look at it, then decide if you want to invert or not.

 

[OCA]

A M23R is a truly omnidirectional mic up to 1kHz then starts getting into cardioid territory Good enough to comfortably cover room transient beyond where digital filters are meaningless anyway. I wonder if rotating the mic only over its long axis would make any difference.

 

[SIY]

then starts getting into cardioid territory

Sort of- the rear null only hits -8dB or so at 16k and only -4dB at 10k.

 

[NTK]

We're talking about 90 degrees calibration file deviations and using a minimum phase version of the calibration file ensures everything lines up better without adding any extra hassle. In REW, you still upload one calibration file and only once. I don't understand the passion to chase inaccuracy.

Sure, for a USB mic with clock issues and other quirks, this kind of precision might not be as important, but with a mic like Earthworks, it’s something worth paying attention to. Ask a speaker designer to move the tweeter 2mm back because it's insignificant to you and you will get the idea.

Let's look at it with some numbers.

If we have a 2-way speakers, "1" is the tweeter and "2" is the woofer. Their center-to-center distance is Δ and let's assume it to be 0.15 m. "M" is the mic (or listener) and it is 1 m away. "y" represents the vertical position of the mic, and y = 0 means it is level with the tweeter. "d1 - d2" is the difference between the distance from "m" to"1" and the distance from "m" to 2. Below is the plot of "d1 - d2" versus "y", normalized to zero when y = 0. I'll leave the interpretation to you as homework.

Additional points:

1. The differences in distances to the drivers are only important (or "problematic") at the cross-over regions, when both drivers are radiating sounds at comparable levels. What are the cross-over frequencies? And what are the phase shifts of the mic at those frequencies? (We may separately discuss line arrays.)
2. Where exactly is the center of a driver? Do you know where it is, to mm precision, in the direction of the driver axis? Hint: The location of the acoustic center is frequency dependent.

 

[OCA]

Let's look at it with some numbers.
View attachment 386975

If we have a 2-way speakers, "1" is the tweeter and "2" is the woofer. Their center-to-center distance is Δ and let's assume it to be 0.15 m. "M" is the mic (or listener) and it is 1 m away. "y" represents the vertical position of the mic, and y = 0 means it is level with the tweeter. "d1 - d2" is the difference between the distance from "m" to"1" and the distance from "m" to 2. Below is the plot of "d1 - d2" versus "y", normalized to zero when y = 0. I'll leave the interpretation to you as homework.
View attachment 386977
Additional points:

1. The differences in distances to the drivers are only important (or "problematic") at the cross-over regions, when both drivers are radiating sounds at comparable levels. What are the cross-over frequencies? And what are the phase shifts of the mic at those frequencies? (We may separately discuss line arrays.)
2. Where exactly is the center of a driver? Do you know where it is, to mm precision, in the direction of the driver axis? Hint: The location of the acoustic center is frequency dependent.

Thanks for the Pythagorean maths overview, having solved Navier-Stokes equations on a regular basis in the uni, guess I can cope with it

It still doesn't explain why would more accuracy without any extra work load hurt. Why not remove as much excess phase as possible from the mic calibration effect by using its minimum phase version?

Here's my floorstanders' crossover/port phase correction with time inverted allpass filters at XO frequencies 2700Hz, 260Hz and port frequency 41Hz:

and here the same correction with same speaker's excess phase response including min phase mic calibration file effect:

I like what I see in rePhase better in the latter.

 

[OCA]

Thanks for the Pythagorean maths overview, having solved Navier-Stokes equations on a regular basis in the uni, guess I can cope with it

It still doesn't explain why would more accuracy without any extra work load hurt. Why not remove as much excess phase as possible from the mic calibration effect by using its minimum phase version?

Here's my floorstanders' crossover/port phase correction with time inverted allpass filters at XO frequencies 2700Hz, 260Hz and port frequency 41Hz:
View attachment 386985

and here the same correction with same speaker's excess phase response including min phase mic calibration file effect:

View attachment 386984

I like what I see in rePhase better in the latter.

I thought would be more helpful if I posted the original EP response before correction

:

 

[OCA]

Do you know where it is, to mm precision, in the direction of the driver axis? Hint: The location of the acoustic center is frequency dependent.

What I knew was that it's half a woofer diameter in front of the cone, I didn't know it was frequency dependent. Thank you for that. Is there a simple formula?

 

[NTK]

What I knew was that it's half a woofer diameter in front of the cone, I didn't know it was frequency dependent. Thank you for that. Is there a simple formula?

Don't you find it interesting that you said the acoustic center of a woofer is in front of the cone? Have you seen any physically "time aligned" loudspeakers that have the tweeter protrudes in front of the woofer? (Let's say your number of half woofer diameter is correct, the acoustic center of a 12" woofer will be 6" in front.)

The concept of "acoustic center", at first glance, sounds deceptively simple. But it is not. For an omni-directional source, the location of the center is obvious. Once the directivity deviates from onmi, where is it? I don't think there is a widely agreed definition.

I looked a little bit, and found 2 papers for your reading pleasure (I haven't read them myself).

EXTENDING THE LOW FREQUENCY ACOUSTIC CENTRE CONCEPT TO DIRECTIONAL LOUDSPEAKERS | Institute of Acoustics

www.ioa.org.uk

https://physics.byu.edu/docs/publication/9867

 

[gnarly]

Here's my floorstanders' crossover/port phase correction with time inverted allpass filters at XO frequencies 2700Hz, 260Hz and port frequency 41Hz:
View attachment 386985

and here the same correction with same speaker's excess phase response including min phase mic calibration file effect:

View attachment 386984

I like what I see in rePhase better in the latter.

Hi, Why do you like the latter better?

 

[KSTR]

I like what I see in rePhase better in the latter.

As far as the noisy plot allows to decipher, they look the same to me, other than the microscopic differences at the very top end -- irrelevant, I would think.

What I knew was that it's half a woofer diameter in front of the cone, I didn't know it was frequency dependent. Thank you for that. Is there a simple formula?

It's frequency-dependent when the cone/dome is not perfectly rigid. Some drivers even purposely decouple the outer parts of the cone to reduce nasty breakup.