Home Theater Forum and Systems banner

1 - 15 of 15 Posts

·
Registered
Joined
·
22 Posts
Discussion Starter #1
Hi,

I'm currently using an ECM8000 microphone calibrated by Cross Spectrum.

My problem is, I have the magnitude response calibrated but not the phase response (the phase column is all zeroes in the calibration file). I would prefer to have a linear phase response: indeed, I'm using DRC to calibrate my system, which is capable of correcting phase.

Question is: can I trust the "natural" phase response of ECM8000 microphones to be linear (or almost linear)? Searches I done gave me contradictory results: one shows a maximum phase difference of 15°, another shows a phase response which is all over the place.

Did someone actually measure the phase response of a fairly recent (post-WM60) ECM8000 microphone?
 

·
Registered
Joined
·
2,604 Posts
I have to admit I don't know. But, one of our members is Cross Spectrum, I have alerted him to your question and I expect he'll give his thoughts.

Welcome to the Shack!
 

·
Premium Member
Joined
·
397 Posts
Hi, this is Herb from Cross-Spectrum Labs.

I don't measure phase response because, frankly, there's no good way to measure the absolute phase response of microphones. Condenser microphones aren't minimum phase devices, so the Hilbert transform isn't applicable (which, AFAICT, if how most of the ECM8000 phase info on the web was calculated).

The only realistic way would be to compare the phase of one mic with another - the problem is that I haven't been able to find reliable phase information for any of my ANSI-rated reference mics. Even Bruel & Kjaer, who supplies a lot of technical info with their mics, doesn't provide that info.

About the best you can do is to assume the phase response is flat - that will certainly be fine for the mid-frequencies between 100Hz to a few kHz. Condenser mics do have a 90-degree phase shift at its resonance frequency... however, we don't know what the resonance frequency of the capsule is (for pro mics the resonance frequency might be anywhere from 10 kHz to 30 kHz). There will also be a phase shift at the lower end of the spectrum because of the capacitor used for the high-pass filter, but again I don't know what the Fc is.

Sorry.
 

·
Registered
Joined
·
22 Posts
Discussion Starter #4
Many thanks for your quick answer.

About the best you can do is to assume the phase response is flat - that will certainly be fine for the mid-frequencies between 100Hz to a few kHz.
If I understand this correctly, this means that correcting a system's phase response with an uncalibrated mic is not ideal, but is probably better than not correcting the phase at all. Guess that will have to do then.
 

·
Registered
Joined
·
22 Posts
Discussion Starter #5
I contacted iSEMcon GmbH, which also offers ECM8000 calibration services:

content.ibf-acoustic.com/catalog/product_info. php?cPath=21_23&products_id=76 (sorry for the space in the middle, the forum forced me to do that)

(see the sample data: these include phase response!)

I asked them what was their methodology for measuring phase response. Here is their (short) answer:

The phase response above 500Hz is calculated from frequency response (inverse fourier transform calculated from my measurement software). In the range 10Hz to 500Hz it is relative to my B&K reference mic (measured in a small pressure chamber).
If their measurements are correct, this means the sample ECM8000 has a quite linear phase response throughout the whole spectrum, which contradicts Herb's affirmations. Now I'm confused...
 

·
Premium Member
Joined
·
397 Posts
The phase response above 500Hz is calculated from frequency response (inverse fourier transform calculated from my measurement software). In the range 10Hz to 500Hz it is relative to my B&K reference mic (measured in a small pressure chamber).
The B&K 4133 is likely to have a flat phase response down to 3 Hz or lower, so the low frequency comparison test is kosher. For the upper freq phase test, calculating the phase from the freq response is only valid if you know that the source has a flat phase response, or the source has a known phase response (which means measured with a mic with a known absolute phase response)

If their measurements are correct, this means the sample ECM8000 has a quite linear phase response throughout the whole spectrum, which contradicts Herb's affirmations. Now I'm confused...
No contradiction, I was talking about the freq range that I new to be flat, nothing precluded the ECM8000 being flatter outside that range (assuming iSEMcon's protocol is valid).
 

·
Registered
Joined
·
22 Posts
Discussion Starter #7
I also contacted Denis Sbragion, the author of DRC. Denis stated, in section 6.2.6 of the DRC documentation (about the microphone compensation stage) that "most microphones suited for measurement tasks are minimum phase systems". I asked him the source of this information. Here is his answer:

I've seen it stated in many academic papers, so I took it for granted. Considering the simple nature of a microphone it looked reasonable to me, but I admit that I have yet to see a clear proof of it.

I've seen many mic phase responses, and they were all smooth and almost linear, which is what's expected from a minimum phase system having an almost flat magnitude response. BTW, never checked it directly, neither I verified that the supplied phase response wasn't simply derived from the magnitude response assuming that the system was minimum phase, i.e. a chicken and egg problem. Apart from this, minimum phase or not, I don't expect the phase anomalies of any decent measurement mic to be audible.

May be you're not correcting [phase] as expected but, for sure, you're not messing with it. The phase response of a measurement mic, even a cheap one like the Behringer, is not the biggest source of phase anomalies, by a wide margin.
 

·
Premium Member
Joined
·
397 Posts
I've seen assumptions made of minimum phase/not minimum phase in several contexts. When I have seen assumptions of minimum phase it's typically a simplifying assumption.

(and just to be clear, when I'm using the term "minimum phase" I'm using it as defined here to mean "there exists a unique relationship between amplitude and phase which allows a complete determination of phase from amplitude.")

There were a variety of things that convinced me that one could not (simply) calculate the phase of a condenser mic based only the frequency response, but what finally convinced me was an analysis I performed based on a figure in the book "Acoustical Measurements" by Leo Beranek. The book shows a plot of the (pressure) frequency response and phase response of the B&K 4145 microphone:



Using GraphClick, I scanned the points off the 4145 curves and imported the data in Octave with the intention of using the Hilbert transform to calculate the phase from the pressure response data. The B&K website also has the free-field data for the 4145 available, so I threw that in the mix as well.

The result:



Assuming I didn't screw anything up, the results I get via Hilbert are not the same as the published phase response. That's the point where I gave up on microphone phase.

With respect to:

Apart from this, minimum phase or not, I don't expect the phase anomalies of any decent measurement mic to be audible.
and

The phase response of a measurement mic, even a cheap one like the Behringer, is not the biggest source of phase anomalies, by a wide margin.
I would agree with those statements wholeheartedly.
 

·
Registered
Joined
·
66 Posts
Is it correct to say that the phase response of a dynamic mic is linear? I have an old E-V RE55 which is just itching to become a calibration mic. The RE55 is the descendant of the 655C which has an output transformer which probably introduces phase non-linearities.
 

·
Premium Member
Joined
·
397 Posts
Is it correct to say that the phase response of a dynamic mic is linear? I have an old E-V RE55 which is just itching to become a calibration mic. The RE55 is the descendant of the 655C which has an output transformer which probably introduces phase non-linearities.
Dynamic mics are basically loudspeakers run in reverse, so theoretically the minimum-phase methods for determining phase (Hilbert transform, etc) should work.
 

·
Registered
Joined
·
20 Posts
Assuming I didn't screw anything up, the results I get via Hilbert are not the same as the published phase response. That's the point where I gave up on microphone phase.
I just noticed this old thread. I wish I had seen it before.
Anechoic, I don't know if you screwed anything up or not but I have had similar results when using Hilbert transforms for phase calculations. The problem appears to be related to be mainly from end effects of the frequency response. I have had some success with extrapolating beyond the curve limits.

However the best results I got came from ditching Hilbert and using other methods.
See
http://www.hometheatershack.com/for...volver-mic-phase-correction-2.html#post492289

and
http://www.edaboard.com/thread238673.html

for details.
FWIW I also agree microphone phase is one of the last of our worries - dunno why I keep persuing it!
 

·
Premium Member
Joined
·
397 Posts
I just noticed this old thread. I wish I had seen it before.
Anechoic, I don't know if you screwed anything up or not but I have had similar results when using Hilbert transforms for phase calculations. The problem appears to be related to be mainly from end effects of the frequency response. I have had some success with extrapolating beyond the curve limits.

However the best results I got came from ditching Hilbert and using other methods.
See
http://www.hometheatershack.com/for...volver-mic-phase-correction-2.html#post492289

and
http://www.edaboard.com/thread238673.html

for details.
FWIW I also agree microphone phase is one of the last of our worries - dunno why I keep persuing it!
Thanks, I'll check it out when I have a moment.
 

·
Banned
Joined
·
516 Posts
May I ask a simple question?

While we obsess over the phase response of the mic, what is the intended use for the system?

And why do we not see a similar awareness or concern regarding of the nature of the actual 'system under test' where the speaker- room environment is very seldom minimum phase which renders the application of EQ to the space moot.

Yet we fail to see a corresponding focus on the causal time domain behavior in identifying and determining appropriate solutions to such issues addressed in the time domain?

I mean, how many multi-driver speakers are minimum phase at the measuring position?????? ...And that is limiting one's attention simply to the loudspeaker itself - independently of its behavior within a bounded space.....:D
 

·
Premium Member
Joined
·
397 Posts
Oh, I agree that the phase response even for budget mics like the ECM8000 or EMM-6 are likely to be far less of an issue that the phase response of any loudspeakers you are measuring, which is why I haven't spent a lot of time & effort trying to measure/derive the phase response (and why manufacturers high measurement mics don't bother publishing phase specs).

However, if there is a relatively easy & straightforward way to figure out the phase response of a mic, it's very much a service I'd like to offer, hence my interest.

As Tsardoz pointed out in another thread, the phase response can be an issue at extreme angles (and for dynamic mics that pretty much act like loudspeakers in reverse), although that shouldn't be much of an issue for well-made condenser mics.
 

·
Registered
Joined
·
20 Posts
May I ask a simple question?

While we obsess over the phase response of the mic, what is the intended use for the system?

And why do we not see a similar awareness or concern regarding of the nature of the actual 'system under test' where the speaker- room environment is very seldom minimum phase which renders the application of EQ to the space moot.

Yet we fail to see a corresponding focus on the causal time domain behavior in identifying and determining appropriate solutions to such issues addressed in the time domain?

I mean, how many multi-driver speakers are minimum phase at the measuring position?????? ...And that is limiting one's attention simply to the loudspeaker itself - independently of its behavior within a bounded space.....:D
Ah well - I know this whole thread may be pointless and and unlikely to have any useful outcomes. I am certain there are many more important things to think about.

For me it is really just an intellectual exercise I want to solve more than anything. I have had a long interest in hifi (I designed a linear phase active speaker in 1983 for my final year elec eng project!) . Since then I have been involved in DSP for biological systems and am just returning to audio (just as a hobby).

It is about baby steps. This was just an idea I had (that I see others have already had before me). Maybe I'll think of better ones later ... (or not!)
 
1 - 15 of 15 Posts
Top