ECM8000 microphone measuring techniques and usage discussion

[hifisponge]

brucek wrote:

It's my understanding that your mic was calibrated to be used in the vertical position, when placed in a room in a diffuse field (such as the listening position). That would mean it is accurate to rely on the results if you use the cal file in REW under those normal conditions.

The only time it would be inaccurate is when you take a close speaker measurement of the tweeter, where you need to point the tip horizontally at the speaker, and so the cal file is not accurate for that upper octave. Is that not a correct assessment?

Comments?

brucek

Herb with Cross Spectrum initially gave me just a horizontal cal file, but upon request also provided a vertical cal file.

In regards to the proper orientation, I was just discussing this with herb last night. Here's what he said:

For in-room speaker measurements / room-correction, 90-degree (vertical mic position) is good enough - these are free-field microphones which assume that sound will be coming around from multiple directions, but there will be a some error at the top end. When doing room measurements, I always recommend people measure at various locations (in three dimensions) to minimize orientation errors as well as the effects of room modes.

When performing loudspeaker freq response measurements, using the mic at a 90-degree angle is a no-no because it's not designed for that at high-frequencies. At the highest frequencies (say, 3kHz and up), the wavelength of sound is small enough that the mic body itself will start to interfere with the wave propagation. At a 90 degree angle, part of the wave collides with the body of the mic and the rest of it grazes across the surface of the diaphragm, so the diaphragm no longer moves as a piston and what the microphone 'sees' isn't actually representative of the wave form if the mic wasn't there.

Personally, because I want more accurate HF measurements, even at the listening position, I'm inclined to use the horizontal mic position and appropraite cal file all the time. I’ve also been thinking for a while now that there is merit to the belief that we humans give priority to the direct sound in the mid and high frequencies (over the combined direct + reflected sound) so it would seem that pointing the mic in the direction of the speaker would give a bit more weight to the direct sound while still capturing the reflected. Then again, I suppose this can also be addressed with reduced gate times.

Thoughts?

[robbo266317]

I totally agree Tim.
Each mic will be different since they are mass produced.
For critical measuring you will need to get it calibrated. Mine seems to be adequate for the purpose using the generic file and the tests I have done shows it is acceptable out to about 25khz.
Since my upper range these days is only about 13.5 khz I am not too worried.
I could calibrate it myself using the graphs supplied with the Ravens.
I'm glad I purchased them when I did, as two months later they doubled in price because the aussie dollar dropped. They are now $3,750 each out here.
http://www.audiomarketplace.com.au/c...art/Itemid,49/

I hope you have air-conditioning. It's going to be another scorcher over there again.

[spearmint]

Thanks for the conducting the tests and posting the results.

Anyone have any negative thoughts regarding using the Mobilepre?

[hifisponge]

Spearmint -

Both Anthony and I have concluded that the issue we were experiencing had nothing to do with the Mobil Pre. It works as advertised.

- Tim

[Anthony]

Yeah, the MobilePre is a great portable sound card. I even use it to tie into my laptop as a mobile music station (laptop to MobilePre to stereo). Good for measurements.

It does have some flaws if you want to use it for speaker building. The line out sags its voltage under even light loads, which makes it tough to measure impedance. It is also a bit noisier than some of its counterparts (again unless it's for really serious speaker testing it's not an issue). But the portability aspect and phantom power make it worth it.

[Anthony]

Okay we've merged the threads back together since they both ended up being a discussion of the ECM8k mic and its usage (originally they were split b/c I thought one would end up being a MobilePre discussion).

So back to one big thread folks!

Carry on!

[spearmint]

hifisponge wrote:

Spearmint -

Both Anthony and I have concluded that the issue we were experiencing had nothing to do with the Mobil Pre. It works as advertised.

- Tim

Anthony wrote:

Yeah, the MobilePre is a great portable sound card. I even use it to tie into my laptop as a mobile music station (laptop to MobilePre to stereo). Good for measurements.

It does have some flaws if you want to use it for speaker building. The line out sags its voltage under even light loads, which makes it tough to measure impedance. It is also a bit noisier than some of its counterparts (again unless it's for really serious speaker testing it's not an issue). But the portability aspect and phantom power make it worth it.

Thanks, very much appreciated...

[brucek]

An interesting response from Cross Spectrum Labs here in regard to mic orientation for measuring.

Does it mean we should stop recommending vertical orientation of the ECM8000 type mic when measuring from the listening position?

brucek

[Sonnie]

The question might be for those measuring 5.1 or 7.1 full room response, for something like Audyssey type measuring, does the mic need to be turned and pointed at each speaker as the tones/pulse moves from speaker to speaker? Or should we assume Audyssey has accounted for vertical placement of the mic?

And how do you point the mic at a sub, since it is omnidirectional? What if you have subs scattered about the room?

[brucek]

does the mic need to be turned and pointed at each speaker as the tones/pulse moves from speaker to speaker?

Well, Herb has clarified my incorrect interpretation of what he said. His recommendation is that the mic is vertical (straight up and down) with a slight 10-20 degree tilt toward the mains speakers.
And, if you're measuring a specific speaker near field, then it's as we already understood - point the mic horizontally on-axis at the source and gate the impulse response.

brucek

[cinema mad]

Sonnie wrote:

The question might be for those measuring 5.1 or 7.1 full room response, for something like Audyssey type measuring, does the mic need to be turned and pointed at each speaker as the tones/pulse moves from speaker to speaker? Or should we assume Audyssey has accounted for vertical placement of the mic?

And how do you point the mic at a sub, since it is omnidirectional? What if you have subs scattered about the room?

As far as Audyssey, Chris co creator of Audyssey has stated that the position of the mic supplied for use with Audyssey should be at Ear hight and pointed verticaly directly at the ceiling, because it is designed for Grazing..

This works fine unless you have A speaker such as center channel tilted slightly up then it "MAY" be best to adjust the mic angle to match the angle of the center speaker, but that creates A huge hassle ....

Cheers....

[Anechoic]

Sonnie wrote:

Or should we assume Audyssey has accounted for vertical placement of the mic?

I would presume that Audyssey has accounted for their suggested orientation of the mic and has their system optimized for that placement.

And how do you point the mic at a sub, since it is omnidirectional? What if you have subs scattered about the room?

For low frequencies it won't matter since pretty much any (condenser) mic will omnidirectional below 1000 Hz. Point it where ever you like.

[Wayne A. Pflughaupt]

In my years in pro-audio, both professionally and dabbling around the fringes before and since, I’ve never seen anyone use a vertical mic orientation when shooting a room. Typically the mic is aimed at the speaker, so when I first saw the vertical-orientation recommendation here I thought it was the most ridiculous thing I had ever heard. However, when I found out that we were recommending vertical because the ECM8000 was calibrated that way, I let the matter drop, even though I don’t buy into it.

Poking around the internet to bolster my case, I was surprised I wasn’t able to find much of anything to support either method. I looked up the manuals for several pro-audio hardware RTAs and even a couple of software-based programs like TrueRTA, and nothing specifically mentioned mic orientation.

The only thing I could find supporting a horizontal orientation was a picture in the manual of my AudioControl R-130 RTA that showed the calibrated mic pointed directly at the speakers, angled slightly upward. My RTA shows that response at the listening position starts drooping as far down as 1200 Hz with vertical mic orientation, and it gets really severe above 8000 Hz. Having used this RTA for well over 10 years, I trust it enough to know that if I EQ’d my system from a vertical mic orientation, it would be horribly inaccurate.

Likewise, the only thing I could dig up supporting vertical orientation was something on Goldline’s website mentioning that THX recommends it for home theater systems. I find that a bit problematic, however, given the wide range of shapes and sizes of rooms that home theaters find themselves in - high ceilings, low ceilings, cathedral ceilings of different angles and pitches, shoe-box rooms, family rooms open to other areas of the house, etc.

I don’t know how THX, Audyssey or anyone else can in good faith recommend a mic orientation that relies on reflections for upper frequency information to be applicable for all residential situations. Let us not forget that reflected sound will exhibit a certain amount of high frequency loss due to a certain amount of absorption from the walls and ceilings. If room treatments are present, the high frequency loss in reflected soundwaves will be even greater. Did Audyssey figure that into their “grazing” theory? Just a few of the reasons why I prefer horizontal orientation: You know what you’re getting with direct soundwaves.

brucek wrote:

Does it mean we should stop recommending vertical orientation of the ECM8000 type mic when measuring from the listening position?

As long as our ECM8000 calibration file was generated using a vertical position, I don’t have a problem with us recommending it for that mic.

However, a while back someone was asking about orientation who was using a dbx mic from one of the company’s RTAs, and we recommended vertical. I think we need to make it a point to tell people to use the orientation that was used to generate their calibration file. If that information is unknown, then I think horizontal should be the standard recommendation, at least for anyone taking full-range measurements. Even if it’s not specifically stated anywhere “on record,” it’s a safe bet that any mic that came from a professional-brand hardware RTA was calibrated horizontally, since that is the orientation traditionally used with that equipment.

Regards,
Wayne

[glaufman]

Interesting information and discussion... although you guys are the experts, my thoughts...
I fully agree with using the mic in the orientation the cal file was generated in, in ALL cases... anything else has potentially zero accuracy...
Assuming you can get a cal file generated for any orientation you want, if you're willing to pay for it, the question of preferred orientation for different applications, has to depend on the characteristics of a given mic... starting the the ECM8000...
Seems to me that any measurement intending to be a direct measurement, i.e. either ignoring/minimizing reflections and/or in an anechoic setup, point the mic directly at what you're measuring, use a cal file taken in that orientation, and go...
But I question the validity of room response measurements taken horizontally as it seems the body o fthe mic would obviously affect the response to room reflections ... polar plots could prove me wrong...

For low freqs the effects of the case surrounding the mic shouldn't affect things significantly, and at high freqs, when talking about room response, are we really looking for a kind of accuracy where a cal file taken in the verticaly orientation would not adequately account for the diffraction/grazing affects?

I can't comment on Audessey as I have no first-hand experience, but I hesitate to trust any automatic process to do it's job without taking manual measurements to confirm results... "trust, but verify"

[JMB]

Herb,
Perhaps, you could enlighten us.
Are your recommendations for mic placement during measurements based upon some literature that you might share or based upon your experience?
What position do you place the mics in during calibration?

Thanks,
Jay

[Anechoic]

Mostly based on my experience, the experience of my colleagues and my understanding of the science involved. I don't really have any literature I can point you to.

As for what position I place the mics - it really depends on a whole bunch of factors. For smaller rooms, I usually just hit the listening spot. For larger rooms where the listening point might be spread out, I like using spacial averaging techniques either my measuring at multiple positions and averaging the curves together, or (for systems where it's not practical to use the multiple-position method) I slowly move the microphone around in large vertical and horizontal circular motions. The "sweet spot" result getting this method isn't as good as the 1-position method, but you get an improvement at the other positions.

[JMB]

Are you saying that for calibration you place the mic in different positions, or for measurement?
I am really asking about the position that you place the mic in for calibration.

Thanks,
Jay

[Anechoic]

Are you asking about home theater system calibration or my microphone calibration?

[JMB]

I am asking about the microphone calibration. Sorry for being too vague.

Jay

[Sonic Icons]

Hi, I hope the following question for Herb isn't considered abusing a dead horse (that has already been discussed extensively in the thread):

Can you comment, based on your measurements, on "typical" variation of the ECM-8000 frequency response between the horizontal (0 degree, on-axis) and vertical (90 degree, grazing incidence) orientation? More specifically, is it possible to give an estimate of the frequency where the splitting between the horizontal and vertical response reaches +1 dB, or the frequency where the splitting reaches +3 dB? Or, does the horizontal-vertical splitting show a large variation between individual microphones, that makes it difficult to estimate "typical" values?

I'm asking this more for general interest in "how it works" than for practical reasons. My guess is that the directionality (horizontal-vertical splitting vs. frequency) is more consistent between individual ECM-8000 microphones than the basic frequency response, because the directionality is determined mainly by diffraction of sound waves from the microphone body, and the size and shape of the microphone body is highly consistent between units. But maybe I'm overlooking something

[Anechoic]

JMB wrote:

I am asking about the microphone calibration. Sorry for being too vague.

Jay

Hmm, I could have sworn that I explained my process here at some point, but I guess I didn't.

Above 100 Hz, I use a free field substitution method. I mount a reference microphone (ACO Pacific 7250, calibrated by Scantek) 27 inches away from a custom speaker with a full-range (coax) driver to measure a reference response. I replace the reference microphone with the ECM8000 and repeat the test. I use the reference response to correct the response of the ECM8000 and that gets me the >100 Hz response for that particular microphone. I use this procedure to measure mics at 0 degrees and at other angles. I use ARTA (swept sine) to make the measurements.

For frequencies below 100 Hz, I use the same substitution method, but rather than doing a free-field measurements, I use a small pressure chamber to obtain the pressure response of the reference mic and the ECM8000-under-test (at low frequencies, the pressure response is equal to the free field response for omni-directional microphones).

These measurements are consistent with IEC 60268 standard for microphone measurements.

For the polar measurements, I have a mic stand mounted on a lazy susan. I mount the mic in front of a speaker playing pink noise and rotate it around 360 degrees (in 10-degree increments) to measure the 1/3 octave band responses. I have the mic mounted so that the mic diaphragm is the axis of rotation - basically the mic capsule stays in the same ~ 1 sq cm of area and the cable end of the mic rotates around the diaphragm (make sense?).

For sensitivity measurements, I have an acoustic calibrator (GenRad 1986 Omnical, also regularly calibrated by Scantek) which puts out calibration tones at several frequencies and sound pressure levels. I place the calibrator over the ECM8000 capsule, hook up the ECM8000 output to my Sencore RTA, and just read the sensitivity figure from the RTA.

For noise floor measurements, I have a soundproof chamber within which the background noise (at night) is lower than 17 dBA. Since the noise floor of ECM8000's are generally higher than 30 dBA, I can just put the ECM8000 into the chamber and read the noise floor off my meter.

[Anechoic]

Sonic Icons wrote:

Hi, I hope the following question for Herb isn't considered abusing a dead horse (that has already been discussed extensively in the thread):

Can you comment, based on your measurements, on "typical" variation of the ECM-8000 frequency response between the horizontal (0 degree, on-axis) and vertical (90 degree, grazing incidence) orientation?

They say a picture is worth a thousand words, so here are plots from 6 random multi-angle measurements:



It looks like they all start to diverge around 2 kHz, which is consistent with the data you see from measurement mic vendors.

I think you're correct that the angle difference from mic-to-mic is more consistent than the frequency response, but I haven't gotten around to actually comparing difference curves to verify that. The "x" factor is the positioning of the mic element in the capsule which I noticed seems to have an effect on the upper frequency response of these mics.

[Terry McGee]

Heh heh, just going back to that schematic of the ECM8000, we've had a couple of interpretations on it, I thought I could really helpfully confuse things by adding another!

My feeling is that the circuit as drawn is probably correct, even if confusing. I think that first transistor (immediately after the capsule) is just there for absolute phase reversal, to make the mic conform to the "positive pressure wave hits diaphragm causes positive voltage excursion at pin 2" standard. Without it, pin 2 would go negative when the positive pressure wave arrived. Not a problem with a single mic, but a trap if you have more than one.

That then raises the question of what the lower transistor of the output pair is doing. If the circuit is wired as shown, its base connects (via a dc isolating capacitor) to earth (follow it round - it's a rather confusing bit of schematic drawing). So with a grounded base, there can be no signal on its emitter feeding pin 3. So it's not there to provide a balanced output voltage.

My guess is that it's there just to provide a guaranteed impedance balance to the real output on pin 2. So that any noise, hum, etc injected by way of cables passing by speaker amp transformers, lighting rigs, or even by way of a mic body in electrical contact with something nasty will be balanced out as well as the preamp's cmrr permits.

They could have achieved the same phase reversal without need for the extra stage by swapping XLR pins 2 and 3, but that would prevent people using the mic in unbalanced mode.

It would be easy to test my theory - there should be signal output on pin 2 only. Of course, we can't be sure that they haven't changed the circuit over time!

The treble rolloff provided by C1, the 3n3 capacitor across the capsule FET, is pretty brutal - I wonder what the the response would be like without it! Ultrasonic? If we ignore the FET for the moment, the combination of R1 and C1 would roll off at 6dB per octave from around 3KHz. If we assume the FET drain impedance is similar to R1, that would lift the start of the treble rolloff to around 6K. Presumably the capsule exhibits a serious resonance somewhere near the top of the audio spectrum and this heavy rolloff helps keep it under control. The economics of commercial electronics manufacture has always been a mystery to me - I wonder how much more you have to spend to get a capsule that doesn't need such aggressive EQ. Another dollar probably!

Looked at another way, you could probably get in there with a more tailored network of components and achieve a far better response before calibration. But hey, that would be overcapitalising (to use real estate language). And you'd probably need an electron microscope to deal with the size of the components.

Terry

[brucek]

My guess is that it's there just to provide a guaranteed impedance balance to the real output on pin 2.

Perhaps, but there would have been a lot easier ways to do that.

Looking at it again, I suspect it's simply drawn wrong and that the first BC118 stage is not only acting as a high impedance buffer for the mic element, but also as a zero gain phase splitter.

I edited the diagram (below) to show where C6 is likely connected to the emitter of BC118 and not to ground. Generally a phase splitter would have equal resistors on the collector and emitter, but they're close, and it's the only solution I could come up with....

brucek

[Terry McGee]

Perhaps, but there would have been a lot easier ways to do that.

Yes, that is possibly a weakness in my hypothesis! Very thorough though - perhaps too thorough for a low price mic?

Looking at it again, I suspect it's simply drawn wrong and that the first BC118 stage is not only acting as a high impedance buffer for the mic element, but also as a zero gain phase splitter.

It's not really acting as an impedance buffer - note the values of R4 and R10 (ignoring other components for the moment) are virtually the same. But that doesn't rule out your zero gain phase splitter interpretation.

I edited the diagram (below) to show where C6 is likely connected to the emitter of BC118 and not to ground. Generally a phase splitter would have equal resistors on the collector and emitter, but they're close, and it's the only solution I could come up with....

Yeah, that would certainly work, and would seem to explain the carefully balanced back end better. I wouldn't be too worried about the 1k2 vs 1k discrepancy. It might have been empirically found to give better balance, or it could be another typo - for example how many 46K resistors (R3) have you come across? Presumably a 47k and a partially colour-blind technician!

On behalf of my hypothesis, we need to note that there is no real need to balance the output voltage of a microphone - it's not as if the enormous voltages concerned are going to make a nuisance of themselves in a jackfield (which is why we balance output voltages). So either of our interpretations will work fine). Yours will have 6dB more output, but if they wanted that they could have put some gain in that intermediate stage (but not if they wanted the phase shift needed in your interpretation). Mine will be less fussy about working into an unbalanced load, but most people needing 48V phantom power will be getting it from a balanced preamp.

I guess there's nothing for it, one of us is going to have to go in there to find out! Or at least check that there is output on both 2 & 3 (your interpretation) or on 2 only (my interpretation). I'll get to that and report back. I'd have to say, I suspect on the balance of probabilities that yours is more likely!

I think the most remarkable element in the circuit is the enormous HF load represented by C1 across the FET. Like sending a boxer into a ring filled to head height with marshmallows! Hmmm, suddenly suspicious, could this be another typo? Perhaps not, given that the turnover frequency would appear to be consistent with the variable top-end boost shown in the various response graphs?

Good to talk audio circuits with you, brucek!

Terry