Four channel recording with two figure 8 mics ???

[666666]

I searched this issue but could not find an answer.

I am aware of how mid-side recording works... to a degree. Typically one cardioid for mid, and one figure 8 for side. Decode the figure 8 signal by doubling it and flipping the phase of one of the copies. Then play them together and you have now effectively separated the left pick-up lobe of the mic from the right pick-up lobe of the mic. Correct? In this case we have taken one single figure 8 signal and split it into TWO separate, unique signals / channels.

So I am wondering if you could take TWO figure 8 mics, position them in a Blumlein type pattern for instance (in an "X"), and then "decode" BOTH of the figure 8 signals later to yield FOUR separate, unique channels.

One possible reason to do this would be if you needed to place microphones in the CENTER of a group of musicians but you only had one stereo recorder handy. If what I described above would actually work, you could now ultimately have control over four physical areas of this performance array by using only two mics / two channels to record.

Would this work? Is this an existing known technique?

I could even see such a technique working for a typical live performance on stage when the mics are in the audience. Set up two figure 8s in Blumlein fashion, then do "side decoding" on both later.... you'd then yield a left and right front, and left and right rear. (Or technically, a left front and right rear on one mic, and right front and left rear on the other mic) Bring all four faders up equal so the decoding works. Then bus each of the 4 individual signals to it's own new separate mixer channel so you could then adjust the level and panning of each individually without messing up the decoding.

I fear I am overlooking or misunderstanding something with respect do "decoding" figure 8 signals... I have yet to experiment with this. If this would work though, it would be awesome. Four tracks for the price of two!

Thanks

[666666]

Side note... I never tried Blumlein, but after reading about it, I don't quite understand how it can be "accurate".

First, I realize there is no decoding involved with classic Blumlein.

If you place two figure 8 mics in an "XY" pattern in front of a source, the lobes of the X mic will be picking up left front and right rear, and the lobes of the Y mic will be picking up right front and left rear. Then, as I understand it, the trick is to simply pan each mic hard, left and right.

It makes sense that the two "front" lobes will represent a "correct" stereo image of what is out in front of the mics.

BUT, what is being picked up by the "rear" lobes will be "reversed" compared to the original stereo field and thus not accurate. Correct?

When the "X" mic is panned hard left, it will place the front lobe (originally pointing left) to the left, great... BUT will place the rear lobe (originally pointing RIGHT) also to the LEFT.

So, am I misunderstanding Blumlein, OR is this "criss-cross" of the actual stereo field done on purpose to achieve a special "euphoric" effect?

I would say that "Blumlein" makes a lot more sense IF you were to introduce "side-decoding" to each figure 8 and then have CONTROL over the stereo placement of each lobe. Then you can put both left lobes to the left, and both right lobes to the right for a more accurate snapshot of reality... plus the beauty of being able to mix and treat each lobe separately to create your own desired custom balance.

[John Willett]

Originally Posted by 666666

I searched this issue but could not find an answer.

I am aware of how mid-side recording works... to a degree. Typically one cardioid for mid, and one figure 8 for side. Decode the figure 8 signal by doubling it and flipping the phase of one of the copies. Then play them together and you have now effectively separated the left pick-up lobe of the mic from the right pick-up lobe of the mic. Correct?

No, not correct.

The fig-8 has the positive lobe to the left and the negative to the right - the polarity invert just puts the positive on the right and the negative on the left.

It's the exact reverse, so no point in recording it at all.

Originally Posted by 666666

In this case we have taken one single figure 8 signal and split it into TWO separate, unique signals / channels.

No - both are identical and if you listened in mono you would get no sound at all.

Originally Posted by 666666

So I am wondering if you could take TWO figure 8 mics, position them in a Blumlein type pattern for instance (in an "X"), and then "decode" BOTH of the figure 8 signals later to yield FOUR separate, unique channels.

No, you would have TWO unique channels and the two channels of those polarity reversed - so why not just record two.

Originally Posted by 666666

I could even see such a technique working for a typical live performance on stage when the mics are in the audience. Set up two figure 8s in Blumlein fashion, then do "side decoding" on both later.... you'd then yield a left and right front, and left and right rear. (Or technically, a left front and right rear on one mic, and right front and left rear on the other mic) Bring all four faders up equal so the decoding works. Then bus each of the 4 individual signals to it's own new separate mixer channel so you could then adjust the level and panning of each individually without messing up the decoding.

You need three channels to do surround - this is MSM or double-MS. Forward facing cardioid, rear facing cardioid + side facing fig-8 - you can matrix 5.1 from this (Schoeps do a plug-in for this).


To understand how MS works, go back to elementary polar patterns ad think of how the cardioid pattern is derived from adding an omni and fig-8 together.

Now imagine the fig-8 sideways and a cardioid instead of an omni - add those two together and the result is a super-cardioid pointing approx. half left. That's the left channel.

Polarity inverting the fig-8 and adding it to the cardioid gives a super-cardioid pointing approx. half right. That's the right channel.

For MSM the above is front left and right, doing the same with the rear facing cardioid gives rear left and right and the forward facing cardioid is the centre channel.

I hope this helps.

[John Willett]

Originally Posted by 666666

Side note... I never tried Blumlein, but after reading about it, I don't quite understand how it can be "accurate".

First, I realize there is no decoding involved with classic Blumlein.

If you place two figure 8 mics in an "XY" pattern in front of a source, the lobes of the X mic will be picking up left front and right rear, and the lobes of the Y mic will be picking up right front and left rear. Then, as I understand it, the trick is to simply pan each mic hard, left and right.

It makes sense that the two "front" lobes will represent a "correct" stereo image of what is out in front of the mics.

BUT, what is being picked up by the "rear" lobes will be "reversed" compared to the original stereo field and thus not accurate. Correct?

When the "X" mic is panned hard left, it will place the front lobe (originally pointing left) to the left, great... BUT will place the rear lobe (originally pointing RIGHT) also to the LEFT.

So, am I misunderstanding Blumlein, OR is this "criss-cross" of the actual stereo field done on purpose to achieve a special "euphoric" effect?

I would say that "Blumlein" makes a lot more sense IF you were to introduce "side-decoding" to each figure 8 and then have CONTROL over the stereo placement of each lobe. Then you can put both left lobes to the left, and both right lobes to the right for a more accurate snapshot of reality... plus the beauty of being able to mix and treat each lobe separately to create your own desired custom balance.

I don't think you understand how Blumlein and MS works - but I hope my post above helps.

[666666]

Thanks John.... clearly I was way off track in my thinking. I was completely neglecting that with MS, the "mid" channel plays a critical part in the decoding... you can't just take the "side" signal only and make it into two separate channels by flipping the phase... of course it would just cancel itself out and disappear... in all seriousness, I did NOT yet have my morning coffee, this is my excuse.... but it's already noon... I will go have a cup of coffee now anyway... I better "wake up" before the day is over.

Actually.... hmmm (still haven't had the coffee)... so I wonder then, in MS recording, let's say you pull the mid channel way down to "widen" your field as much as possible... the bulk of the remaining signal would then be just the two side "splits" (one being phase flipped)... wouldn't this then not work well (with very little "mid" in the equation)?

I better just go pick up some mics and start experimenting first-hand, that will probably help me to really soak in what's actually going on here.

[John Willett]

Originally Posted by 666666

.... but it's already noon... I will go have a cup of coffee now anyway... I better "wake up" before the day is over.

Early afternoon where you are - almost 6PM here.

Originally Posted by 666666

Actually.... hmmm (still haven't had the coffee)... so I wonder then, in MS recording, let's say you pull the mid channel way down to "widen" your field as much as possible... the bulk of the remaining signal would then be just the two side "splits" (one being phase flipped)... wouldn't this then not work well (with very little "mid" in the equation)?

The normal way is to not touch the mid and do the "steering" by changing the level of the fig-8.

You may find it useful to get hold of the AES publication "Stereo Recording Techniques" - there is a great paper in there by Ron Streicher and (sorry I forget) on MS which explains it all very well.

[Gutbucket]

A Mid/Side recording made using a figure-8 as the mid mic and a figure-8 as side mic decodes to Blumlein.

Inversely, you can use M/S encoding on a recording made with a Blumlein array to derive sum and difference signals. In that particular case, the sum signal is equivalent to a forward facing figure-8, the difference signal is equivalent to a side facing figure-8.

[John Willett]

Originally Posted by Gutbucket

A Mid/Side recording made using a figure-8 as the mid mic and a figure-8 as side mic decodes to Blumlein.

In fact, this was Blumlein's preferred method.

[SoundWeavers]

the answer is NO. 4-channel recording with two figure-8 mics is not possible.

In the meantime, I enjoyed the additional knowledge, information and clarification from Mr. John Willett, KUDOS!

[Petrus]

With double MS you can get 4.0, but not real 5.0 (forget the 0.1 in this context), as L, C and R are all synthesized from just 2 mics. It still is a neat and compact way of doing surround in field conditions with minimum possible number of mics. Soundfield and similar arrays use 4 mics, even though they might be even smaller in size, and need more complicated matrix to code the signal to something usable. Double MS can be coded and adjusted with any DAW.

[ghellquist]

If you record three channels you can do all kinds of interesting things.

Say, as an example, that you run 2 channels with crossed figure 8 mics. And then one channel with an omni mic. If the mics are really close you may do "mixing" of these signals. Theoretically you could then add one of the figure 8 mics to the omni mic and get a cardioid pattern. In this way you can simulate 4 different cardioids going in four different directions. Now, theory and practice is close but not the same thing.

Side note, this is actually a very nice experiment to do. I´ve tried it and it works surprisingly well. When I tried it I got some artifacts from the figure8 and the omni not beeing quite close enough. I used a Royer SF24 at the time with a Sennheiser MKH20, and as those two mics sound a bit different from each others this also influenced the end result. It worked perfectly for the top figure 8 of the SF24, here the omni mic ended up real close. (I had the omni business end pointing down on top of the SF 24 ) . For the bottom figure 8 the distance was large enough to create a little bit of interference, not bad but not necessarily perfect.

In a similar manner, one cardioid pointing forward and one backward combined with a single figure 8 ( the Schoeps double MS method ) records three channels and gives you a lot of options i post processing.

// Gunnar

// Gunnar

[John Willett]

Originally Posted by Petrus

With double MS you can get 4.0, but not real 5.0 (forget the 0.1 in this context), as L, C and R are all synthesized from just 2 mics.

Not true really as you can use all 3 mics to synthesise L, C and R.

My ideal MSM set-up is an MKH 800 TWIN and an MKH 30.

The centre is the front and rear cardioids added or subtracted to give you whatever polar-pattern you want for the centre channel.

Front left and right can be a simple MS of the front cardioid and fig-8 - but it can also be MS using the rear cardioid to make the front mic any pattern you like between omni and fig-8.

The rear left and right are done in the same way as front left and right - but with cardioids reversed.

So you *do* get a full 5.0 and *not* 4.0 (and I agree you ignore the .1 as this is low frequency effects and not bass as most people think).

[audibell]

Hi, it's the algebra that's missing - one of the few cases where the math actually makes it clearer for the non-numerically inclined. (L+R)+(L-R)=2L & (L+R)-(L-R)=2R The second part is where you flip the phase on the side channel. I heartily recommend the schoeps DMS plug-in as a way to learn about how patterns affect each other. It's available here: Order PlugIn - SCHOEPS.de
You have to answer a few questions in exchange for receiving their excellent tool for free.

Use any fig-8 and 2 lipstick cardioids to make it work, experiment with it in any VST-capable daw.

The Streicher-Everest book is a useful tool on any location recordist's bookshelf, as are the very informative .pdfs on Wes Dooley's AEA site.

rgds
WalterT

[Petrus]

Originally Posted by John Willett

Not true really as you can use all 3 mics to synthesise L, C and R.
---snip---
So you *do* get a full 5.0 and *not* 4.0 (and I agree you ignore the .1 as this is low frequency effects and not bass as most people think).

I think it is 4.0 with L+R summed to C. There is no real independent C channel even if you mix a dash of rear cardioid in it. If this was pure, real 5.0, why everyody is using 5 mics for proper 5.0 if you could just get it by jugling 3 mics? For example there is no time domain differences beween L, C and R. This does not diminsh the value of the DMS system for cases where compactnes etc. is needed.

It is true that the Schoeps matrix box puts out 5.0, but also 4.0 for honesty's sake...

[John Willett]

Originally Posted by Petrus

I think it is 4.0 with L+R summed to C (or actually just the front cardioid alone as C, same difference). There is no real independent C channel. I this was pure, real 5.0, why everyody is using 5 mics for proper 5.0 if you could just get the C by summing L+R? Actually we could just have 4.0 and people would sum as many intermediate channels as they wish? 5.0, 7.0 and so one.

It is true that the Schoeps matrix box puts out 5.0, but also 4.0 for honesty's sake...

No, no, no......

The centre channel need *not* be left right combined - nor the cardioid on it's own.

You have back-to-back cardioids - just like any switchable-pattern mic.. So the centre channel is whatever pattern you want it to be - which can be a different pattern from what you use in MS to get left and right.

The centre mic. is a totally separate mic.

It is definitely a 5.0 and *not* a 4.0.

It's a *lot* more flexible that you seem to be envisaging.

[apple-q]

More info here: http://www.schoeps.de/documents/SCHO...d-brochure.pdf

Page 9 onwards.

[aracu]

"So I am wondering if you could take TWO figure 8 mics, position them in a Blumlein type pattern for instance (in an "X"), and then "decode" BOTH of the figure 8 signals later to yield FOUR separate, unique channels."

Check out the available info on the Josephson C700S stereo mic, you might find it interesting.

[David Spearritt]

Just to correct the algebra. Because the matrixing is reversible, the correct maths is:

L = (M+S)/sqrt(2)
R = (M-S)/sqrt(2)

Reversability occurs because you matrix back the same way.

M = (L+R)/sqrt(2)
S = (L-R)/sqrt(2)

[Tommy-boy]

Originally Posted by John Willett

In fact, this was Blumlein's preferred method.

I did not realize that. Interestingly, I did some recordings with 2 fig-8 pattern mics with one mic point directly at the center of the sound stage and the other at a 90 degree angle. I then decoded like normal M-S decoding. I thought I was doing Blumlein, but then thought I must have been doing something wrong after recently reading articles on Blumlein recording that all showed the mics being positioned so that the "V" from the two mics pointed at the sound stage. So either way it's the same. Hmmmm...

[Celloman]

When done properly, Blumlein does not need decoding. The method you used is perhaps a variation on m/s?

Mike

[John Willett]

Originally Posted by Tommy-boy

I did not realize that. Interestingly, I did some recordings with 2 fig-8 pattern mics with one mic point directly at the center of the sound stage and the other at a 90 degree angle. I then decoded like normal M-S decoding. I thought I was doing Blumlein, but then thought I must have been doing something wrong after recently reading articles on Blumlein recording that all showed the mics being positioned so that the "V" from the two mics pointed at the sound stage. So either way it's the same. Hmmmm...

Originally Posted by Celloman

When done properly, Blumlein does not need decoding. The method you used is perhaps a variation on m/s?

Mike

What he was doing *was* MS!

This was Blumlein's preferred method, I have been told.

MS fig.8s matrix into Blumlein - but - using the MS method you can "steer" the patterns to narrow or widen the image and, uniquely, the shape of the matrixed fig.8s do not change as you steer them.

[Celloman]

Yes, I think I read about that. but have never tried it. I always equated Blumlein with the 45 degree, non MS setup.

Thanks,

Mike

[Adebar]

Even more powerful than 2 figure 8 are two crossed figure eight plus an omni.

If you take a look at Josephson Engineering Microphones -> C700 -> Users Guide there you will find a very good description for matrix and levels.

Instead of only 2 figure eight you are able to create every pattern in every direction in as many channels you like by just matrixing the 3 signal of the 3 capsules. So you could generate a 5 channels for typical surround.

[John Willett]

Originally Posted by Adebar

Even more powerful than 2 figure 8 are two crossed figure eight plus an omni.

If you take a look at Josephson Engineering Microphones -> C700 -> Users Guide there you will find a very good description for matrix and levels.

Instead of only 2 figure eight you are able to create every pattern in every direction in as many channels you like by just matrixing the 3 signal of the 3 capsules. So you could generate a 5 channels for typical surround.

Not quite - you need a 4th mic. - a fig.8 for height - and those equal the Soundfield B-format.

[David Spearritt]

Originally Posted by John Willett

What he was doing *was* MS!

This was Blumlein's preferred method, I have been told.

MS fig.8s matrix into Blumlein - but - using the MS method you can "steer" the patterns to narrow or widen the image and, uniquely, the shape of the matrixed fig.8s do not change as you steer them.

If you change the relative gain or patterns of either mic to anything other than pure fig 8 then its NOT Blumlein.

John, do you have a reference to his preference for MS? I find this hard to believe.

[John Willett]

Originally Posted by David Spearritt

If you change the relative gain or patterns of either mic to anything other than pure fig 8 then its NOT Blumlein.

Agreed - the Blumlein set-up is XY figure 8s.

Originally Posted by David Spearritt

John, do you have a reference to his preference for MS? I find this hard to believe.

This was told to me in a lecture a whie ago and we were viewing the original films of Alan Dower Blumlein doing the stereo "walking and talking" tests.

But I said that he was using two fig.8s in MS which matrix to XY fig.8s.

In fact I would be surprised if he didn't prefer this method, as you have a mic. pointing directly at the centre of the action and this matrixes to perfect "Blumlein" - you also have the added advantage that if you alter the side level the pattern widens or narrows, but without any change in the shape of the figure 8 pattern.

[Adebar]

Originally Posted by John Willett

Not quite - you need a 4th mic. - a fig.8 for height - and those equal the Soundfield B-format.

As long as there is no real distribution format I don´t care about height and I´m glad with the 3 capsules of the C700S.

[John Willett]

Originally Posted by Adebar

As long as there is no real distribution format I don´t care about height and I´m glad with the 3 capsules of the C700S.

That's fine - but originally you said "every pattern in every direction" and you need the height fig.8 for that.

And - with the height information you can tilt the mic. if it's badly angled.

[Adebar]

That´s fine.

Beside of that the figure eight and onmi capsule of the the Josephson C700 do sound really great. It´s not only the concept of what makes it unique - it´s also the built quality and detailed reproduction of sound.

[aracu]

Originally Posted by John Willett

MS fig.8s matrix into Blumlein - but - using the MS method you can "steer" the patterns to narrow or widen the image and, uniquely, the shape of the matrixed fig.8s do not change as you steer them.

One of the most natural sounding, problem solving recording techniques.