Issues with M/S recording - 90 degrees off axis positioned in center

Question

Hello everybody,

I am having an issue with Mid-Side recording that is really confusing me. I am using a Rode NTG3 as the mid, and a MKH30 as the side mic. I have been conducting tests using my voice and also a sine wave. The issue is that the stereo field is not properly translated after MS decoding. When I speak at 90 degrees off axis, this is translated after decoding as if my voice is more or less in the center. Whenever I speak at about 45 degrees off axis, the stereo positioning after decoding is correct. This issue happens on both left and right, so both 90 degrees L and R are positioned in the center.

I have done various tests in order to find the root of the problem. With regards to decoding, I have tried three different types: manual, using the freeware MSED plugin in my DAW, and using the MS decoder in my field recorder (Tascam DR680). The result is all the time the same.

I have also tried adjusting the width of the M/S decoding. The issue stays pretty much the same regardless of the width I set. The stereo field changes according to the width setting, but 90 degrees off axis is always centered.

I have done tests using my Rode NT1a and NT2a, with the NT2a set to figure of eight and used as the side mic. This configuration behaves pretty much normally. The wider I set the MS width, the more pronounced the 90 degrees sound is placed at either hard L or R.

I have done various adjustments of the recording gain of the NTG3 relative to the MKH30. As the NTG3 is more sensitive (31 mv/Pa) than the MKH30 (25mv/Pa), I'd assume that normally you would turn the gain of the NTG3 down a few dBs in order to get an even stereo field. Whatever adjustment I made to the gain, it never sounded like proper stereo in case of sounds coming at 90 degrees off axis.

I've also had briefly access to a Neumann 191, and in fact had exactly the same issue as with the NTG3/MKH30, so this confuses matters even more. Neither of the mics that I have used are having any issues of themselves, it's just that when doing this test I get these results. What I'm trying to say is that I'm pretty sure that none of the mics I am using is faulty.

I have also tried various cables, this is not the issue either. Neither is the issue with my recorder. I have used both my Tascam and my M-Audio soundcard, both with the same results. The MKH30 is mounted as close to the NTG3 as possible, and moving the position of the MKH30 did not help.

I have uploaded a recording to showcase the issue. For this test, I played back a sine wave (440 hz) at about a meter distance at both 90 and 45 degrees off axis and recorded it simultaneously with the NTG3/MKH40 setup and my Rode NT4 for reference.

I've added comments in the sound file to explain which is which. One can clearly hear that at 90 degrees, the MS setup captures the sine wave as if it's coming from the center, whereas at 45 degrees things are positioned correctly.

MS test by Daan Hendriks

If anybody has any idea what is going on here, I would be very grateful because I am out of ideas how to solve this. I will be renting another MKH30 with MKH50 and MKH60 to conduct more tests. Perhaps the issue has to do with the Rode and Sennheiser not being properly matched, but looking at the specs of both mics, there is not too much of a difference - one would still expect these mics to behave normally in a MS setup, I would think?

Thanks

-- EDIT: alt text http://daanhendriks.co.uk/wp-content/images/ms_setup.jpg This is how I had the mics positioned when I posted my question. After Ryans response, I have moved the MKH30 more backwards but the problem still persists.

Answer 1

Hi,

If you have a sound coming into your mic setup at an angle of 90 degrees off axis, theoretically all of the sound pressure will be on the S microphone, and no sound pressure will be on the M microphone.

So if you look at the waveforms in MS you will probably see the M signal as a flat line, and the S signal as a waveform. You could say that the M signal has a constant value of 0.

If you then decode the MS recording into LR, either by manually summing og using a plugin, the sound in the Left speaker will be M+S, and the sound in the Right speaker will be M-S. If we equal M to 0, the signal in the Left speaker will be S, and the signal in the Right speaker will be -S. Identical, but opposite phase.

Out of phase signals give you no sense of direction, if they are at the same level. So you will have a feeling of the sound coming from somewhere between the speakers, or next to your ears or perhaps even behind you. But you will not be able to pinpoint the direction.

I use MS for recording stereo sounds most of the time. I like the sound of it and the ease of use, but it does have some limitations when it comes to very wide stereo sounds. Anyway most of the action is almost always within +-45 degrees off axis. I wouldn't worry that much about losing the action at +-90 degrees.

If you really wish to capture a very wide stereo field, you should perhaps use an A-B setup or an X-Y setup, perhaps even a binaural setup.

Answer 2

Not entirely sure what's to blame here, Daan, but some thoughts for you just the same, to see if this inspires fresh or oblique thinking about your challenge.

1. Apologies if I missed this, but have you tried another mic with the MKH30? I know you tried MS with another pair, but the first troubleshooting step would be to choose a variable and either eliminate or alter it. The first thing I'd do is pick a mic that's not of an interference-tube (shotgun) design: An omni, cardioid, or even hypercardioid would be fine. I'd recommend renting or borrowing one.

2. Your tests sound valid, but IMO doing tests with just one sound source can be misleading. When I rig up my MS mics, I stand in front with my arms stretched. I announce "right" and snap my right-hand fingers, then "left," and snap my left-hand fingers, repeating a few times. This helps calibrate the stereo spread when I edit the final sound and even helps me identify if I'm mistakenly un-matrixing a MS recording that's been matrixed to LR stereo already!

3. I didn't read much about your signal chain between your mics and the finished recording. Did you mention what recorder you're using? Are you matrixing through a preamp? Are you Is it possible you're using the recorder to matrix the MS recording to LR and you're re-matrixing it again in post?

Answer 3

I would suggest renting a standard sennheiser cardioid so that you can hear the difference between a standard cardioid and a shotgun in an m/s setup. It can make a huge difference.

Answer 4

_{(source: rodemic.com)}

Looking at the NTG-3 polar response, we can see that at 4K the mic null point is around +/- 90 degrees. So depending on frequency, a MS rig with that mic will result in almost no signal in the M channel and max signal in the S channel for a sound source at + or - 90 degrees. Once decoded, this will result in equal amplitude in L/R but with opposite polarity. The safe pickup angle with the NTG-3 in MS would be around +/- 45 degrees as your tests confirm.

If you want to do MS with a large pickup angle, you should use a less directional mic for the M (omni to cardio).

Answer 5

it is possible you have a wiring problem. do your 2 XLR female cables go to 2 XLR males, or is there a 5 pin XLR in between?

If you ever plug the side (figure 8) mic into the mid input, and the cardiod into the side, you might get the problem you are describing.

just a thought,

-greg-

Answer 6

Hello all,

I just wanted to confirm that after having done more tests, including the same test with the same microphone types (NTG3+MKH30) but different mics (a friend of mine owns them both as well), my basic conclusion is:

This combination is not ideal for MS if you want to capture a very wide stereo spread.

Other people have pointed it out already, but yes, I'm convinced of that now also. It will work fine for not-so-wide (i.e. within 45 degrees pickup angle) source, so for spot fx work it's good. But obviously, this is not the setup to use for any sort of ambience recordings.

I had read several times before that a shotgun is not ideal for MS, but now I see how it manifests itself, so it has been quite revealing. Interestingly, my friend did the same test with a MKH60/MKH30 setup, and even though that also revealed pretty much the same issue, it did sound a bit smoother. The jump from 90 degrees to 45 degrees wasn't as pronounced, so to speak. I presume it may have to do with the MKH60/30 combination being better matched in terms of frequency response curve.

What surprises me is that I got similar results with a Neumann 191 as with the NTG3/MKH30. You'd expect the 191 having better matched capsules (frequency wise) and therefore have more of a smoothness to it in the stereo pickup angle, as with the MKH60/30 combination, but in the tests I did at least this was not the case.

Thanks for your help, everybody.

Answer 7

NOTE: I just noticed you have a newer post from about 1 year after this one where you find an MS setup that you liked. However, the below response may be relevant still for those trying their first matrix and frustrated by what they hear and/or don't like.

Listening to your samples, I hear phase problems on the MS setup. I listened to your tests on a set of JBL LSRs and the MS ones clearly stuck out like a sore thumb has not having a clear phase coherence without a doubt.

Example being, the NT4 @ 45 degrees sound like it came from mid way between center and my L speaker (but along the same diameter/distance - as in, it sounded like it came from a point in 3d space exactly along the arc line between LCR, as it should). But from the MS, it sounded like the tone origin was emanating from near by right ear in 3d space, but hearing a perceived reflection near where the NT4 recording felt like it came from in 3d space. But at the same time, it was hard to localize the MS one. So this tells me immediately that you have a phase problem.

The thing with MS is that it's not really as easy or simple as matrixing it and calling it a day. I have an MS setup of an MKH8040 / MKH30 I set up fairly recently and I'm still playing with the matrixing signal flow of a few test recordings (namely, ambiences) to find what I like before heading out to do extensive recording with it.

Because the signals are split when recording (or rather, they SHOULD be - it's highly advisable to never bake the matrixing during recording), these give you many, many options in processing them. Set up your chain correctly, and I you can do pre-matrixing processing such as EQ, gain staging, and time delay on each individual channel before it ever touches a matrixing tool like the Waves S1.

In my own case, I found that after many A/B tests of delaying the side channel (and ONLY the side channel) by 0, 4, or 8 samples - 4 samples produced the most pleasing stereo image to me. At 0 delay (or neutral), the imaging was strange and didn't feel like it sat along the imaginary LCR 'arc' - it popped out three dimensionally weird, almost like how yours sound, but not quite that extreme. The differences at this sample level had a drastic affect upon the result. Even these small differences had a drastic affect upon the noise characteristics of traffic wash in the background. So this is why I feel MS is tricky because it requires a trained and discerning ear as well as a reliable monitoring set up to even trust that what those ears are hearing is accurate.

I mention, rather stress this, because you say that at 45 degress it sounds like the correct stereo position (a la the NT4) - however I can say with 100% confidence, from hearing your sample, that this is not correct at all, it does not sound like a correct 45 degree matrixed 'XY' position within the stereo field. In fact, it sounds like a phase problem. If a sound recording was given/sold to somebody like how that 45 degree MS one sounds in your sample (assuming that what you said about it sounding correct to you means that you would let it pass Quality control), most sound editors would actually kick it back to you as being a non-usable stereo recording because of a phase/imaging problems. That's where it can get dangerous with MS. I'm not saying this as a scolding or reprimand, but I feel it must be stressed because of how critical of importance it is to mastering to be able to trust both your ears and that what is feeding your ears must also be dialed in.

And even before this, I have to attenuate my side mic by about 6.4 dB for correct LCR balance. so the pre-matrix gain staging helped find the correct "balance", and the delay helped find the correct" alignment". I believe that playing with the pre-matrix delay and/or gain staging on per channel basis will give you a better result than what is being had before writing it off as a problem.

Also, the thing to remember about shotguns - the rear node is inverse phase (think of it like a lopsided/squeezed bi-directional), so whatever it captures back there will move right to left (if the front is moving left to the right). If the room you recorded this in has reflections, than this could be also what you're experiencing - the inverse phase early reflection of your tone from the (R) side + M (rear) is not phase coherent with the source sound coming from (L) side + M (front). Using a cardiod like the 8040 or 40 fixes this problem.

Remember that an MS is effectively 3 cardiod polar patters of axis (cardiod + phase-coherent bi-directional), which means it captures as far around as about 270 degrees - 180 degrees of it being the centrally-on axis range. The NT4 is more of a 180 degree range max because of how the capsules are aigned at 45. So with MS you will run the risk of experiencing bleed through in what's captured potentially off-axis. Playing a tone in the room on the left will still be captured by the mid and the right, either as an early reflection or a combination of source + reflection. So you can see how this can easily be a phase problem waiting to happen. of course, the phase-coherent bi-directional assumes that the mic you are using is a true, single-diaphragm bi-directional like the MKH 30. Not all bi-directional mic are made equal, and some cheap ones can use two cardiod diaphragms glued together, which means that since they don't share a single element, both sides are not phase-coherent inverses. For using bi-directional in, MS this can be a problem, because the side requires exact phase coherence. Luckily, you're running an MKH 30 so that's not the problem here. But it does add another thing to keep in mind.

Just a few things to consider. And make sure the diaphragms on your mics are aligned! While I don't know the internal workings of the NTG3, I would take a guess that your MKH 30 is way too far forward from where the NTG3's element likely is - so much of a gap potentially by what I see in the picture, that this could be very well contributing to the phase problems.

Good luck! MS is awesome with so many possibilities, but there's no free lunch with it either!