Three Coincident Overheads

[UXB]

Thanks, Jeremy, for such a concise overview.
I had investigated using a decca tree (and later abandoned the pursuit)for a drummer with a particularly large kit, and in my research, I thought I had found that the Neumann M50 was not a "true" omni. It's been a while, but if my failing memory serves, I thought it had some mild frequency specific directionality which made it even more suitable to this technique? I never used one hands on, so I can't speak from direct experience, but was wondering what your thoughts were.

Re. various micing techniques, I find it depends greatly on the drummer's approach to drumming. Some drummers, whom I think of as "one mic" drummers (I use this terms because their approach and dynamics gives you a pretty perfect mix of the whole kit from one mic, placed about anywhere! - roughly speaking of course), or those who play with such intensity that any strike feels like a gunshot, but requiring more work at mixdown. For the one mic type, almost any good technique will yield great results, but if you have someone crushing the snare (albeit with great technique), the positioning and angles of the overheads becomes more critical if you are using them to capture the "picture" of the kit. If you are planning to "reassemble" the kit in mixdown, then perhaps close micing the cymbals becomes more relevant. I have battled with mixes on sessions where an unfamiliar drummer will kill the snare, dominating the overheads.

Best,
H

[ckeene]

thanks, Jeremy, that was a really great post. The thrust of my initial post was mostly centered on blending mics to get a (personally) likable tone that was phase-coherent. The question that remains is in regard to the quote below, where you discuss stereo image: regardless of whether that middle LDC was cardiod or omni, wouldn't your stereo image simply be increased or reduced depending on how the middle mic was mixed with the stereo pair? I can absolutely understand that the L-R image would necessarily contract in this scenario, but to the point where volume cues would effectively render it mono?

Now the catch with all of this would be if you tried to do something like adding in a third microphone. You might well get a great tone from the three mics all stacked up with a 90? XY pair and a cardioid LDC in the middle, but you're not really going to get much of any kind of stereo action because there will be no timing cues, and most of the volume cues would be minimized because the cardioid mics just don't drop off fast enough at only 45? off-axis. So you would probably get just about the same result by using a nice LDC omni as your mono mid.

Happy Listening.

-Jeremy

Another thing that was really interesting, though a little O-T, was your implication that adding capsules in a soundfield reduces phase cancellation. I never really thought about that before.

[Professor]

Regarding the M-50 mic, it is a spherical omni capsule, meaning that the diaphragm is kind of like a flat spot on the side of an otherwise smooth sphere. BLUE is now making a capsule like that, Neumann still makes that capsule design in the current M-150 and you can get add-on spheres for SDC omnis from Schoeps, Neumann, MBHO, etc. The idea is that the sphere creates a little more of an acoustic shadow of sounds arriving from the rear of the microphone and reduces signals above about 5kHz by maybe 2-3dB when they are arriving from the rear. It means the mic is more directional at high frequencies and more omni at low frequencies. But then again, that is true of just about every omni and cardioid microphone out there - just to varying degrees. I have some interesting video clips I made of various mics being rotated about the capsule axis (approximately) in front of a speaker playing pink noise as a demonstration of how much sound the mics still hear off-axis. I don't have them posted anywhere, but I'll have to see if there is a place I can put them up where I can preserve the audio quality well enough for them to be useful. A simple experiment you can do is to take a side-address LDC in cardioid or omni (or best of all a multi-pattern LDC) and simply talk into the mic while you rotate it in front of you. I do that one in class to demonstrate polar patterns, but it also shows where the real patterns don't live up to the specification (meaning they might be omni at 1k, and at 100Hz, but no at 10kHz).


As for the stereo imaging, as you keep stacking more mics into the XY array, think about what each mic would be hearing for a source at a particular direction. With a 90? XY pair, a sound on axis with the Left mic will be 90? off-axis from the Right and so will be -6db down in that mic but otherwise identical (except as descrived in the last paragraph). A sound on axis to the Right mic will be 90? off from the Left and so will be -6db lower there. A sound at the center, 45? off-axis from each mic will be identically loud in each mic at about -3dB in each.
Now if I add a third cardioid mic up the middle, a sound on axis to the Left is 45? off-axis to the center and will be -3dB lower there, and still 90? off-axis & -6dB lower in the Right. A sound on-axis at the center will be full strength there and -3dB each to the Left & Right mics.
So if a regular 90? XY gives you about 6dB difference from 'hard left' to 'hard right' then adding the extra coincident center will reduce that to a little less than 3dB difference from hard left to hard right. That means less stereo activity, or a more mono sound.
But you could try widening the angle on the XY pair to say 135? which will return some of that stereo spread though not quite all of it. And while going out to a 180? spread seems like an obvious next step, it also happens to get you into an even more interesting possibility. Two back-to-back cardioids would give a very significant stereo spread and would sum to a simple omni mic in mono. But if you flip the phase on one of those mics, the two would sum to form a figure-8. That's how most multi-pattern LDCs work, as two cardioid capsules placed back to back inside the grille and either summed to mono to make omni, or with one flipped and then summed to mono to make a figure-8. Knowing that, you could take those three mics and turn them into a Mid/Side pair. Invert the Left, add it to the Right and it becomes the 'Side' mic, then you just decode the Mid & Side like normal.
So yeah, while adding an extra 'middle' mic to your 90? XY pair will make your overall sound more mono than the XY pair alone, you could swing that 90? XY out to a 180? XY and either mix it as you were thinking initially, L-C-R, or create a Mid/Side pair out of them. And you don't even have to choose which way to go during tracking, you can track the 3 mics individually and go either direction during mix down.


Oh, and a very quick bit on when I mentioned the 3 mic reducing the phase problems that might exist, that's mostly because we were talking about them being a coincident array. If they are spaced that will get a little more tricky, especially since being close to a drumset would mean a moving target of sound sources. But close together they would be a lot like the difference between two players in unison and three players in unison. Two players on the same note will sound a little phasey because the signals will almost never line up quite right, but adding a third singer means that at least two will always be somewhat close to being in phase. Likewise, when we have the mics that are really, really close to being at the same point in space but just a little off, sure, two of them might be slightly out of phase with certain frequencies arriving from certain directions, but adding the third mic will mean that at least two of them will always be in phase.


-Jeremy

[chris harris]

Great posts, Jeremy!
This may be a little off topic... but, it was mentioned. Could you clear up some of the confusion about the "3 to 1 rule"? I've seen it mentioned often, but never really explained, as a way to minimize phase issues. But, it's my understanding that the technique is actually intended to facilitate the recording of multiple sources in the same space, with minimal bleed. It would be awesome if you could clear this up.

[UXB]

Jeremy,
You are a fount of knowledge with a gift for articulating and communicating it.
If you wanted to host anything like that, just pm me, I may have something.

Just wanted to say thanks for sharing so much info.

Best,
H

[japmn]

I do the XY above and a LDC in front (exactly the same distance from the midle of snare as is the XY) just above kick bass rim with the results sounding the best. Is the good sounding drums this way.

[the finger genius]

You're right that the 3:1 rule is meant to minimize bleed. In practice, if you minimize bleed, you also to some degree are controlling phase issues, because you end up with (mostly) one different sound source in each mic, and therefore less phase cancellation.

Keep in mind that the 3:1 rule is also dependent on having sources with equal volume. For example, if one source is acoustic guitar and the other is a cranked Hiwatt stack, you will not be able to minimize bleed with the 3:1 rule. You're basically going to have 2 electric guitar mics, one which might hear some acoustic guitar, and which also may have phase issues with the other.

Edit: Let me add that 3:1 isn't a magic number for reducing your phase issues. It is about bleed.

Hopefully I didn't just make things more confusing.

[weatherbox]

When I track drums in my home space I always use an xy SDC pair and an omni mono "shoulder mic" to good effect.

Sessions I'm working on now I'm using a pair of KM84s as spaced OHs at a low height - sort of between OH and spot mic placement, and an SM69 stereo mic centered over the kit up higher plus an LDC out front in cardioid. The SM69 capsules are currently in omni (I know, but it sounds good...) and planning to mebbe try m/s over the kit today.

Usually I like to do a mono overhead and stereo room mics but some of these songs called for a tighter and faster sort of feel - less of that room echo - but I feel lost without my room mics; the mono overhead and front of kit mic for me fill in the kit and link the close mics into one instrument in a way I can never seem to make happen with just stereo overheads. This current setup has definitely required a lot of tweaking to find a phase situation that's usable (what lead to going omni on the sm69) but is sounding really excellent. We've got the definition and crispness we need and the kit still sounds "in a room" vs. all close mic'd. It's not quite the method you detail in your post but a variation of it that seems to be working well.

[firesine]

Here's another vote for putting the LDC in front of the kit instead of above, if you haven't already tried. I think two overheads is enough. You can get a much different tone that will add more to your overall sound by lowering the mic. To me, three mics in the same place seems like a waste of a channel.

[Professor]

Yeah, there always seems to be a lot of confusion out there about the 3:1 rule because it's very rarely described well or correctly online, in print, or even in most classrooms. I've typed descriptions here several times already, but I know they tend to fall way back into the depths of old threads, so I don't mind posting another quick one here.

First of all, let's change the name... from the "3 to 1 rule" to the "Greater-than-3 to 1 recommendation".
Here's the idea as we usually see it:
I have one instrument and two microphones that are capturing sound from that instrument. Usually this happens when you have one mic close to your target instrument-A and another that is targeting another instrument-B but still close enough to hear the inst-A. The idea is that the sound from Inst-A reaches the close mic first and the next mic a little while later. That time difference means the signals captured at the two mics will be out of phase at a few specific frequencies throughout the spectrum. If the signals are combined in mix down, those out of phase frequencies will end up canceling each other (at least a little) and messing up your sound.
So the idea of '3:1' is that you would measure the distance from the instrument to mic-A and from the source to mic-B, and that those distances should be different by at least a factor of 3. That is, if you snare mic is 2" off the drum and the hi-tom mic is 2" off that drum then the mics should also be at least 6" from each other. And since we said at least that means it's even better if they are 8", 10" or even further apart. With drums it's actually really easy. It gets a little more tricky if, for example, you have three horn players lined up because you might have the mics 12" in front of each player, and if they are standing shoulder-to-shoulder then the mics might only be 24" apart. There are various solutions for that including getting the mics closer to the horns, spreading the player out more, and using tighter mic patterns while turning the players a bit.
The reason the rule works is not because it puts things back in phase (the only way to do that is to place all the mics at the same distance from the source). It works because of another law, the 'inverse-square law' that says that for every doubling of distance the volume of the original signals drops by half, or -6dB. (Actually it's not really a 'square' for sound, that would be 1/4 the volume, but it is square for light, and we're really just borrowing the name because it rings better than the "inverse proportional law".) Either way, at 3 times the distance, the signal reaching the second mic should be about 9dB down, which means that even if those out-of-phase signals are combined, they won't be the same volume and the closer signal will survive. So it's only a guideline that ensures we have a slight volume advantage to prevent total cancelation. That's why using polar patterns to your advantage will help too - if the second mic is a cardioid pattern and you're catching it 90? off-axis then it's dropping the signal by an extra 6dB right there.

The confusion usually sets in when people forget the "at least" part of the 3:1, which is why I like ">3 : 1 " as a description. If 3 times the distance is good, then 4x is better and 5x is better still.
Another bit of confusion is that distance is the only factor. Like I said, we just want a volume drop, so using polar patterns to your advantage will help too. Of course, it's also true that if the signal is naturally stronger in the second mic because it is more sensitive or the gain is set higher at the preamp, then that will counteract our other efforts.
Another common one is the idea that if you're 10-feet from a source, like you might be at a concert recording, that you would need to have the mics 30-feet apart. In this case, we want the microphones to slightly different volume levels and timings to define the position of the source in the stereo field. We compromise between the intensity of that stereo imaging and mono-compatibility.
And one last common misconception is that it only has to do with one instrument reaching two mics. It can also involve two instruments reaching one mic.


But for now that should cover enough for this thread, and was a lot more than I'd planned on typing.

Happy Listening.

-Jeremy

[Future One]

Well said Professor!

[@?,*???&?]

Ambisonic mic'ing might be something interesting to experiment with. 3 Bi-directional mics placed coincident...