MS stereo asymmetry

[The Listener]

I occasionally encounter the following problem. When I record single instruments with MS technique I get asymmetrical stereo picture after encoding (eihter by hand - duplicating S track, pan both S tracks L-R and reversing phase on the right channel or by Waves MS matrix inserted on a group to which I send M and S track, panning M - L and S - R). The stereo picture usually leans a little to the left. Why is that so and did someone else had the same problem? I am using Shure SM81 cardioid as MID and Gefell UM70 in figure 8 as SIDE. I am doing all correct as I checked, but unless I really lower the SIDE track volume, I get the asymmetrical stereo picture.
Maybe it is something wrong with Gefell's figure 8 characteristic?

[John Willett]

Originally Posted by The Listener

Maybe it is something wrong with Gefell's figure 8 characteristic?

Well - the published polar diagram seems to be identical front and rear - it certainly looks OK.

I notice the mic. has a round body:

Are you *sure* you have it set up perfectly - if the mic. is rotated in the mount slightly it could account for a lop-sided L/R result (eh: left pointing down slightly and right pointing up) ??

I hope this helps.

[d_fu]

Turn the Gefell by 180° - what happens?
Make some mono test recordings with both sides - do they sound the same, are the levels ok?

[_Morph_]

Let's get down to the basics.

There are two signal sources: one from the cardioid microphone and one from the figure-of-eight. So none of them should give you unbalanced results, since both are mono.

When you mult and flip the polarity of the figure-of-eight on the right channel, the signal is actually the same, only reversed, so there should be no problems here as well.

Finally, when you combine the figure-of-eight tracks (left and right) with the centered cardioid, there will be summing involved, thus giving you that stereo spread, where things appear from left to right. This sum is indeed less balanced - and rightfully so - because the cardioid signal is being summed with two different signals (from that cardioid signal's point of view). If everything was done correctly, it's normal for the stereo soundfield to appear a little bit skewed, if the microphone placement was skewed from the start. Maybe that's what happened...

[The Listener]

Originally Posted by d_fu

Turn the Gefell by 180° - what happens?
Make some mono test recordings with both sides - do they sound the same, are the levels ok?

I just did the test... since I am used to some figure 8 mics (like chinese ribbon T-bone RB500 to have non perfect response on both sides) I thought this is normal.

This is how front and back side sound in a quick test:

gefell figure 8 test

gefell omni test

Omni seems fine on both sides, but in figure 8 the front has much more high frequencies and definition, while the back side sounds dull. I guess this is not normal for this mic? Any ideas what this defect (?) might be. After that I'll write to Gefell.

btw - it is a vintage MV691 + UM70, not the new UM70S version.

I guess this causes the asymmetrical MS picture...

[The Listener]

Originally Posted by _Morph_

Let's get down to the basics.

There are two signal sources: one from the cardioid microphone and one from the figure-of-eight. So none of them should give you unbalanced results, since both are mono.

When you mult and flip the polarity of the figure-of-eight on the right channel, the signal is actually the same, only reversed, so there should be no problems here as well.

Finally, when you combine the figure-of-eight tracks (left and right) with the centered cardioid, there will be summing involved, thus giving you that stereo spread, where things appear from left to right. This sum is indeed less balanced - and rightfully so - because the cardioid signal is being summed with two different signals (from that cardioid signal's point of view). If everything was done correctly, it's normal for the stereo soundfield to appear a little bit skewed, if the microphone placement was skewed from the start. Maybe that's what happened...

I know MS technique quite well.

No, I was especially careful to be at the right angle from both sides of S mic, not leaning more to any side - I was dead in the center in front of the MID mic - a little further away to be sure that some strange projections would not cause that asymmetry - I tested with ethnic flute, kalimba, frame drum - and the result was the same.

here is the MS example with leaning towards left:

kalimba MS

It still sounds decent, but I am puzzled by this asymmetry.

edit: The asymmetry is consistent - left channel always louder - I noticed that already once in the past when recording acoustic guitar - again - left channel slightly louder.

[_Morph_]

Don't forget that bass frequencies usually have more SPL than higher frequencies. This is accurately translated whether it's a kalimba, an acoustic guitar, etc. (I don't mean to sound judgemental, just trying to cover everything that might lead to this result).

Also, there's nothing preventing you from panning the Middle signal a little bit to the right, to compensate for this. The mono compatibility is preserved and might help balancing the stereo sound field...

[The Listener]

Originally Posted by _Morph_

Don't forget that bass frequencies usually have more SPL than higher frequencies. This is accurately translated whether it's a kalimba, an acoustic guitar, etc. (I don't mean to sound judgemental, just trying to cover everything that might lead to this result).

Also, there's nothing preventing you from panning the Middle signal a little bit to the right, to compensate for this. The mono compatibility is preserved and might help balancing the stereo sound field...

I tried panning MID towards right and I can balance the stereo picture that way a little, but I think there is a problem with that mic.

I also notice that the asymmetry is bigger with high frequency sounds - notice how it shifts with high hits on the frame, while bass hits stay more in the center:

MS with MID panned slightly right to correct the "anomaly"

MS with no "correction" of MID mic


Again - I was NOT standing left from the MS mic position.

[_Morph_]

Nice work!

I agree: by panning the middle signal, it became more balanced. I think it sound really good, fow what it's worth. The stereo is, well, stereo, and the mix sounds big and alive. The hand slapping and rubbing is coming from the left, which tells me that the microphones were relatively close to the instrument, which increases the differences between left and right channels, but it's really nice this way.

[QUOTE=The Listener;4483591]I also notice that the asymmetry is bigger with high frequency sounds - notice how it shifts with high hits on the frame, while bass hits stay more in the center:[QUOTE]

That's the nature of acoustics. Lower frequencies tend to be more omnidirectional, and higher frequencies are more directional. The only way to I can think of counteract this is by recording in mono or lowering the Side signal in the mix.

[The Listener]

[QUOTE=_Morph_;4483676]Nice work!

I agree: by panning the middle signal, it became more balanced. I think it sound really good, fow what it's worth. The stereo is, well, stereo, and the mix sounds big and alive. The hand slapping and rubbing is coming from the left, which tells me that the microphones were relatively close to the instrument, which increases the differences between left and right channels, but it's really nice this way.

[QUOTE=The Listener;4483591]I also notice that the asymmetry is bigger with high frequency sounds - notice how it shifts with high hits on the frame, while bass hits stay more in the center:

That's the nature of acoustics. Lower frequencies tend to be more omnidirectional, and higher frequencies are more directional. The only way to I can think of counteract this is by recording in mono or lowering the Side signal in the mix.

Oh, yes, I completely overlooked the lower freq being more omnidirectional, harder to locate... I was rather close with frame drum - 20cm, but I was in the dead center - it should not sound so much on the left (without correction). With correction (panning the MID mic) it sounds ok, yes. Thanks for the ideas, anyway.


I don't know... I wrote to Gefell if the figure of 8 in UM70 should have such difference between front and back response...

btw - it was not mix - it was just a quick banging on the drum for test purposes - what causes "big and alive" sound is the DAV BG1U I am using and the Gefell mic (and also Shure SM81 is very decent).

I just want to settle this little annoyance.

[leddy]

Wouldn't the room be the likely cause of these issues? I just did an MS recording where the bass jumps between left and right a bit on a few notes. It was a live recording in a club and the mics were 10 feet from the bass. No reason it should move in the stereo image except for different notes exciting different resonances in the room, no?

[The Listener]

Originally Posted by leddy

Wouldn't the room be the likely cause of these issues? I just did an MS recording where the bass jumps between left and right a bit on a few notes. It was a live recording in a club and the mics were 10 feet from the bass. No reason it should move in the stereo image except for different notes exciting different resonances in the room, no?

The thought crossed my mind, too. But the "anomaly" is consistent - always left channel and quite obvious - like one would be standing left of the mics.

[boojum]

Run an ORTF pair in parallel and see what you hear on the ORTF's. ;o)

[Storyville]

Originally Posted by leddy

Wouldn't the room be the likely cause of these issues? I just did an MS recording where the bass jumps between left and right a bit on a few notes. It was a live recording in a club and the mics were 10 feet from the bass. No reason it should move in the stereo image except for different notes exciting different resonances in the room, no?

That's normal. Any stereo room capture on a bass will have this effect. XY, AB, spaced pair, MS, whatever.

As for differences on the front and back of the mic, ask Gefell. Perhaps you have a more reflective room on the left side?

[iluvatar]

How would differences between the front & back of a Fig-8 mic cause something to pan one way or another? It's not a stereo mic. It's a mono mic. The front and back are added together inside the microphone.

-Dan.

[ghellquist]

Originally Posted by iluvatar

How would differences between the front & back of a Fig-8 mic cause something to pan one way or another?
-Dan.

Because then it does not act as a figure 8 mic. If, say, one side removes high frequency sounds, then high frequencys will be attenuated from that side in the stereo mix.

Gunnar

[The Listener]

Originally Posted by ghellquist

Because then it does not act as a figure 8 mic. If, say, one side removes high frequency sounds, then high frequencys will be attenuated from that side in the stereo mix.

Gunnar

Exactly my thoughts.

The front and back are quite different - the back being much duller. It could not be the difference in reflectiveness of each side of the room.

listen: gefell figure 8 test (first front, after short pause the back side)

I wrote to Gefell and await the answer.

[John Willett]

Originally Posted by The Listener

Exactly my thoughts.

The front and back are quite different - the back being much duller. It could not be the difference in reflectiveness of each side of the room.

listen: gefell figure 8 test (first front, after short pause the back side)

I wrote to Gefell and await the answer.

The polar-pattern shows the front and back to be identical - if the rear is duller then it looks like a faulty mic. that needs fixing (probably a new capsule).

[Nobilmente]

I have used the UM70 on occasion as the side mic in M&S, but not been aware of the problem you have encountered. Could it be a dirty connection on one of your leads, what happens if you put a fresh cable on there?

What happens if you speak to the front and then the rear of the UM70 alone and removed from the M&S set up, are you aware of a difference?

There is a difference with a number of ribbon microphones, just something to live with, but the front and back of the UM70 should be the same.

[The Listener]

Originally Posted by Nobilmente

I have used the UM70 on occasion as the side mic in M&S, but not been aware of the problem you have encountered. Could it be a dirty connection on one of your leads, what happens if you put a fresh cable on there?

What happens if you speak to the front and then the rear of the UM70 alone and removed from the M&S set up, are you aware of a difference?

There is a difference with a number of ribbon microphones, just something to live with, but the front and back of the UM70 should be the same.

Yes, there is a major difference. Gefell said to send them the mic to service, it should be the same on both sides. I'll just do that.

[zoom]

Every fig8 microphone is slightly asymmetrical to its corresponding angles but I would say this is not relevant in practice.
Usually I recommend to configure a M/S setup with small diaphragm mics but in principle your Gefell UM70 large diaphragm should work in fig-8 position as an side-microphone too. The only problem is the laeger distance of the mics in the array which causes a problem for the matrix in the higher frequencies.

Some tipps for you are already mentioned and it seems to be an issue for the service. Maybe there is some dirt on one diaphragm which causes a decreased sensitivity?
Ask Gefell to deliver some measurement data together when they return your microphone. (once I've got these data on request, when I had 2 mics in service)



If you are interested in the theme of measuring the symmetry of a fig8 microphone, I had posted some stuff in the geek-departement of this forum some time ago. Probably it was too geeky, because I've got no responses :-)
this is the link:
Symmetry of a pressuregradient microphone (fig.8)
and you see, also a Schoeps MK8 not perfectly symmetrical but this is only interesting for the measurement engineers ;-)

cheers
stefan

[iluvatar]

Originally Posted by ghellquist

Because then it does not act as a figure 8 mic. If, say, one side removes high frequency sounds, then high frequencys will be attenuated from that side in the stereo mix.

Gunnar

No. You're still thinking of it as a stereo mic. It's not a stereo mic; it's a mono mic. The front and back are summed inside of the microphone. For M/S, you take that one signal and make a copy of it: the "left" and "right" channels of the "side" are exactly the same except for their polarity.

If we were talking about paint, imagine one side of the mic as a pint of black and the other side as a pint of white. The mic mixes them together into a quart of grey then sends it off to the preamp. M/S processing doesn't separate the two sides of a fig-8 mic (to keep the analogy going; it doesn't separate the grey into black & white). M/S is essentially fake stereo - it's not an accurate l/r image. Pointing an M/S setup off to the side of an instrument won't make the center image pan left or right, it'll make the instrument sound wide and off to both sides.

-Dan.

[ghellquist]

Originally Posted by iluvatar

No. You're still thinking of it as a stereo mic. It's not a stereo mic; it's a mono mic.
-Dan.

Of course the S mic is a mono mic. No person has even implied differently.

I guess there are different ways to think about things. Now, as an example, if I take my Sennheiser MKH30 it has exactly one diaphragm and has a pretty good figure 8. Frequency response is same both back and front. This makes it a good choice for the S mic.

On the other hand we have a faulty mic that has quite different frequency response on the front and the back. This might make it very useful for a lot of things, but makes it into a lousy S mic in i MS setup.

The effect of radically different frequency response can be heard in that things that are "smack in the middle" in reality will end up panned in the decoded MS stereo landscape. Different amounts depending on the frequency of the "things" we record.

Next thing is that we know that this actual mic, in contrast to the MKH30, has two diaphragms. And these gets summed together in different ways or proportions to get an omni response, a cardioid response or a figure 8 response. This is pretty standard mic technology for the last 50 years or so. So the very obvious conclusion is that something is wrong inside the mic. And the very obvious solution is to send it back to its maker for tender care and repair. Once it has been repaired we can expect this mic to be a good or even excellent choice as S mic in a MS setup, alas not right now.

Gunnar

[_Morph_]

Originally Posted by iluvatar

Pointing an M/S setup off to the side of an instrument won't make the center image pan left or right, it'll make the instrument sound wide and off to both sides.

-Dan.

Fortunately, no. The instrument will pan accordingly. At least in my experience. And the mp3s posted here are proof of that. The stereo sound field is accurately represented, and the Side microphone adjusts the "width" of that sound field.

[John Willett]

Originally Posted by zoom

Every fig8 microphone is slightly asymmetrical to its corresponding angles .......

Where did you get this idea?

If the acoustic conditions are identical both sides of the diaphragm, then the rear lobe will be identical to the front lobe.

The Sennheiser MKH 30 is truly symmetrical both sides, so is the Neumann KM 120, AK 120 and KK 120 - all these are single diaphragm figure-8 microphones with identical acoustic conditions both sides of the diaphragm.

Yes, the Schoeps MK 8 is not symmetrical (according the the published polar pattern on the website) and shows attenuation on the 16kHz plot on the rear lobe.

[iluvatar]

Originally Posted by ghellquist

Of course the S mic is a mono mic. No person has even implied differently.

Actually, you did when you said, "If, say, one side removes high frequency sounds, then high frequencys will be attenuated from that side in the stereo mix."

If it's not a stereo mic, how does that HF attenuation get applied to one copy of the Side channel, but not the other?

There are elements of psychoacoustics and phase cancellation between the M & S mics that can cause the stereo image to appear to shift one way or another, but it's not as simple as saying that since the darker back-side of the Fig-8 mic is pointing to the right of the sound source, then the right side of the final mix will also be darker. That darkness is going to be applied to both the Left and the Right and how it affects the final mix when combined with the Mid mic is anybody's guess.

I've got a session open right now where I recorded an M/S pair and a spaced pair at the same time, with the sound source moving around in the coverage area. There are bits where the M/S pair sounds like it's panned a bit left or right (and even the meters show an imbalance), but it's not predictable and it doesn't coincide with what actually happened (i.e. if something happened off to the right, the image doesn't necessarily pan right).

I guess there are different ways to think about things. Now, as an example, if I take my Sennheiser MKH30 it has exactly one diaphragm and has a pretty good figure 8. Frequency response is same both back and front. This makes it a good choice for the S mic.

On the other hand we have a faulty mic that has quite different frequency response on the front and the back. This might make it very useful for a lot of things, but makes it into a lousy S mic in i MS setup.

That's because when the front and back have different responses, they don't cancel out completely and the null (where the Mid mic is pointing) isn't as deep as it should be and thus there's more overlap between the coverage of the two mics than there should be. Actually, that could account for the panning - one copy of the Side signal is constructively interfering with the Mid, while the other is destructively interfering with it. I bet that if the OP reversed the polarity on his Side tracks, then the perceived panning would reverse.

-Dan.

[zoom]

Originally Posted by John Willett

Where did you get this idea?

If the acoustic conditions are identical both sides of the diaphragm, then the rear lobe will be identical to the front lobe.

The Sennheiser MKH 30 is truly symmetrical both sides, so is the Neumann KM 120, AK 120 and KK 120 - all these are single diaphragm figure-8 microphones with identical acoustic conditions both sides of the diaphragm.

Yes, the Schoeps MK 8 is not symmetrical (according the the published polar pattern on the website) and shows attenuation on the 16kHz plot on the rear lobe.

ok, is there anyone near Berne / Switzerland who can borrow me a MKH30? I want to measure it by myself...
Usually I don't trust smoothed manufacturer graphs and you have to be careful if you have a look to the polarpattern, because these are normalized curves in simplified octave intervals. Complete angle dependend frequency responses tell us much more in that case.
The deviation of 0° and 180° around 16kHz of the Schoeps MK8 is caused probably due to the mechanical fixture of the diaphragm in the capsule. There is a little dip in the 180°response I have measured (see here).
If anyone want to listen to the MK8 polarpattern: a wav-file recorded with a noise signal on a turntable in an anechoic chamber, you can analyze both lobes by yourself at the beginning (= 0°) and the end (=180°) of this track.

...and - I have to point out again - what I have measured is only of academical interest. It is almost neglectible in practice (under the condition you use single (small)-diaphragm fig8 microphones for your M/S setup).

(probably we are a bit off-topic now ;-))

[The Listener]

Thanks for additional ideas and info.

UM70 should have the same response on both sides, so says Gefell. If one side is radically duller it is an error that should be serviced.

I know how to use MS technique in practice, but as I see I am not the only one confused about the more obscure technical details. I know the basic theory of it, but I have no idea what it means if figure 8 mic is asymmetrical in MS set-up - obviously it messes with the panning... How? I also thought what does it matter - one mono mic mixing with another one through MS matrix, so what does it matter if one side of figure 8 is different, but obviously it is. Probably someone knows how to explain it theoretically.

[John Willett]

Originally Posted by The Listener

Thanks for additional ideas and info.

UM70 should have the same response on both sides, so says Gefell. If one side is radically duller it is an error that should be serviced.

I know how to use MS technique in practice, but as I see I am not the only one confused about the more obscure technical details. I know the basic theory of it, but I have no idea what it means if figure 8 mic is asymmetrical in MS set-up - obviously it messes with the panning... How? I also thought what does it matter - one mono mic mixing with another one through MS matrix, so what does it matter if one side of figure 8 is different, but obviously it is. Probably someone knows how to explain it theoretically.

OK...................................

MS works in the same way as a fig.8 and omni wired in parallel = cardoid.

The cardioid, for example, mid mic and the side mic are added together = a super-cardioid pointing half left.

If you polarity reverse the fig.8 you get a super-cardioid pointing half right.

So - polarity reversing the fig.8 means that the rear lobe becomes the front lobe for the right mic.

If the rear lobe lacks treble this will affect the response of the right stereo channel.

This is why the fig.8 needs to be truly symmetrical both sides of the diaphragm.

As far as I know there are only 3 condenser fig.8 mics that are single diaphragm (all the others being made from twin diaphragm mics of back-to-back cardioids).

These three are: Sennheiser MKH 30, Neumann AK/KK 30 and the Schoeps MK 8.

The Sennheiser MKH 30 is a symmetrical capsule mic. with identical active front and back plates.

The Neumann AK/KK 120 (same head with different threads for KM 100 and KM-D series) is also symmetrical with identical front and back plates, but only the rear plate is active, the front being there for acoustic purposes to make the front and rear acoustically the same.

The Schoeps MK 8, as I understand, has a back plate but not an identical front plate - this accounts for the slight HF droop on the rear lobe which is shown in the published polar-pattern (NB: it only shows on the 16kHz plot and not really on lower frequencies). In practice, because the mid mic. will be equal on both left and right channels, the HF droop in the right will be extremely small and probably not really noticeable in practice.

I hope this helps.

[ghellquist]

It is not too difficult to understand really.

Assume that we have a perfectly good M mic (any type really, but I will use an omni as example to simplify things)

Assume that we have a really bad S mic that is more or less deaf towards right. It can only hear things to the left. This is of course extreme, in the real world it might only dampen things by say 10dB, but that is enough to skew the stereo panorama.

The two signals from the mics are called M and S.

The operation we do in decoding is described by a very simple mathematical formula.

L = M + S
R = M - S

Now we put a source all the way to the left. Remember that the signal will be heard both by the omni M mic and the faulty S mic. Lets keep things really simple and say that the signal from both mics is 1Volt.

What we get is:
L = M + S = 1 + 1 = 2
R = M - S = 1 - 1 = 0.
So clearly this signal is heard fully panned left.

Now we put the source to the right instead. Remember that the S mic in this is example is deaf to the right so the signal from that one is 0.
L = M + S = 1 + 0 = 1
R = M - S = 1 - 0 = 1
So clearly this signal is heard panned straight in the middle. The faulty S mic has destroyed the panning as things will only be panned in the middle or to the left, never to the right.

Small note: if the S mic had been working as a figure 8 the signal would have been -1 Volt, fully out of phase with the M signal. In that case we get:
L = M + S = 1 + (-1) = 0
R = M - S = 1 - (-1) = 2
which gives us the expected full pan to the right.

Hope this simplified example helps. Not meant to be the full details.

// Gunnar