Okay, I understand how to record M-S stereo, the steps you go though are clear to me but...
I can't see how it works, and so far can't find anyone who can explain it.
One feature of stereo is that (sometimes) you'll hear stuff in the left ear, or from the left speaker, but not from the right, and vice-versa. In other words if, during recording, a room door to the left closes then when you play back the recorded sound the clunk comes from the left speaker. Recording in A-B (or X-Y) then this sound gets picked up by the "left" mic, and placed on one channel. The Right mic hears nothing and there's no sound recorded on the right channel. In playback the clunk comes out of the left speaker and the stereo image is made.
With a figure8 microphone it records left and right on one track (assuming that you record the mid microphone on one track and the side microphone on another) and then in post you use the recorded side track as the "left" side and then "flip" the track to get the "right" side. But... If you take this track and flip it, it sounds just the same.
My question is... Using the closing door example... How could the "side" track reproduce the clunk of the closing door just on the left speaker and not on the right?
I asked this question at various stands at IBC and no-one could explain it, the usual answer was just to try and sell me a microphone. (John Willett wasn't there unfortunately as it was him I wanted to ask).
So, can anyone explain it? this is a serious question, I genuinely don't know the answer and would like to be enlightened.
Actually, when you record using the A-B method, both mics will pick up the sound of the door closing, but the right hand mic will be quiter and will take a fraction of a second longer to receive the sound. If you play back the recording it sounds as if only the left mic has picked up the sound but if you listen closely using a pair of headphones you should hear a slight sound from the right too. This diagram (with the sound-source coming from the right) is quite useful.
Re. M-S Stereo. It's not my field but I found this article on Google which might help? How MS Stereo Works.
Okay, I understand how to record M-S stereo, the steps you go though are clear to me but...I can't see how it works, and so far can't find anyone who can explain it.
With a figure8 microphone it records left and right on one track (assuming that you record the mid microphone on one track and the side microphone on another) and then in post you use the recorded side track as the "left" side and then "flip" the track to get the "right" side. But... If you take this track and flip it, it sounds just the same.
No, you have it wrong.
The fig-8 mic is NOT recording left and right, it is recording "side".
You record mid and sides.
The mid mic. records the front, central sound and the side, fig-8, mic. records the side information (+ve lobe pointing left).
Your left stereo channel is an addition of the mid and the side mic. and is the equivalent of a super-cardioid mic. pointing about 45° to the left.
If you polarity reverse the fig-8 mic and add that to the mid you get another super-cardioid pointing about 45° to the right.
Does this make it clearer?
Look at the maths for a few seconds:
Suppose you have two mic channels labelled L and R. You can convert these signals to M-S easily:
M=L+R, S=L-R
And, you can convert back to L and R just as easily:
L=(M+S)/2, R=(M-S)/2
The merits of recording and manipulating sound in MS form is that you can change the gain of M or S and not change the centring of the sound image. Make S smaller and the image width shrinks, or bigger and it grows, but it stays centred. If you manipulate the gains of LR, you change the image centring unless you do exactly the same thing to both.
Typically, a microphone set that delivers MS has an omnidirectional mic for the M signal (so it picks up everything in all directions), and a cardiod (figure or 8) for the S signal, set at right angles to the sound field you want to record, so it has peak responses at extreme left and right, and has a null response (i.e. no output) fore and aft.
That's exactly what John described, but in different words.
In reasonable laymans terms, the key to understand is that the "plus" and "minus" referred to represent phase in real terms.
Hence, to take the example of a door slam, it will be present in the M channel in all cases. If there is nothing on the S channel, the sound will be in the middle. Have the same signal on both M and S channels and it will be panned left. If you then INVERT THE PHASE of the S channel, the result will be panned right.
M & S was especially useful in systems which had two asymmetrical tracks, such as 16mm mag film with a centre and edge track. The S signal would be recorded onto the least reliable of the two, when any problems with the channel would show up as variations of the width - not as serious as if the problems were on the right channel. A gain control on the S channel then acts as a width control.
I think I expressed myself badly.
When I said that the figure-8 mic is recording left and right, I should have said that the figure-8 mic is just recording what is happening to the left and right sides, leaving the M (usually cardiod rather than omni) to record what's going on in front. I didn't mean that it was working as a L-R pair... That's the problem.
Now (ignoring the M mic completely for now) you have a door slamming to the left of the figure 8 mic. It will be "heard" by the figure-8 mic and recorded as a mono track onto tape.
In the edit suite play back the track, which is mono, and you'll hear the door slam. Invert it and play it back and you'll still hear the door slam.
So how can a mono track suddenly be converted into a true left/right reproduction? Surely you'll never be able to reproduce the original door slam on left side only? (okay you could pan it completely to the left but then there'd be nothing at all on the right side, which is not "stereo")
What I mean is (and hopefully explain what I'm trying to ask) if you were recording a situation where a bass singer was on the left side of the room and a tenor on the right. (Still ignoring the "mid" signal completely) and you recorded this on a figure-8 mic onto a mono track. How could you later (in post) reproduce this split effect with only the bass out of the left speaker, tenor from the right?
I'm aware this is a technically anal question, you don't need to know how M-S works to know that it does, but this is a question which I'd love to have answered.
All the parts of the explanation are already before you. Neither the M nor the S signal alone carries the whole signal; the door slam on the left is recorded in the M signal, and in the same phase in the S signal, so when added together they make the L signal (door slam heard) and when subtracted they make the R signal (door slam not heard, it cancels). You cannot ignore the M signal any more than you'd expect to be able to ignore the L or R signal of a LR pair; in both systems each channel is a mono feed, the spatial content of stereo exists only when you take them as a pair.
Actualy now that I gather you can't ignore the "mid" signal, it becomes clearer.
"M=L+R, S=L-R
And, you can convert back to L and R just as easily:
L=(M+S)/2, R=(M-S)/2"
was a little bit too obscure for my tiny brain.
Also John Willett's answer of "The mid mic. records the front, central sound and the side, fig-8, mic. records the side information " didn't make me think that it was essential for the mid to record everything that the side records to get the stereo effect.
So, does this mean that any sound which falls outside the range of the "mid" microphone can't be reproduced as a purely "left" or "right" sound? In fact the result would be that sounds which occur way, way to one side or other (out of the "mid" footprint) would then be reproduced from both speakers and thus appear to be in the middle?
Did you do any maths at school? This is very basic algebra. We've all explained it in very simple terms, I can't see how we can make it any simpler. Go back to the top of this thread and read all our responses again, you'll find that we've explained it all several times.
Typically, a microphone set that delivers MS has an omnidirectional mic for the M signal (so it picks up everything in all directions), and a cardiod (figure or 8) for the S signal, set at right angles to the sound field you want to record, so it has peak responses at extreme left and right, and has a null response (i.e. no output) fore and aft.That's exactly what John described, but in different words.
Sorry Alan, not quite.
The "S" signal MUST be a fig-8 and NEVER a cardioid.
The "M" signal is normally a cardioid but can also be omni, super-cardioid, or even another fog-8.
A cardioid mid gives you the equivalent of a pair of super-cardioids at 110°
An omni mid gives you the equivalent of a pair of cardioids back-to-back
A fig-8 mid gives you the equivalent of a crossed pair of fig-8s
Thanks John. I was getting a bit unsure.
Alan, I appreciate that you've told me what the M-S system is (I've used it and know how it works) my question was why it works. I was after an "idiots guide" to why M-S works.
(and although I haven't done any serious math since uni, I can follow a simple equation.)
So far no-one has been able to offer a simple (non-technical) explanation. The sort of thing which you could quote to someone who left school at fifteen and is new to the sound field.
but thanks for your time anyway.
Thanks John. I was getting a bit unsure.
Alan, I appreciate that you've told me what the M-S system is (I've used it and know how it works) my question was why it works. I was after an "idiots guide" to why M-S works.
(and although I haven't done any serious math since uni, I can follow a simple equation.)
So far no-one has been able to offer a simple (non-technical) explanation. The sort of thing which you could quote to someone who left school at fifteen and is new to the sound field.
but thanks for your time anyway.
OK John, you're right, I've always been a bit shaky on the names of the various polar diagrams.
But, I still can't see any way to make the explanation any simpler, it's already extremely simple. The maths is of the sort that I was doing at Junior School, aged 8, it's not degree stuff.
So far no-one has been able to offer a simple (non-technical) explanation. The sort of thing which you could quote to someone who left school at fifteen and is new to the sound field.
It is very simple, actually, and I have a presentation explaining microphones (which I am going to give to a college in Taunton on the 29th).
Firstly - polar patterns - there are only two real patterns:
Pressure microphone (reacts to changes in pressure) = Omni = picks up all the way round.
Pressure-gradient (reacts to differences in pressure) = fig.8 = picks up front and back.
Every other pattern is a combination of these.
To get a cardioid pattern you add an omni and fig.8 as shown below.
You can see at the front 1+1=2, at the sides 1+0=1 and at the rear 1+(-1)=0. Do this for all angles and you can see how we get the cardioid pattern.
OK.
For MS you take (for example) a cardiopid microphone for the mid and the fig.8 (turned sideways) for the side (+ve lobe to the left):
If you do the same additions as we did to get the cardioid pattern above, you will see we get a super-cardioid pattern pointing half-left (remembering that a fig.8 has +ve on the front and -ve on the rear).
So, reversing the polaroty of the fog.8 and doing the calculations again results in another super-cardioid pointing half-right.
Is this clear now?
You may find THIS useful.
Thanks John!
That's exactly what I was after. Obviously the diagrams make it easier for me to visualise but by you going through it step by step means that it's now logical. I couldn't see how a figure-8 on its own could produce two different sides but now seeing that it's the interaction with the mid mic which creates the effect, makes it much more understandable.
Cheers mate!
But, I say again, the figure-8 on it's own doesn't produce anything other than mono, it's only when used in combination that you get stereo. It takes two mikes to produce two channels, you can't get two channels out of one mic.
At risk of confusing the issue, M-S relies on signals adding and subtracting depending on their phase. If you have access to a mixing console and a source of say 1kHz tone you can listen to this in action (and watch it on the meters) If you plug the tone into two adjacent channels and fade one up followed by the other to the same level you will get an increase in level on the meters of (6dB if memory serves). If you then fade one channel down and reverse its phase then fade it up again the overall level drops. in a perfect world with 100% phase accuracy the signal would disappear completely.
The principle problem with M-S microphones is that many of them have a tendency to sound "phasey" and in fact recordings made with these mics can wreak havoc with the old Dolby Stereo matrix encoding system. (which also relies on accurate phase) I've always suspected but lack the facilities or maths to prove that these undesiarable characteristics are caused by a) the capsules having different frequency responses and or b) different acceleration of the capsule membranes resulting in an unwanted phase difference. It's probably also worth noting that some mixing consoles (and DAWs) provide a stereo width control by converting A-B stereo to M-S allowing the S to be varied to change the width then converting back into A-B
If the frequency response of the microphone combination is to be constant (“flatâ€) across the audio-frequency band, then there must be ZERO time-delay difference between the two acoustic paths. (the two paths from any arbitrarily placed acoustic source to each microphone).
This zero delay difference could only be achieved if the two microphone capsules were spatially coincident. In practice it would mean that the capsules and their spacing are small compared to the wavelength of the highest audio frequency to be used.
Alan
And there lies another problem. At the very frequencies where phasing anomalies are most noticeable, the wavelength is comparitively short. Therefore, the capsules need to be really close together. There is of course the alternative approach using a tetrahedral array of identical capsules, perhaps best exemplified by the Soundfield mic.
As an aside, certain M-S mic designs exacerbate the problem by using an intereference tube on the front element.
Okay Rob, thanks for that.
My main reason for going towards M-S stereo rather than X-Y is portability. I'm looking at a Sennheiser combination of an MKH30 and an MKH40 together in a Rycote windjammer. The sort of thing you can have at the end of a fishpole and not have to worry about it too much.
I was going to get the Senny MKH stereo mic (MKH 418S) but then John W (my sound guru and the best brains to pick in my not so humble opinion) suggested going for the 30/40 mix. So, hopefully the problems associated with "stereo" microphones won't apply if the 30 and 40 are as close together as possible (ie almost touching).
My expertise with M-S is mostly on the receiving end. I.e. I was a dubbing mixer by profession. As a result I heard an awful lot of frankly awful M-S material. FWIW the best I ever heard was recorded with Schoeps capsules clipped together in a very neat Rycote, next best was the combination you suggest and the worst was from a Neumann, purpose designed mic. (Apart from the clowns that used a cardioid or shotgun for the front and an omni slung over their shoulder for the S) I too would avoid the 418S on exactly the grounds I cited earlier.
If it is of any interest there are a couple of guidance notes we used to issue to freelances recording in M-S. Firstly, that you should never compromise the mono when recording dialogue. I.e. keep the mic in just as close as you would for mono and follow the artist(s) in exactly the same way without worrying about disturbing the stereo image, the mixer can sort it out in post ;-)
Secondly, you may wish to apply 3 or 6dB extra gain on the S signal when recording (obviously drop the monitoring accordingly) I think most location mixers that have M-S decoding will have this option anyway.
Hope this is of some interest at least.
It's of more than interest. thanks for your time .
At the moment, on the few occasions I record stereo, I use 2x MKH40 mics in an X-Y configuration, tips almost touching and, so far I've been lucky. The only thing is that whilst this is okay for plonking on a stand, I don't fancy it waving around at the end of a pole. Also I like the fact that poor M-S can be sorted in post a lot easier than poor X-Y providing that the Mid signal is okay (you just end up with mono, as I understand it).
(I'm not a sound mixer, by the way, it's just something I have an interest in, recording is something I occasionally have to get involved in.)
I was going to record 2 x mono, ie: the fig8 (mkh30) on one channel the mid (mkh40) on another and sort it out in post. The tip about having a bit "extra" on the side channel makes sense, thanks for that.
When I get home I'll try that experiment with the 1kh tone, just for fun.
I'd love to go for the Schoeps capsule but it's a fair bit outside my budget. Also I've become quite a fan of the Sennheiser MKH microphones.
The Sennheiser MKH 30/40 works superbly - I have had my own set for 20 years with never a problem.
For outside use they are clipped together with a special clip inside a Rycote Stereo Windshield kit with ConnBox.
The two microphones are pretty-well identical, except for polar-pattern.
The Schoeps fig.8 is not a true fig.8 as the rear lobe is not identical to the front lobe and will affect the resultant stereo because of this.
MKH 30:-
[IMG]http://www.sennheiser.co.uk/uk/icm.nsf/resources/C1256F140044E469C1256F4E005D79A3/$File/02872_pd.jpg[/IMG]
[IMG]http://www.sennheiser.co.uk/uk/icm.nsf/resources/C1256F140044E469C1256F4E005D79A3/$File/02872_fg.jpg[/IMG]
Schoeps MK8:-
That does surprise me, not that the Sennheisers are a "true" 8 but that the Schoeps aren't.
Anyway, it just makes me the more determined to get a MKH30 now.
There was one on ebay last week which I started bidding on, it eventually went for just over €400 but... in the small print the seller mentioned that there was a screw missing and it had a loose connection somewhere...!!! Luckily I read that early on in the auction, sounds like a bit of DIY repairs gone wrong.
Glad I didn't win that one!
Thanks again for the advice John.
As I said, my observations are based on extensive empirical listening in post rather than more scientific evaluation. Of course I often had no way of judging how good the recordists technique was. The polar plots never tell the whole story although they can be a useful guide.
On balance, My instinct would be to go with the Sennheisers. I never liked the 'Nextel' finish on the Schoeps and their price is even more scandalous. More to the point, the Sennheiser combination will be a better match when circumstances demand that some material is shot on a 416 or even 816. Having said which, the new Schoeps short-shotgun is attracting a lot of attention...
Bottom line is that technique is more important than the kit once you get past a certain threshold. It is this threshold that seperates pro kit from the rest and both Sennheiser and Schoeps qualify. (Apologies to everyone else I'm leaving out) There will always be the occasional anachronism like the Neumann I alluded to before but, basically, once you know what you are listening for, class will out. I speak as one who nowadays compromises with Rode and similar on purely budgetary grounds. But I'm always aware I have to work a lot harder to achieve what I want. It's a bit like the difference between 2/3" cameras and 1/2". In short, if you cannot control the circumstances then you need the kit to give you the widest margin for error.