Topic: microphone matching

[page]

DSatz pretty much has said it. I will only add some small collected tidbits.

First of all, as already said, what the manufacturer does when "matching" mics is not defined anywhere.

Agreed, I noticed on my ck930s that they are matched as a pair of specific cap and specific body against another specific cap and specific body. Now, do I run them that way in the field? Probably not (I have a 50/50 chance), but it's stated on the plots I have that the following pair result in the following data that was used for evaluation.

Unfortunately, we tend to find explanations for our observations that may fit those observations, but are nonetheless based on junk science.  We humans like to be able to explain everything we encounter, even if we don't fully understand what's going on.  That's not to say that our badly founded explanations are not worthwhile.  Usually there is a strong correlation between our explanations and our observations.

That is an awesome articulation on the subject of fear of the unknown as it pertains to reasoning. I've thought about this in the back of my mind for years now when doing any type of research (academic or otherwise).

[DSatz]

tonedeaf, I wouldn't call ultra-precise matching a requirement by any means, but it is certainly an advantage for stereo recording with coincident or closely spaced main microphones.

By the way, when I mentioned earlier that the more serious manufacturers would prefer if they could make microphones with even greater consistency than they do now, there is also an interesting contrary viewpoint which I've come across, especially among aficionados of "vintage" microphones. As you may know, there are people who practically worship certain older designs such as the Neumann U 47, M 49 and M 50, KM 54 and U 67, the Schoeps M 221 B, and the AKG C 12 and Telefunken Ela M 250 and 251.

But for each engineer or producer who swears up and down that no microphone made today is in the same league as these classics, you will usually find that they own one or a few of these microphones which they have particularly become fond of, and it's only those particular examples that they mean. As a thought experiment, if I could bring together, say, the twelve or fifteen most outspoken devotees of these microphones and have them put all their individual microphones on a table, then if I scrambled up whose microphone was whose and returned them, those people would probably hang me from the nearest lamppost in short order.

What this comes down to is that if someone is looking for a particular sound quality which is not exactly the "center line" of the microphone's tolerance field, then if production tolerances are wider, that person has a greater likelihood of finding a microphone they really like than if all the microphones of that type sound the same.

It also implies what I think is the truth in this case: If companies such as Neumann or AKG could re-issue the old microphones exactly as they were made in the past (something which some people loudly and insistently demand), many of the most ardent fans of vintage microphones would dislike the result--because what they really like isn't the sound quality which that type of microphone was originally intended to have, and (on average) did have; instead, they like an individually altered version of that sound quality which happens to have come about due to the more or less random circumstances of component tolerances and aging in the particular instance(s) of their own favorite microphone(s).

--best regards

[ero3030]

sounds alittle like vintage wine/new bottle off the shelf!?   ed

[Gutbucket]

Great discussion. I'm primarily posting to follow the thread.  As for the vintage mic craze, DSatz's last post soundly reinforced my personal thoughts.  I've never seen it stated so clearly:

(snip) But for each engineer or producer who swears up and down that no microphone made today is in the same league as these classics, you will usually find that they own one or a few of these microphones which they have particularly become fond of, and it's only those particular examples that they mean. (snip)

Question: how useful is the basic home matching method which Richard outlined (ie- recording pink noise and visually comparing the frequency spectra) in determining how well mics are matched, or perhaps more importantly, in determining if the response of the capsules have drifted from each other over time?  Is it worth putting some time, thought and discussion into how to best perform a test or might I just end up playing mind games and fooling myself? Would I be better off forgetting about trying to determine the deviation of frequency match between mics myself and instead find a lab to do it for me? Any idea what a ball-park cost to do that might be?

I have an older Gefell pair which seem off.  I also have four DPA miniature omnis which I use for surround recording that should be pretty close to each other.  I bought one pair new and paid a bit more to have the two matched by the US distributor. That pair has remained very closely matched to my ears. I've become suspect of one mic of the second pair which I bought second hand.  I'm not concerned with overall average sensitivity which is simple enought to correct for (as long as the sensitivity difference is linear across differing dynamics), but differences in frequency response.

[stevetoney]

I almost hate to go there because there might be some waffling, but there has been alot of discussion about manufacturers that do a good job with quality control and how that affects matching and those that don't do such a great job.  A logical follow-on would be, is it fair to ask which manufacturers fall in which category, or can we simply use price point as a good indicator of the answer to that question?

Answering in part, it seems that this thread has already ascertained Schoeps, Neumann, Senn, DPA, and MG, as high quality manufacturers that use tighter tolerances on what they'll allow to leave the mothership.

What about some of the others; AKG, Peluso, Josephson, Shure, Milab, Audio Technica, and so on?

[illconditioned]

Question: how useful is the basic home matching method which Richard outlined (ie- recording pink noise and visually comparing the frequency spectra) in determining how well mics are matched, or perhaps more importantly, in determining if the response of the capsules have drifted from each other over time?  Is it worth putting some time, thought and discussion into how to best perform a test or might I just end up playing mind games and fooling myself? Would I be better off forgetting about trying to determine the deviation of frequency match between mics myself and instead find a lab to do it for me? Any idea what a ball-park cost to do that might be?

I have an older Gefell pair which seem off.  I also have four DPA miniature omnis which I use for surround recording that should be pretty close to each other.  I bought one pair new and paid a bit more to have the two matched by the US distributor. That pair has remained very closely matched to my ears. I've become suspect of one mic of the second pair which I bought second hand.  I'm not concerned with overall average sensitivity which is simple enought to correct for (as long as the sensitivity difference is linear across differing dynamics), but differences in frequency response.

For small omni mics (lav mics), this is a good sanity check.  Certainly for newly purchaced mics, you can get a quick sanity check.  I wouldn't trust anything until I test it, especially used mics.

As for bigger mics I think the problem is harder.  I have tried putting the mics in the same place in the room and recording the same (pink) noise pattern for each mic.  But this is more tricky (to line them up).  Not to mention, you're only measuring incident, not off-axis, performance.

Comments?  I'm willing to bet that some micbuilders/vendors use this method, as a sanity check, or to check a batch for defective capsules.  Though they may not disclose their "secrets".

  Richard

[Gutbucket]

It's very effective if you have a good tool to analyze the resulting recording.  Although you won't know about past capsule drift unless you have an earlier calibration.

Cross-Spectrum Labs does basic testing for a low price:

http://www.cross-spectrum.com/measurement/services.html

Thanks, I'll check that link..

What would qualify as a good analysis tool?
Can I simply open an RTA frequency analysis window in Samplitude- probably with a very long time average, wait for the oscillation in the displayed curve to settle and take a screen shot of each mic's average curve for comparison?

How about using something the Room EQ Wizard java application (freeware), which probably provides the ability to save and compare response curves and even calculate the difference between curves? [Hmm, i thought we could indent without making a list.. oh well here is the overview blurb]-

• Room EQ Wizard is a Java application for measuring room responses and correcting modal resonances. It includes tools for generating test signals; measuring SPL; measuring frequency and impulse responses; generating spectral decay plots, waterfalls and energy-time curves; generating real time analyzer (RTA) plots; calculating reverberation times; displaying equalizer responses and automatically adjusting the settings of parametric equalizers to counter the effects of room modes.

Talking that line of thought one step farther.. Could I then use the measurement of a known-good mic as a reference, figure the difference between that measurement curve and the subsequent measurement curve of an off-specification mic, invert the difference and apply that curve as a corrective eq band-aid for the 'problem mic' which I could apply after I've made a recording?

Isn't that partially what the corrective portion of the software for Len's Ambisonic 'Tetramic' does, allowing Core to use less expensive capsules and furnishing 'correction files' which are specific to each mic?  In that case there is no burden in applying those response corrections since the 'A-format' signal needs to be matrixed in the software before it is useful anyway.  I'd imagine that any tetrahedral, near-as-possible-to-coincident 4-capsule design would be more demanding in terms of mic matching than most any other application. 

[Gutbucket]

You nailed it.  I basically want to get an idea of how well matched my four identical mics are, and use that information to EQ the deviants in a broad sense back to usefullness- or perhaps just help me to varify and accept that I may just need to replace a couple of them.  I don't use Wavelab, but I'll check the FFT in Samp for looking at this.  I have a good grasp on the general test rigor required of keeping everything unchanged except for the single variable of the mic capsule under test.  When I get some free time I'll give it a try. 

As it is, I currently EQ the playback channel with the one suspect mic by ear to achieve a similarity of tone with the others when I'm using them all in a single array.  I agree completely on the slippery slope of attempting fine granular corrections like the impulse response manipulation.  The room correction stuff is most interesting to me for analysis of the bottom-most few octaves.  I'd think attempted correction of high-Q variations in mics would create problems similar to those found in agressive electronic room correction- loosing the view of the forest for the trees.  I'm not going that route either, but it is interesting to think about.  It may be that electronic correction of small omnis to an appropriately smoothed curve might be practical, but again I mostly want a general picture as an 'EQ-ing guide'.

Fortunately I'm not looking to duplicate the sound of a completely different mic, just one that differs from it's siblings, and to make that correction in a broad, low-Q tonal sense. Both the off and on-axis response of the suspect mic will be adjusted by the same amount and appropriately so since the models are identical, small and about as omnidirectional as achievable in the audible range. 

As for my Geffells, I'd just like to test them to see where they stand before I send them both in to a pro for a checkup and re-work.

[stevetoney]

Hey Gut and MSHil...what language are you speaking in?  I'd like to use a translator to try to understand your discussion!  So far, nothing comes up for the word 'anechoic'.  LOL! 

[Gutbucket]

Those fancy terms make it easier to get the meaning across without writing so much. 

I’ll try to explain a bit on what’s behind what we’re talking about and explain what anechoic is-

One part of measuring the microphone response is the physical measuring part, which includes: Choosing and producing a sound picked up by the mic (that's called the test signal); Placing the mic in front of the sound source that produces that signal (a speaker); The environment in which the source and mic are located and the measurement is made.  Each of those things effect the measurement.  Consider the last one.

Since the microphone and sound source are in some kind of room or enclosure, the microphone will pick up all the sound that bounces around that space, not just the direct sound from the source.  You end up measuring not just the microphone, but the combination of the microphone, the speaker that produced the sound, and the reflections from the room that the sound bounced around in.  Of those three things, the microphone often has the most neutral response, which means that what you're trying to measure is buried beneath a lot of extra information. 

One way to eliminate some of that extra information and clarify things would be to eliminate the room itself.  A perfectly anechoic room allows no outside noise in, and produces no reflections from sounds produced inside it.  All sound produced inside is absorbed completely with no reflections or echoes: an-echoic.  Those rooms are absolutely unnatural to be in (i've been told) and very, very expensive to build so only governments, universities, test facilities and big audio companies typically have them.  Even then, they are not perfectly anechoic to the very lowest frequencies, just good enough for most uses. 

Since we don’t have a real anechoic room, we do other things eliminate the influence of the room on the measurement.

[page]

there is an anechoic room up in Boston, at MIT I think. If I remember correctly, if you go by and it's a slow day (and you ask nicely), they will let you sit in it for a few minutes. It's been on my list of things to do for a while now.

[Gutbucket]

Cool.  I wonder if it begins to sound like screaming ear ringing in there after a while, or not.

[rjp]

My dad worked for a manufacturer of audible signals, and they had an anechoic chamber for testing and tuning purposes. When I was a kid, I got to check it out a few times... it was downright weird. If you have tinnitus (and I've always had a bit) it becomes very noticeable.

[SparkE!]

After about 5 or 10 minutes in an anechoic chamber, I begin to hear the blood moving in my veins.  It's weird hearing your blood moving in your arms and neck.  It's also weird hearing the movements of your internal organs.

[dactylus]

After about 5 or 10 minutes in an anechoic chamber, I begin to hear the blood moving in my veins.  It's weird hearing your blood moving in your arms and neck.  It's also weird hearing the movements of your internal organs.

^
Wow - I have had tinnitus since I was a kid so I'm guessing that it would be pretty horrible for me...