Skip to main content

Hi there,

I've been using this setup more and more in the last year and I thought I'd share a little thing I found.
First for those who doesn't know how M/S capture is done :
(link removed by UA)

Now my secret : It doesn't work ! (well not perfectly that is)

Got your attention ?

Let me explain, when recording in M/S the 2 mics are recorded in phase and the figure of 8 is then splitted on 2 tracks and one goes out of phase. That out of phase signal added to the mid mic will cancel some frequencies and the result is a stereo signal but never equal on both sides. Every time I do M/S recording I find it good for creating a stereo image but if you put the level of the Side tracks too high, the instruments starts to shift to one side of the stereo image.

Solution ?
I found that if I delay the mid mic a few sample (5-10samples) that resolves the problem.
what happen is that if you move the mid signal so it's not out of phase with any of the Side tracks the M/S array is now more effective at any ratio between Mid and Side...

I share this with you so we can experiment and maybe prove me wrong or find that it creates other problems..
Or even get the technic better

So far I had good results with that..

One thing that still occure is that in mono the Side tracks will cancel, but that's the trade of M/S mic

Comments

MrEase Mon, 11/09/2015 - 08:44

I don't have this problem! However mic placement is critical and I always set up my mic's one above the other (with the upper one upside down - I use a Rode NTK for mid and a K2 for side). This ensures that the X-Y position of the diaphragms are identical with only a couples of inches or so difference in height and neither mic is in the "shadow" of the other. Are you doing this or placing the mic's as close as possible alongside each other? This could explain why you have to shift by a few samples.

Also, I'm confused by your last statement. I don't see the side tracks cancelling in mono as any sort of trade off as that is exactly what you want in mono! You're simply left with the Mono (which is the mid signal) and not the side....

EDIT: I just looked at the link you provided and my point on mic placement was already made!

MrEase Mon, 11/09/2015 - 10:07

The key here is that the theory dictates that the mics should occupy the same place which is clearly impossible. To avoid any inevitable delays the trick is to place one above the other. Putting them side by side will always create a small timing difference which, as you have found, will need to be corrected even if it's only a few samples. Otherwise, you will see exactly the problem you've encountered.

Of course the two side tracks will cancel but that's exactly what you want to produce mono. It leaves only the mid recording which is exactly what you want for mono irrespective of the level. The side signal only contains the stereo information so it must be removed to produce mono. What I don't see is why you see that as a trade off? All this is irrelevant if you're talking about a mixed down signal as it is then conventional stereo anyway.

Boswell Mon, 11/09/2015 - 10:25

M-S works provided that (a) the M and S microphones are precisely co-incident in at least two axes (e.g. the two horizontal axes) and (b) you use microphones of the same transducer principle for the M and the S (e.g. both ribbon or both condenser). If either of these two rules is broken, you get improper phasing of the M and S channels, which delaying one channel with respect to the other may appear to correct over a limited range of frequencies. I have on occasion witnessed the "same transducer type" rule being broken by recording professionals who should have known better. Note that figure-8 patterns on condenser microphones are often generated internally by using two opposing cardioid transducers, and I have always been suspicious that the result is never as accurate in phase terms as a velocity microphone such as a ribbon.

If you feed the M-S pair of microphones through a proper M-S decoder, the input to the mix is two channels carrying X and Y information, behaving exactly as though they came from a 90 degree crossed pair of microphones. In the absence of a hardware M-S decoder, you can emulate a decoder by taking the raw S microphone signal into two mixer channels panned hard L and hard R respectively and inverting the polarity of the channel panned hard R. The M microphone signal is taken into a standard mono channel, panned centre. The faders for the two S-channel components must move exactly together. When the levels of the M channel and the two S channels are brought up, you hear an image with width. The "correct" width is when the level of the two S channels is 6dB less than the M channel, since there are two S channels going into the mix and only one M channel. The width of the image can be varied by changing the level of the S channels from zero (mono) through the correct point (true stereo) to wider than reality. Going more than about M+6dB gives a very non-realistic image, but could be used as an effect in rare instances (such as live theatre) where mono compatibility is not essential.

pcrecord Mon, 11/09/2015 - 11:27

MrEase, post: 433739, member: 27842 wrote: Of course the two side tracks will cancel but that's exactly what you want to produce mono. It leaves only the mid recording which is exactly what you want for mono irrespective of the level. The side signal only contains the stereo information so it must be removed to produce mono. What I don't see is why you see that as a trade off?

It's a trade off if you want to have a nice mono mix compatible. Don't get me wrong, I know most listeners are gonna listen in stereo.. But If I mix the Sides tracks very strong and the mid track at lower volume.. that instrument will be weak once in mono... (not a big trade of.. but something to be aware.. ) Hope I'm more clear.. ;)

Boswell, post: 433740, member: 29034 wrote: M-S works provided that (a) the M and S microphones are precisely co-incident in at least two axes (e.g. the two horizontal axes) and (b) you use microphones of the same transducer principle for the M and the S (e.g. both ribbon or both condenser).

I'm not saying there is no chance I'm not doing it wrong..
I'm using a brand new SM81 as mid and a ksm44 as side. Honestly I'm surprised you don't see this. We are using 2 mics at which the sound will arrive at the same time and we reverse the polarity on on side.. To me it's obvious, if we mix them at equal volumes that the reverse side will cancel itself with a part of the mid signal. . . therefor you'll get a signal stronger to one side..

Now Bos said the exact information that I needed, saying the the sides should be at least 6db lower in volume. But what if you want them to be at equal or higher volume ?? That's when my little trick is comming handy ! ;) Ok it might not sound realistic, but in a world of plastic as today, many exagerated stereo image are used these days ;)

bouldersound Mon, 11/09/2015 - 11:36

pcrecord, post: 433744, member: 46460 wrote: We are using 2 mics at which the sound will arrive at the same time and we reverse the polarity on on side.. To me it's obvious, if we mix them at equal volumes that the reverse side will cancel itself with a part of the mid signal. . . therefor you'll get a signal stronger to one side..

This is what's supposed to happen if the source is not centered on the mid mic's axis. If the source is centered then it should be centered in the image when the mid mic is at all present, and practically disappear when the mid mic is lowered completely. If that's not how it's happening then there's a fault somewhere in the signal chain, or maybe it's a room effect, a reflection different/stronger from one side of the mic array.

pcrecord Mon, 11/09/2015 - 11:43

bouldersound, post: 433746, member: 38959 wrote: This is what's supposed to happen if the source is not centered on the mid mic's axis. If the source is centered then it should be centered in the image when the mid mic is at all present, and practically disappear when the mid mic is lowered completely.

That's one thing I havent considered !! :confused:
While having the problem, I'm recording an acoustic guitar and the axis is near the 12th fret so it's not really centered.. My bad !!
I'm gonna test with another instrument and place it in the center and see if the problem still occurs

paulears Mon, 11/09/2015 - 13:24

It would never occur to me to use M/S close in. My own experiments over the years have always shown it's not nice close in, distorting the natural left/right sound field as the inverse square law makes big differences in level where they shouldn't be. M/S works for me for distant mic techniques on sound sources that are really wide - so sort of string quartet/small vocal group/at least 2m distance grand piano. M/S on a solo guitar seems to me to be a poor choice of technique because the instrument has a limited width in real life, and close miking it produces an instrument that in normal M/S balance is far too wide, so you drop the side channel and get an almost mono recording. If you use two mics to get the usual sound hole and fingerboard sound, then I pan them fairly close, maybe ten to two, and just adjust the balance between them for a nice tone with a bit of spread. Guitars just don't have the size to be a suitable sound source for M/S to capture properly.

Just my view, but just what would the side capsule be actually hearing, pointing off axis so much - a very thin and weedy sound.

pcrecord Mon, 11/09/2015 - 16:35

Ok guys, I made some samples :
First it's the accoustic played at 6 inches of the M/S array then at 2 feet than an electric amp with 1 x 12'' at 3 feet.
On each parts I automated the side tracks to go from zero db to the same level of the mid track.
to me, they all are more powerfull on the left speaker at the end

[MEDIA=audio]http://recording.or…
 

And then with a delay on the mid channel, they sound more equal each sides
[MEDIA=audio]http://recording.or…
 

Attached files MS test mid and side phase aligned_1.mp3 (2.4 MB)  MS test mid and side with delay.mp3 (2.4 MB) 

MrEase Tue, 11/10/2015 - 03:47

pcrecord, post: 433744, member: 46460 wrote: It's a trade off if you want to have a nice mono mix compatible. Don't get me wrong, I know most listeners are gonna listen in stereo.. But If I mix the Sides tracks very strong and the mid track at lower volume.. that instrument will be weak once in mono... (not a big trade of.. but something to be aware.. ) Hope I'm more clear.. ;)

You seem to be assuming that once the M-S is mixed down it still retains it's M-S characteristic. This is not so, once you mix down the signal becomes conventional L-R stereo. If you find that the "mono" weakens a track that is a characteristic of your mix and is not due to the M-S recording technique.

It is quite easy to produce conventionally recorded mixes with the same problem and as you say, you are trying to get a particular stereo effect. It is this effect you're trying for that is causing the mono problem and not the M-S recording per se.

paulears Tue, 11/10/2015 - 03:54

I've totally lost the plot. The results are interesting, perhaps a bit predictable, but essentially, an effect - like top and bottom miking of a snare. It's a valid technique of course, and it's got a place, but it's a morph of M/S, in the same way recording a guitar close in with A/B is. These techniques are designed and tweaked to give realism and accuracy - hence all the talk about moving the capsules in space, occupying the same spot etc etc. These are essentials in recreating accurate and realistic sound fields. Using them like this means the separation in time and space is deliberately increased for effect. It's isn't really M/S at all, it's two interfering channels that 'coagulate' into a new 'whole' where one channel brings shifts in the stereo field that because of the separation and timing are producing peaks and troughs in the waveform as they modulate and interfere.

Would two cardioids do something similar? Perhaps. I'm not certain it works as well in headphones - I started on them to listen to this and it sounds like there is an effect in the chain, when it's just, I presume, the comb filtering taking place?

An interesting experiment, but it's no longer M/S, it's kind of channel 1/channel 2

DonnyThompson Tue, 11/10/2015 - 04:05

Honestly, Mon Ami, I've never encountered what you have, and you're right - I can certainly hear the signal moving. LOL... Like, You're not walking around while you're playing or anything, right? LOL

Seriously, I've never experienced this anytime I've used an M-S array.

My favorite go-to's for MS are 414's, as their inherent physical shape kind of makes it easier to visually line them up correctly - that's just what works for me.
You don't have to use two exact same models, as long as each mic has the same type of transducer - meaning you probably wouldn't want to use a dynamic mic for your M and a Fig 8 condenser for your Sides...

And, there's nothing that says that you have to have a 'natural" type of M-S, either. You can spread them very wide if you want for special effect.

Here are a few hints I feel work well for me:

• Place the Fig 8 mic directly on top of the cardioid.. and I mean, right on top, just so they're barely not touching.

• If you are going to use an HPF on one mic, ( or input channel) make sure you set the other mic/channel the same way ( you can always add further HPF afterwards in the mix).

• When you clone the track of the Fig 8 mic in the DAW, after inverting the phase on one of those channels ( I always use the Right channel, just out of habit, I guess)... make sure to LINK those two tracks/channels before you pan hard Left/Right, or adjust the Side channel in any way, so that when you are panning or bringing them up in the mix, you are doing so in exact equal increments. The slightest variation - volume or tone - between those two Fig 8 tracks will alter the image.

One of the main things I like most about using an M-S array, besides the beautiful depth and space that it achieves, is because of its mono compatibility.
When you switch the mix to mono, the two Side channels cancel each other out, ( or should if they've been placed and decoded correctly) which then leaves only the direct (Middle) mic, giving you a mono signal.
Because the Sides contain mostly the sound of the space/room , if you switch the mix to mono, it eliminates that sound, resulting in a more direct, primary sound of the instrument/track.

FWIW

pcrecord Tue, 11/10/2015 - 05:22

MrEase, post: 433762, member: 27842 wrote: You seem to be assuming that once the M-S is mixed down it still retains it's M-S characteristic. This is not so, once you mix down the signal becomes conventional L-R stereo. If you find that the "mono" weakens a track that is a characteristic of your mix and is not due to the M-S recording technique.

I don't get you on that. Either you listen in mono or in stereo. If content in the right channel is out of phase of the left channel, they will cancel in mono. This is valid for any stereo recording, right ?

paulears, post: 433763, member: 47782 wrote: An interesting experiment, but it's no longer M/S, it's kind of channel 1/channel 2

I agree, let's call it, Marco's M/S trick ;)

DonnyThompson, post: 433765, member: 46114 wrote: Place the Fig 8 mic directly on top of the cardioid.. and I mean, right on top, just so they're barely not touching.

That's what I'm doing. If the mic would be missplaced, the wave files wouldn't align...

DonnyThompson, post: 433765, member: 46114 wrote: make sure to LINK those two tracks/channels before you pan hard Left/Right, or adjust the Side channel in any way, so that when you are panning or bringing them up in the mix, you are doing so in exact equal increments.

Yes, I do that as well !

DonnyThompson, post: 433765, member: 46114 wrote: When you switch the mix to mono, the two Side channels cancel each other out, ( or should if they've been placed and decoded correctly) which then leaves only the direct (Middle) mic, giving you a mono signal.

That's the thing, I'm not decoding, just a copy of the side and reverse polarity.. I'm gonna look for ways to decode the M/S with a plugin or a DAW fonction..
I could test with my FF800 which has the m/s fonction.. I'll test that with the next musicien that gets to the studio ;)

DonnyThompson, post: 433765, member: 46114 wrote: Because the Sides contain mostly the sound of the space/room , if you switch the mix to mono, it eliminates that sound, resulting in a more direct, primary sound of the instrument/track.

That's another thing that might affect my results, my room is nearly dead, so the signal of the sides maybe more similar than it should be for the M/S to work properly

Boswell Tue, 11/10/2015 - 09:44

DonnyThompson, post: 433765, member: 46114 wrote: When you clone the track of the Fig 8 mic in the DAW, after inverting the phase on one of those channels ( I always use the Right channel, just out of habit, I guess)...FWIW

You have to invert the phase on whichever direction the rear of the S-microphone points. Convention is that the microphone front (where an increase in air pressure or in air velocity results in a positive electrical output) is aimed to the left, and hence the rear of the microphone aims to the right, so the S channel that is panned right has to be inverted. Spinning the S-microphone through 180 degrees and inverting the left-panned S-channel instead would give identical results, asuming the microphone response is symmetrical.

pcrecord, post: 433767, member: 46460 wrote: That's the thing, I'm not decoding, just a copy of the side and reverse polarity.. I'm gonna look for ways to decode the M/S with a plugin or a DAW fonction..I could test with my FF800 which has the m/s fonction.. I'll test that with the next musicien that gets to the studio.

By copying the S-channel, inverting one side and mixing with the M-channel you are decoding. A full mix is always done in L-R stereo - most mixing arrangements (hardware or DAW) can't handle M-S at the point of mix.

pcrecord, post: 433767, member: 46460 wrote: That's another thing that might affect my results, my room is nearly dead, so the signal of the sides maybe more similar than it should be for the M/S to work properly

M/S will always work properly if set up and decoded correctly - it doesn't depend on particular room acoustics to work any more than other microphone configurations would. However, what you hear from an M-S configuration after decoding to L-R may be a result that is not normally achieved using X-Y stereo miking, so may either be unfamiliar or not what is best for a given sound source.

I remember recording a well-known blues guitarist in a studio and he wanted me to try M-S miking as an experiment. I had on hand a pair of Rode NT2-As, which have switchable omni-cardioid-fig8 patterns. I set up the mics and went to the control room. I was amazed by the quality of the sound, but wondered what this really odd width effect was. It took me several minutes to check everything out, and finding nothing wrong with my wiring, decoding and balance settings, I was forced to look at the microphones. I found that the switch on the S microphone was not pushed fully into the fig-8 position, so I was getting a cardioid pattern for the S-channel. Nudging the switch fully over produced a great result, and by varying the S-channel level we were able to set the width of the stereo field to be what the guitarist was really happy with. Interestingly, the chosen final S level was less than that for standard width stereo, a result that would not have been possible to achieve using X-Y miking without going through an XY-MS-XY process.

pcrecord Tue, 11/10/2015 - 10:53

Boswell, post: 433775, member: 29034 wrote: Spinning the S-microphone through 180 degrees and inverting the left-panned S-channel instead would give identical results,

I'm gonna verify that tonight...

Boswell, post: 433775, member: 29034 wrote: I found that the switch on the S microphone was not pushed fully into the fig-8 position,

Not the case here ;)

I know I'm pushing the bounderies of what M/S should be, but thanks to it, we're having this nice discussion.
Thanks guys !

pcrecord Tue, 11/10/2015 - 16:50

Boswell, post: 433775, member: 29034 wrote: You have to invert the phase on whichever direction the rear of the S-microphone points. Convention is that the microphone front (where an increase in air pressure or in air velocity results in a positive electrical output) is aimed to the left, and hence the rear of the microphone aims to the right, so the S channel that is panned right has to be inverted. Spinning the S-microphone through 180 degrees and inverting the left-panned S-channel instead would give identical results, asuming the microphone response is symmetrical.

Ok Bos, I tryed both channel once with the Side left inverted and once with the side right inverted.. Same results.. Unless my KSM44 is defective...?
I use both condensers, both without HPF, both in identical pre (ISA)


I even tested with a tone generator pointing at the M/S at 6 inches, the result is the same a side is always louder.. but if move the generator of center the sound goes to on site and when I go the other way the sound goes to the other side but at equal as when on the other side.. My G.. there is too much use of the word side in this sentence.. ;)
So since my signal is equal in volume when I put it to each side of the figure of eight, I'm guessing it's not defective..
In the end my whole story might hold to the fact that I'm too near of the M/S mics when recording.. !!!
Next time I get a drummer here.. I'll try that again.

As a reminder, I'm able to equilize the volume on both sides by sliding the Mid channel in time (about 1.5ms)
And it is still compatible in mono ! ;)

Boswell Wed, 11/11/2015 - 02:59

It's this sort of difficulty that seems to crop up when using the fig-8 pattern on certain types of double-capsule condenser, and it makes me suspicious of the phasing from these models. As I mentioned previously, I had no problem with the fig-8 mode on a Rode NT2-A (once I had it selected correctly), but I have to say that for the majority of M-S recordings I use ribbons - not only is good phasing automatic with them, but their sonics do not degrade off-axis in the way that happens with some condenser microphones.

Unmentioned so far in this thread is a configuration I use from time to time: M-S Blumlein. This employs two fig-8 pattern microphones, one lined up with the centre of the acoustic source and the other at right angles to it. It's only useful where you have a good acoustic behind the microphones, as it is just as sensitive to rear sounds as to those from the front. I tend to use it for sources such as large choirs and orchestras, and position the microphones at least 3m off the ground on a tall stand behind the conductor and angled downwards so the rear lobe captures the natural reflections from the roof space. In the right type of building, this configuration gives a great sense of air and life to a choir's sounds, as well as controllable width through the level of the S-channel. However, visually, it's not great for a paying audience to have to see past a large sandbagged microphone stand, so I tend to restrict usage to advertised recording sessions with or without an invited audience.

pcrecord Wed, 11/11/2015 - 04:00

Quite frankly, I don't think it's a phase issue of my setup.. I think the signal from both mics are producing signals that are too similar and since we reverse the polarity on one side, it's no surprise that phase cancellation happens... To emphase this, if we take one mono track pan to center and make 2 copy pan left and right with one that has the polarity reverse, I'm sure I'll get similar to worst results..
A part of the center and the reversed side is cancelling each other so it leaves the other side with more power..

But I don't rule out anything, tonight I'm gonna test M/S with 2 KSM44 and I also can test with a ribon and the sm81. So I'm guessing that the signals will be so different that they won't null.
(just a guess ;))

OR all that time, I'm testing with a defective mic !! (hope not) ...

MrEase Wed, 11/11/2015 - 04:50

pcrecord, post: 433807, member: 46460 wrote: A part of the center and the reversed side is cancelling each other so it leaves the other side with more power..

Yes, exactly, that's surely what we're trying to achieve but you only seem to be considering one half of the equation.

For example, let's take an orchestral layout with a violin on your left and a cello on your right with the Side mic pointing to the left. When recorded the mid mic will be in phase with the side mic violin but not with the cello. Mix them and the violin signals will be in phase and the cello's will be out of phase so all you should hear will be violin. That's your left channel. Now invert the side signal and the opposite happens, the violin cancels and the cello adds to give the right channel. QED. If there were no cello yes there would be more power in the left channel but that's what you want, the violin is after all to the left. You also have to consider the other scenario though for the cello. This will give you more power in the right channel if there's no violin.

You are undoubtedly getting some odd results but, like Boswell, I'm convinced that there is something amiss with your setup and it's not a fundamental problem with M-S recording!

pcrecord, post: 433767, member: 46460 wrote: I don't get you on that. Either you listen in mono or in stereo. If content in the right channel is out of phase of the left channel, they will cancel in mono. This is valid for any stereo recording, right ?

I agree, let's call it, Marco's M/S trick ;)

I don't understand this either. The quotes from Paulears and myself seem to me to be paraphrasing the same thing yet you only agree with one?

What I was really trying to say (in response to one of your early comments) is that when you add any stereo spatial effects (either with a stereo enhancer plug in or changing relative M-S levels) then there is always a risk that a mono mix will not translate well. This was in response to your comments about poor mono reproduction when you had boosted the side signals.

pcrecord Wed, 11/11/2015 - 06:30

MrEase, post: 433808, member: 27842 wrote: What I was really trying to say (in response to one of your early comments) is that when you add any stereo spatial effects (either with a stereo enhancer plug in or changing relative M-S levels) then there is always a risk that a mono mix will not translate well.

That, I'm with 100% :)

I think that recording an ensemble is another games.
I'm recording an acoustic guitar in a small dead room and the distance from it is too short. In that lies the defects of my M/S setup, I think.

I'm gonna get back to you with the next tests, which I hope will be the last and will conclude this challenge..
One thing I'm proud of is that I found a way to make it sound right even if something isn't..

pcrecord Wed, 11/11/2015 - 16:22

Ok I did it..
tested with the SM81 and the KSM44
tester with 2 ksm44
tested with the ksm44 and a cascade ribbon

Every single time I get the same result.. on a single instrument at about 6inches, there is always a side stronger...

BUT if I sing about 6 feet away.. the energy is nearly identical to each side..

Conclusion : standard M/S is accurate only for ensembles or distant instruments... :)

JayTerrance Wed, 11/11/2015 - 18:18

Boswell, post: 433805, member: 29034 wrote:
Unmentioned so far in this thread is a configuration I use from time to time: M-S Blumlein. This employs two fig-8 pattern microphones, one lined up with the centre of the acoustic source and the other at right angles to it. It's only useful where you have a good acoustic behind the microphones, as it is just as sensitive to rear sounds as to those from the front. I tend to use it for sources such as large choirs and orchestras, and position the microphones at least 3m off the ground on a tall stand behind the conductor and angled downwards so the rear lobe captures the natural reflections from the roof space. In the right type of building, this configuration gives a great sense of air and life to a choir's sounds, as well as controllable width through the level of the S-channel.

Also great for picking up the overtones that a choral group can generate as these overtones leave the sound source at a rather steep angle...hence the tall stand for the Blumlein.

Boswell Thu, 11/12/2015 - 02:49

pcrecord, post: 433828, member: 46460 wrote: Conclusion : standard M/S is accurate only for ensembles or distant instruments... :)

I suppose it depends what you mean by "accurate". If you want to make a CD that when played gives a sound field that is as near as possible to what you would receive when listening to the same performers in a concert hall, then M-S will do that as well as any other configuration, and sometimes better. However, M-S is a great techique for experimenting with to produce sound fields that are not easily achievable in other ways, and, for me at any rate, there is no concept of "accuracy" in this case. In particular, I find M-S can work really well for close recording of acoustic guitar:

Boswell, post: 433775, member: 29034 wrote: I remember recording a well-known blues guitarist in a studio and he wanted me to try M-S miking as an experiment. I had on hand a pair of Rode NT2-As, which have switchable omni-cardioid-fig8 patterns. I set up the mics and went to the control room. I was amazed by the quality of the sound, ...... and by varying the S-channel level we were able to set the width of the stereo field to be what the guitarist was really happy with. Interestingly, the chosen final S level was less than that for standard width stereo, a result that would not have been possible to achieve using X-Y miking without going through an XY-MS-XY process.

I didn't mention in that previous post that separate EQing of the M and S channels was a great way to control sound-hole boom, and allowed mic positioning that would have been a no-no for most other placements.