Matched VS Unmatched Sennheiser mics?

[ktime]

Hey all.

Im wanting to by 2* Sennheiser MKH8050's, but do not have the time nor the money to order a matched pair from the factory !.

i plan on using the mics for environmental field recordings, and my question is what are the consequences when using an unmatched pair for recordings?. will they produce a more false spatial relationship, dose having slightly different sonic characteristics in stereo field make the recordings off-putting.

any feedback would be greatly appreciated

marty

[staudio]

Some mics manufacturers are more consistent and others are not. I have two AKG 414 XLII's that were purchased months apart and I use them for stereo without any noticeable difference between them. Given my experience with the older Sennheiser MKH line, I do not think you will have any meaningful issues if you do not purchase a matched pair. All the multiple MKH mics that I have used were identical in sound/sensitivity even though they were not matched. So, when talking about high quality mics from reputable manufacturers I personally would not worry too much.

On the other end of the spectrum, I once had to use a pair of MXL's that were actually purchased as a pair but were almost 10dB different in sensitivity from each other. What crap.

This is my experience, your experience may differ.

[Don S]

My Stereo Set of 8040's does NOT have consecutive serial numbers, but a graph shoing the mics to be decently consistant. I don't hear any difference between the 2. Go for it! You'll love these new senn mics.

[audio ergo sum]

Neumann and Sennheiser developers both have explained to me, that their manufacturing these days is so precise that matching is not necessary. Same with Schoeps. As long as you don't use a MKH from 15 years ago with a more recent one in a stereo setup you should be fine. Most of this matching offered by companies is a hoax anyway. It would require considerable additional resources to *really* match microphones based on relevant parameters...

[John Willett]

MKH 8020 and 8040 stereosets are matched after manufacture (to within 1dB for both frequency response and sensitivity). The serial numbers are engraved on the body before manufacture, so matched stereosets will *not* have consecutive numbers normally.

MKH 8050 matched pairs *can* be supplied and the price is exactly the same as buying two at random (ie: there is no extra "matching" charge). These would be special order from the factory and I would guess be about 3-4 weeks for delivery.

Taken at random, I would say they are not quite so tight as the MKH 20/30/40 series. The earlier series are a lot more expensive than the MKH 8000 and the larger body makes for easier manufacture. But they will still be pretty close.

Personally I would get a matched pair - or, if time is short, see if the dealer can get you a manually matched pair. This would be done by just opening a few and holding the frequency response graphs up to the light on top of each other and picking the two that are closest.

I hope this helps.

[David Rick]

I picked up a B-stock pair of unmatched 8050's that was used at the Aspen Music Festival. Let me try to describe the differences I see in the factory sweeps.

Serial No. 10636 is commendably flat between 100 and 10,000 Hz. There's maybe a 1/4 dB bump at 150 Hz, and a -0.5 dB depression between 500 and 2k.

Serial No. 10647 is down -1.0 dB at 100, and the midrange depression is broader and a bit deeper: it goes from 200 to 2k, and is mostly down -1.5 dB. It also has a 0.75 dB bump at 8k.

I can summarize by saying these two curves are within 1 dB of one another over their complete range.

David L. Rick
Seventh String Recording

[Audiop]

I have a stereo set of MKH8040 with consecutive SN and a matching that even beats my QTC1MP. It's hard to see but the pair is within 0.3dB at least. Identical (as far as the eye can tell on these graphs of poor resolution relatively speaking) over most of the range with 0.25-0.3dB difference "peak to peak" in a small range.

My pair also have flatter response than the litterature and web page graphs show. Spec indicates +/-1dB over a wide range but my pair has closer to +/-0.5dB

I'm impressed!!

The MKH8020 pair I have is within 0.5dB and is slightly flatter than spec. +/-0.75dB instead of +/-1dB and do not have consecutive SN.


/Peter

[ktime]

hmm .. decisions decision...

thanks for all the reply's... its good to know that an unmatched pair sounds like a safe enough option to go through with...

i think ill try get a matched set just for peace of mind, but as im based in australia, going through any shops/distributes might prove to be to much of an expensive nightmare, due to our GST and harsh price hikes aussi electronics retails love to give us!.

m

[audio ergo sum]

just some food for thought:

Manufacturer published on axis frequency responses tell the whole story of a frequency response of a mic, specially omnis ?

What is more problematic for stereo imaging: small deviations in frequency response or deviations in timing/phase between mic pairs ?

Anyone seen a HRTF for human ears? And their difference between left and right ear?

What is the weighting filter for the manufacturers frequency graph, 1/3rd octave, 1 octave? Which leads to the question how relevant this graph is for matching in the first place?

[mathieujm]

Originally Posted by audio ergo sum

just some food for thought:

Manufacturer published on axis frequency responses tell the whole story of a frequency response of a mic, specially omnis ?

What is more problematic for stereo imaging: small deviations in frequency response or deviations in timing/phase between mic pairs ?

Anyone seen a HRTF for human ears? And their difference between left and right ear?

What is the weighting filter for the manufacturers frequency graph, 1/3rd octave, 1 octave? Which leads to the question how relevant this graph is for matching in the first place?

+1 for this interesting question.
JM

[Audiop]

Originally Posted by audio ergo sum

just some food for thought:

I'm hungry! :-)

Manufacturer published on axis frequency responses tell the whole story of a frequency response of a mic, specially omnis ?

Nope, but if you know the size of the capsule of an omni you get a good indication of its energy response. Sennheiser and most manufacturers with good reputation also publish polar graphs that shows curves off axis.

What is more problematic for stereo imaging: small deviations in frequency response or deviations in timing/phase between mic pairs ?

Mic's are min-phase devices so phase follows frequency. IOW if they have tightly matched FR they will have equally matched phase response.

Anyone seen a HRTF for human ears? And their difference between left and right ear?

I don't know what you're thinking of here but the HRTF and various individual differences is not relevant since all sounds we hear is filtered, live or recorded.

What is the weighting filter for the manufacturers frequency graph, 1/3rd octave, 1 octave? Which leads to the question how relevant this graph is for matching in the first place?

Depends on manufacturer. Since the ear has critical bands actually some smoothing can make sense. Also in the low range the response is always smooth and only when the wavelength gets short in relation to the capsule we start to see narrow deviations. Smaller capsules having smoother FR and better phase/time response generally speaking.

I'm planning a mic study for next month and hopefully you can get some answers from that.


/Peter

[audio ergo sum]

Originally Posted by Audiop

Nope, but if you know the size of the capsule of an omni you get a good indication of its energy response. Sennheiser and most manufacturers with good reputation also publish polar graphs that shows curves off axis.

I wish they also published not only the (heavily smoothed) polar graphs but also the frequency responses off axis. How many times lazy engineers have said: we should point the mic to the source? Why, it is an omni!

Originally Posted by Audiop

Mic's are min-phase devices so phase follows frequency. IOW if they have tightly matched FR they will have equally matched phase response.

In an stereo image the time shift needed to shift something from center to full side is app. 1200 µs (frequency dependent, not linear). The equivalent level shift necessary is app. 18 dB.
The phase shift needed for an equivalent of 120 µs (app. 20% image shift, not linear!) for 100 Hz is only 4.3 deg.

For the same image shift a level difference of app. 3 dB is needed...

While frequency dependent phase shifts are normal and the human ear is not sensitive to them for stationary signals (phase linearity is not constant phase!), they to the contrary are becoming an issue when differences in phase linearity are introduced in an stereo image through stereo mic pairs.

So my point is: Matching frequency responses is secondary but matching phase responses is primary. (And matching in general is not too important.)

Originally Posted by Audiop

I don't know what you're thinking of here but the HRTF and various individual differences is not relevant since all sounds we hear is filtered, live or recorded.

Depends on manufacturer. Since the ear has critical bands actually some smoothing can make sense. Also in the low range the response is always smooth and only when the wavelength gets short in relation to the capsule we start to see narrow deviations. Smaller capsules having smoother FR and better phase/time response generally speaking.

Very true. My point again is the huge difference in the frequency response between left and right ear for every individual. This makes the precise matching some prefer somewhat irrelevant IMO. As long as the tonal balance over wider frequency bands is sufficiently similar and the overall sensitivity is similar, there should be no problem.

We tend to underestimate the ability of the brain to compensate for frequency deviations. Hearing is believing, literally. Not many are aware of the simple fact, that simply the neuronal signal chain from the left ear takes longer travelling to the hearing center in the cerebral cortex than from the right ear. All these differences are compensated in the brain, we hear with the brain, not with the ears...

Actually my thesis regarding matching frequency response in stereo setups is the antithesis to frequency matching. In spaced AB setups every side should sound its individual best, small equalisations are not only possible but preferable. In a simplified nutshell that is my experience.

[Audiop]

Hi!

Originally Posted by audio ergo sum

I wish they also published not only the (heavily smoothed) polar graphs but also the frequency responses off axis. How many times lazy engineers have said: we should point the mic to the source? Why, it is an omni!

The angle of a omni would depend on the "voicing" of the mic. IOW if it's a direct field or diffuse field mic or something in between. For real close up miking on axis you want a flat curve for natural results. For distant set up you want a boost in the top of the mic (on axis) in order to compensate for the off axis roll of towards higher freqeuncies in the diffuse field.

So my point is: Matching frequency responses is secondary but matching phase responses is primary. (And matching in general is not too important.)

I think you missed the essential thing I wrote. :-)

Microphones have something that is called a minumum phase response. That means that phase follows frequency. If the frequency response is matched so is the phase response.

Very true. My point again is the huge difference in the frequency response between left and right ear for every individual. This makes the precise matching some prefer somewhat irrelevant IMO.

If you have different response in your left and right ear you are likely used to it and have calibrated your ears to that on live sounds. You still need good matching between left and right channels in stereophony in order not to spread the image and loose focus.

Of course an individual can adjust chanell balance and frequency response in all stages of recording and playback but that is not something that can be done universally.

As long as the tonal balance over wider frequency bands is sufficiently similar and the overall sensitivity is similar, there should be no problem.

It depends on the stereo technique but high performance speakers for example are usually matched to tight tolerances. For mic's, pure intensity stereo (coincident X/Y, Blumlein) would obviously be more sensitive to differences in FR between the chanells than the other extreme, AB stereo but I see no reason to accept errors when they can easily be avoided even if they are small. :-)

We tend to underestimate the ability of the brain to compensate for frequency deviations.

Tonality yes, stereophonic distortion, nope!

Actually my thesis regarding matching frequency response in stereo setups is the antithesis to frequency matching. In spaced AB setups every side should sound its individual best, small equalisations are not only possible but preferable. In a simplified nutshell that is my experience.

Do you mind clarify this part. It seems to go against what you mentioned earlier.




/Peter

[audio ergo sum]

Originally Posted by Audiop

The angle of a omni would depend on the "voicing" of the mic. IOW if it's a direct field or diffuse field mic or something in between. For real close up miking on axis you want a flat curve for natural results. For distant set up you want a boost in the top of the mic (on axis) in order to compensate for the off axis roll of towards higher freqeuncies in the diffuse field.

That's well known basic knowledge. As there is nothing 'natural' about recording in the first place, I try to avoid this term as much as possible. Also in practical applications often the brilliance boost is applied for the direct field mics and the room mics are added with a flat response.

The 'natural' frequency response of a mic membran is the on axis treble boost, depending on its diameter. Flat mics are artificially treated to be so. DPA does this acoustically with their 4006, all others electrically.

What would be ideal is a compensation of the off axis roll off only for off axis signals, but that is impossible. DPA actually has these cones, but I have never seen anyone using them for music. Anyone?

Originally Posted by Audiop

I think you missed the essential thing I wrote. :-)

Microphones have something that is called a minumum phase response. That means that phase follows frequency. If the frequency response is matched so is the phase response.

And you missed what I wrote :-) I understood you but tried to make the point, that when frequency response and phase response are matched, the small differences in phase responses are more critical than the differences in frequency responses. As you pointed indirectly out yourself here:

Originally Posted by Audiop

Tonality yes, stereophonic distortion, nope!

Also do you happen to know some real world phase deviations in microphones? I have no charts at hand at the moment.

[John Willett]

Good manufacturers will publish both the frequency response and polar pattern of an omni.

The two together tell you what the off-axis frequency response will be like.

Any omni (other than sup-miniature tie mics) will go directional at higher frequencies - this is purely down to the size of the mic. body in comparison to the wavelength of the sound.

[Audiop]

Originally Posted by audio ergo sum

That's well known basic knowledge. As there is nothing 'natural' about recording in the first place, I try to avoid this term as much as possible.

Ok, time for semantics again!
A source have some tonal characteristics and for that to be captured on reproduced with HIgh FIdelity we can't allow linear distortion of significant amounts.

If a recording and playback chain can give the same timbre and tonality as in real life I do call that reproduction 'natural'. If there is a clear proximity effect or presence boost that is audible I call it colored and less natural.

Also in practical applications often the brilliance boost is applied for the direct field mics and the room mics are added with a flat response.

I do not like a brilliance boost or presence boost or any boost on the direct field mic's as to me that gives the unnatural colorations mentioned above. OTOH a room mic often gives a better picture of the ambience if it has a flat curve in the power reponse curve as opposed to a flat on axis curve since sound is comming in from all angles typically.

Schoeps for example has omnis with three different voicings (AFAIK) and they recommend these in applications as outlined by my text above.

The 'natural' frequency response of a mic membran is the on axis treble boost, depending on its diameter. Flat mics are artificially treated to be so. DPA does this acoustically with their 4006, all others electrically.

Yes, the exception though being ultra small omnis for HF measurements that has it's "baffle step frequency" or "4pi > 2pi transition point" just above the audioband.

What would be ideal is a compensation of the off axis roll off only for off axis signals, but that is impossible.

Yes unfortunately!

And you missed what I wrote :-) I understood you but tried to make the point, that when frequency response and phase response are matched, the small differences in phase responses are more critical than the differences in frequency responses.

Ok, great, we're comming closer to and understanding!

I have actually not looked into to it in detail. I'd say it depend on the situation and also I'm curious on how you came to that conclusion?

Also do you happen to know some real world phase deviations in microphones? I have no charts at hand at the moment.

No, but if we assume totally flat response in the passband and a first order lopwpass function with -3dB 200kHz we would still have several degree of phase shift at 20kHz if memory serves me well. There will always be significant measurable phase shift in mics and speakers and what we really would need for transparency in that regard I'm clueless about.

I tend to like gear that has good bandwith though but lots of gear is described as musical and natural by some people even when there's a roll off setting in even before 20kHz. Ribbon mic's and old LDC's for example.


/Peter

[audio ergo sum]

Originally Posted by Audiop

I do not like a brilliance boost or presence boost or any boost on the direct field mic's as to me that gives the unnatural colorations mentioned above. ...

You must not like then the typical Decca-Tree sound with M50s, a very common recording technique, acclaimed for its good representation of timbre.

Originally Posted by Audiop

I have actually not looked into to it in detail. I'd say it depend on the situation and also I'm curious on how you came to that conclusion?

See above where I'm talking about time vs amplitude differences for interchannel signals.

No, but if we assume totally flat response in the passband and a first order lopwpass function with -3dB 200kHz we would still have several degree of phase shift at 20kHz if memory serves me well. There will always be significant measurable phase shift in mics and speakers and what we really would need for transparency in that regard I'm clueless about.

Linear frequency dependent phase shift is normal. IIRC there are speakers available with linear AND constant phase by KS Digital.

It is only the differences and nonlinearities between interchannel signals/stereo imaging that are interesting. And overlooked IMO, coming back to the original topic of the thread.

[Audiop]

Originally Posted by audio ergo sum

You must not like then the typical Decca-Tree sound with M50s, a very common recording technique, acclaimed for its good representation of timbre.

I'm sure I've heard some of it but I don't know when and for sure so basically, I wouldn't know.

That said those mic's were/are placed in the diffuse field. Direct field is close to the source where the direct sound dominates which is not the case with a Decca tree in a reverberant hall. So, I'm not surprised that a top end lift is prefered in such a set up since flat omnis (direct field omnis flat on axis) would sound like a bag of lambs wool!

See above where I'm talking about time vs amplitude differences for interchannel signals.

I don't see an explanation to your claim in the Decca tree.

Linear frequency dependent phase shift is normal. IIRC there are speakers available with linear AND constant phase by KS Digital.

Well, transducers is still basically min-phase devices and digital processing of the signal that is fed to the speakers don't change that fact.

It is only the differences and nonlinearities between interchannel signals/stereo imaging that are interesting.

What other differences could there be?


/Peter

[audio ergo sum]

Originally Posted by Audiop

I'm sure I've heard some of it but I don't know when and for sure so basically, I wouldn't know.

That said those mic's were/are placed in the diffuse field. Direct field is close to the source where the direct sound dominates which is not the case with a Decca tree in a reverberant hall. So, I'm not surprised that a top end lift is prefered in such a set up since flat omnis (direct field omnis flat on axis) would sound like a bag of lambs wool!

Nope. Decca tree typically is around or mostly within critical distance, so direct sound is stronger than diffuse, particularly for the boosted high frequencies. In a typical concert hall it is usually within 3 m or closer to the nearest instrument.

Originally Posted by Audiop

I don't see an explanation to your claim in the Decca tree.

Misunderstanding. Above but not in the same post, some posts before.

[Audiop]

Originally Posted by audio ergo sum

Nope. Decca tree typically is around or mostly within critical distance, so direct sound is stronger than diffuse, particularly for the boosted high frequencies. In a typical concert hall it is usually within 3 m or closer to the nearest instrument.

And many more meters to many of the instruments which likely puts them in the reverberant field.

MKH8020 (basically flat in the directfield) is relatively small (at least compared to M50) and for example when recording grand piano in a hall they sound much darker than QTC1 at only a couple of meters. QTC1 is flat on axis as well but as you know the small size pick up more energy off axis in the top octaves.

Put the M50 at 0.5 meter or 1 meter and it will sound very bright (unnatural ;-) due to the boost in the top decade, as would any diffuse field omni with a 5dB boost from 2k-20k.

Would be fun to try them out but I'm afraid I would have to sell my car AND my house to get money for a pair. I'm afraid I'll have to settle for KM183.




/Peter

[audio ergo sum]

Originally Posted by Audiop

And many more meters to many of the instruments which likely puts them in the reverberant field.

...

It's a little more complex than that. Even for woods and brass an M50 of a Decca tree is not clearly in the diffuse field. This is off topic but think about parameters like:
-directionality of the instruments (depending on frequency)
-the same for the Mic
-time domain
-the resulting actual critical distance is therefor different for every source and for every frequency at any moment at any point in the room --> chaos (and teachers, authors and scientists don't like chaos, it's the realm of the empirical study)
-critical distance is only defined for stationary sound field, for the attack/first wavefront the critical distance is always infinite !

etc.
(critical distance is the distance from a sound source in a room, where direct and reflected/diffuse energy are equal.)