Two microphones as matched pair, what properties are matched?

[MicUlli]

Hi All
I am a little bit confused what matched pair regarding microphones means. Some sellers "match" their mics by consecutive serial numbers
Others select nearly equal sensivity (at only one frequency) or similar frequency response.
My approach for putting together a matched pair is as follows:
1. Nearly identical polar pattern over frequency is the foremost rule,
2. Frequency response on axis should be very similar but can be equalized,
3. Sensivity is not as important because it can be easily adjusted.

What are your thoughts?
BR MicUlli

 

[cyrano]

I've never believed in "matched pairs". It's possible some manufacturer is doing it correctly, but the ones I've seen were no more than consecutive serial numbers. Now I didn't have a Neumann, or a Schoeps, or... just some Rode, AKG, T-Bone, Shure.

In general, with modern professional mics, the difference between random mics from the same model is already minute. Nothing I can hear, just measurable.

An example: I bought seven used AT835 hanging mics. Six of those are very, very close. The seventh one is about 1,5 dB lower in level but still matches frequency respons quite closely.

 

[molke]

The only matched pair I own is the Oktava MK-012 (I don’t count the t.bone SC 140), so no Neumann or Schoeps. Judging by the supplied curves and notes, they seem to be matched for sensitivity and on-axis frequency response.

The individual curves contain 90° and 180° responses, and those look rather matched as well. If you did measure some off-axis responses, could on-axis be a good enough indicator, that is, if on-axis is matched, then off-axis will be matched as well?

Your approach is probably more useful, but I’ve yet to find a way to measure off-axis response reliably.

 

[kingkorg]

1. Nearly identical polar pattern over frequency is the foremost rule,

It is pretty difficult for two capsules of same construction to have different polar pattern if the frequency response matches. They are intrinsically tied together. I can't think of any capsule that would be exception.

Some sellers "match" their mics by consecutive serial numbers

Nope, there are many ways to go about it depending on construction, circuit, capsuøes, but many manufacturers match two mics by their criteria and give them consecutive serial numbers at the end of the process.

Matched microphones would have all the parameters you named matched. The s/n and output level would have to be matched too. However i am aware of many even reputable brands that do lousy job at this. As well as some that are so tight in production they don't need any matching, and they don't even advertise it.

 

[MagnetoSound]

I don’t believe Neumann or AKG ‘matched’ anything back in the day ... (pre-1980s ...?)

Users would select mics that were sufficiently close for certain applications.

It’s hype, really.

 

[MicUlli]

I agree in most of the items stated. But under certain circumstances matching a pair may be very useful and necessary. The question is how close the mics should be...
Find attached a BBC paper which enlights the audible interchannel differences of stereo material. Probably up to 2 dB interchannel difference is not an issue at all..
But the paper is related to speaker playback. May be that playback via headphones gives different observations.

 

[Icantthinkofaname]

I don’t believe Neumann or AKG ‘matched’ anything back in the day ... (pre-1980s ...?)

Users would select mics that were sufficiently close for certain applications.

It’s hype, really.

My impression is that if a mic is to spec with a frequency response 2 dB higher or lower than advertised, it miiight be a way to not have a 4 dB difference between mics. I'm trying to think of models like that off the top of my head, I know there are a couple with published graphs that show what variations are to spec, I just can't recall which specifically, though I believe CAD had some like that. Other than that I've seen "matched pairs" that look about as even as any two to spec mics of the same model. 6 of my Oktava MK-012s sound close enough and have close enough output that you'd think they're matched pairs (I do buy matched sets, but all 6 are basically the same), I have another matched pair that sounds a bit different, but output is still basically the same.

 

[Basser75]

With bidirectional ribbons, it's pretty easy. You can use the test software to divide the frequency response of one mic with another mic, and anything within a half dB (or whatever your standard is) is a match. And as long as the divided response is between 0.75 and 1.25 dB (centered around 1), then the sensitivity is matched as well. The polar patterns are gonna be matched because bidirectional ribbons are very consistent.

 

[Grim Reaper]

I have a matched pair of Sennheiser MKH416T-F microphones. I ordered them that way. Their frequency responses are matched to within 0.5dB. I still have the frequency response measurement chart from Sennheiser for one microphone but I seem to have lost the chart for the other one.

 

[RuudNL]

Neumann says their micrphones have such a narrow tolerence, that if you have bought a microphone and two years later you buy a second microphone of the same type, you have a 'matched pair'.

 

[mappee]

Thought consecutive serial #'s meant the mics were assembled from the same batch of components giving a good chance of matching performance specs. As mics age the values may drift and then what? In the end it's an ear thing
over specs. I've learned to take specs a bit random but they are a starting point.

 

[abbey road d enfer]

Thought consecutive serial #'s meant the mics were assembled from the same batch of components giving a good chance of matching performance specs.

Being from the same batch is not a guarantee that they will perform identically.
People who assemble mics do not necessarily take capsules by order of manufacture, rather by order of appearance. A particular capule may have been delayed for a significant period of time, for any of various possible reasons in manufacture, among them sampling for QC, being misplaced and later found, being reworked or built on Friday.
What actually makes consistent products is QC.
Any serious manufacturer guarantees that their "matched pairs" have published specs within tolerance, so, as mentioned earlier, good QC ensures that any random pair will be matched.
Consecutive numbers are purely cosmetic.
Manufacturers with less stringent QC need to make extra tests for matched pairs.
Electronic components do not need matching, unless they are from dubious origins.

 

[MicUlli]

@RuudNL
Neumann speach is a joke, look at the pictures attached for 2 Neumann KM184.
@kingkorg
measurement of Neumann KM184 tell us that polar response (at least at 90°) match perfectly but frequency response is completely different. Seperate equalisation and level adjustment would bring up the Neumanns to a well matched pair (but i would never spend 1350€ for them).

For comparison i have attached the same figures for 2 Telefunken TC600. These mics are very well matched (for my opinion). I would spend 300€ or more for them...

 

[RuudNL]

measurement of Neumann KM184 tell us that polar response (at least at 90°) match perfectly but frequency response is completely different. Seperate equalisation and level adjustment would bring up the Neumanns to a well matched pair (but i would never spend 1350€ for them).

I wouldn't buy a KM184 anyway! They were introduced as a successor of the well known KM84, but they sound nothing like a KM84...

 

[kingkorg]

@RuudNL
Neumann speach is a joke, look at the pictures attached for 2 Neumann KM184.
@kingkorg
measurement of Neumann KM184 tell us that polar response (at least at 90°) match perfectly but frequency response is completely different. Seperate equalisation and level adjustment would bring up the Neumanns to a well matched pair (but i would never spend 1350€ for them).

For comparison i have attached the same figures for 2 Telefunken TC600. These mics are very well matched (for my opinion). I would spend 300€ or more for them...

Don't mean to question your findings, but IMHO, and experience with measurements, there is no way this kind of result can happen where off axis and on axis can diverge that much. Allign off axis but not on axis? There must be something wrong. I'm not aware how physics would allow this.

Not saying either Neumann is flawless, but if on axis diverge this much, off axis has to diverge too.

I have sold the SB Acoustics drivers you recommended, i never managed to get the flat performance you suggested they have. No matter what baffle i used, they are nowhere near published specs or your reports.

This has lead me to a month long questioning and testing of my own setup, and i'm nowhere near answer. Either i'm doing something terribly wrong, and all my measurements are off, or the error might be on your side.

I'm not sure why you publish 90° divided by 0° of the same mic, off course they will line up perfectly with capsules of the same construction. It's exactly the point i made. But it's misleading. I hope you get my point.

Edit:

I think i know what happened. You need to take a measurement of your reference mic at 0°.

Then measure DUT mic at 0° and 90°. Alling spl of all the measurements at 1Khz. Then divide both DUT 0° and 90° mic measurements by reference 0° measurement.

You also have to work on the space you are measuring in, or do it outdoors. This much smoothing isn't the way to go. Especially if you are measuring off axis.

 

[MicUlli]

@kingkorg
"Don't mean to question your findings, but IMHO, and experience with measurements, there is no way this kind of result can happen where off axis and on axis can diverge that much. Allign off axis but not on axis? There must be something wrong. I'm not aware how physics would allow this."

PLS interpret the diagrams correctly. There is no off axis response BUT the relationship between 90° frequency response and 0° frequency response. Therefore it is named polar response PR.

"Not saying either Neumann is flawless, but if on axis diverge this much, off axis has to diverge too."

Yes of course, it does. But polar response is what you have to discuss (again correct interpretation of my measurement results helps)

"I think i know what happened. You need to take a measurement of your reference mic at 0°."
Thanks for the hint I ALWAYS publish the FR graphs after dividing the DUT measurement by the REF measurement !

"Alling spl of all the measurements at 1Khz."
Doesnt make sense because FR @ 90° / FR @ 0° tells the whole story about mic directivity type (super- hyper- cardioid).

"You also have to work on the space you are measuring in, or do it outdoors."
Why that? Comparing all measurement results is independent on room influences because the measurement conditions were the same for all DUTs.

"This much smoothing isn't the way to go. Especially if you are measuring off axis."
The same as before: measurement conditions are the same for all DUTs. Therefore smoothing doesnt change anything.

But you didnt get my point: frequency response does NOT tell you the whole story regarding off axis response. Matched frequency responses are no guaranty for matched directivity. Same applies vice versa as proved here. Matched directivity can lead to significant deviations in frequency response.

Therefore i repeat my requirement for matched mic pairs:
1. matched directivity rules,
2. frequency response can be equalised (of course only in the DAW),
3. sensitivity can be corrected easily by gain knobs or in the DAW.

By the way: I am wondering why your SB Acoustics drivers do not perform well. Try to measure frequency responses at 10 cm, 31,6 cm and 100 cm distance and compare them to each other. They should have 10 dB distance from each other and match to 2 dB over frequency...

 

[abbey road d enfer]

@RuudNL
Neumann speach is a joke, look at the pictures attached for 2 Neumann KM184.
@kingkorg
measurement of Neumann KM184 tell us that polar response (at least at 90°) match perfectly but frequency response is completely different.

Are these two mics supposed to be reasonably close w/regard to history?
It's a quite well-known fact that doing the same type of comparison between two U47's will lead to the same type of discrepancies, if only because the diaphragms have not aged identically.

 

[terry setter]

I had a friend who worked for Electro Voice many years ago. He swore that they would put two mics in a pink noise field, gain match them, and sum their outputs. They then put one of them out of phase. If they saw 20dB of signal reduction from the summed pair, they considered the mics to be "matched". Oh the other hand, I bought two Blueline M7 capsules from Thiersch about one year apart. It turned out that I needed a a matched pair, so I contacted Thiersch about swapping them for a matched pair. They wrote back and got the serial numbers of my two capsules I already had. Then they went into their records and were able to assure me that those two capsules were identical enough to be called a matched pair. They keep extensive records of the testing they had done before shipping the capsules to America.

 

[MicUlli]

Are these two mics supposed to be reasonably close w/regard to history?
It's a quite well-known fact that doing the same type of comparison between two U47's will lead to the same type of discrepancies, if only because the diaphragms have not aged identically.

The 2 KM184 were rented, the SNRs were NOT consecutive. Probably one of the mics was replaced. I have no idea what manufacturing date they had...
Nevertheless i was not very impressed of their sound

 

[kingkorg]

@kingkorg
"Don't mean to question your findings, but IMHO, and experience with measurements, there is no way this kind of result can happen where off axis and on axis can diverge that much. Allign off axis but not on axis? There must be something wrong. I'm not aware how physics would allow this."

PLS interpret the diagrams correctly. There is no off axis response BUT the relationship between 90° frequency response and 0° frequency response. Therefore it is named polar response PR.

"Not saying either Neumann is flawless, but if on axis diverge this much, off axis has to diverge too."

Yes of course, it does. But polar response is what you have to discuss (again correct interpretation of my measurement results helps)

"I think i know what happened. You need to take a measurement of your reference mic at 0°."
Thanks for the hint I ALWAYS publish the FR graphs after dividing the DUT measurement by the REF measurement !

"Alling spl of all the measurements at 1Khz."
Doesnt make sense because FR @ 90° / FR @ 0° tells the whole story about mic directivity type (super- hyper- cardioid).

"You also have to work on the space you are measuring in, or do it outdoors."
Why that? Comparing all measurement results is independent on room influences because the measurement conditions were the same for all DUTs.

"This much smoothing isn't the way to go. Especially if you are measuring off axis."
The same as before: measurement conditions are the same for all DUTs. Therefore smoothing doesnt change anything.

But you didnt get my point: frequency response does NOT tell you the whole story regarding off axis response. Matched frequency responses are no guaranty for matched directivity. Same applies vice versa as proved here. Matched directivity can lead to significant deviations in frequency response.

Therefore i repeat my requirement for matched mic pairs:
1. matched directivity rules,
2. frequency response can be equalised (of course only in the DAW),
3. sensitivity can be corrected easily by gain knobs or in the DAW.

By the way: I am wondering why your SB Acoustics drivers do not perform well. Try to measure frequency responses at 10 cm, 31,6 cm and 100 cm distance and compare them to each other. They should have 10 dB distance from each other and match to 2 dB over frequency...

I understand now, but i dont think anyone else does. Just publish 0°, and 90° of both mics, without dividing 90° by 0°. You just can't do it this way without ending up chasing your tail. There is no reason to divide 90 by 0 unless you want to confirm what i stated, which is that off axis sound will follow the changes of on axis sound with same capsule constructuon.

I am the first to claim you don't need anechoic chamber for accurate measurements, but you can't just smooth the responses and be done with it. In that case no one would ever build anechoic chamber.

The issue is your reference mic is pointed at 0°, while DUT is 90°. Without proper anechoic measurement you are measuring reflections, and room nodes created by the difference in the position + pattern difference. Just smashing smoothing to make it look nicer is brute force approach, and doesn't give correct measurement.