Why you should never use multi pattern mics

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
While it is true that most LD condensers are optimized for cardioid mode, some assumptions in the first post are not quite correct.
Let's take double sided/dual membrane K67(or k47, c12) capsule as example. Now imagine only one side is active, the other is not and for all intents and purposes doesn't exist in cardioid mode. You get frequency response like this:

yNwSBvY.jpg


We are interested in the green line which is 180 degree response. Most people think of cardioids as if they reject everything coming from the rear. This couldn't be further from the truth. 180 degree response is just 5 db bellow the 0 degree response at 5K where our ears are quite sensitive. This 180 degree response is created by the sound hitting the FRONT diaphragm coming through the mechanical delay network of the capsule's backplate, and not the REAR diaphragm which is in this mode disabled.
Even though the rear diaphragm is electrically inactive in cardioid mode, it is still acoustically active.
Now imagine we activate the rear diaphragm in such manner that we form figure of eight pattern. And we do get indeed, with an ideal capsule, F8 pattern with identical responses. The sound coming from each side of the capsule is now dominated by the sound coming from both physical diaphragms. We have achieved this by applying polarization voltage, and now both diaphragms are engaged. But what happened to that green 180 degree response we had in the previous image which was created by the sound coming through the rear diaphragm and the backplate and eventually hit the front diaphragm in opposite phase? Did it disappear? Well that's the thing people don't usually think about. It is still here, masked by the electrical signal created by the rear diaphragm. It is still just -5db at 5K, and it is mixed with the electrical signal. However it is PHASE SHIFTED by the distance between the diaphragms (approximately backplate thickness)!!! Of course this is true for the sound coming from both sides, there is no front or rear diaphragm per se, the capsule is symmetrical.
It's easier to understand what's really happening, if you don't think about polarization voltages but think in terms of sum and difference. Cardioid is the electrical signal of the front diaphragm, Omni is front + rear diaphragm (in phase), Fig 8 is the front - rear diaphragm (anti-phase).

A multi-pattern LD condenser in figure 8 often has an almost perfect pattern, at least with quality microphones with a precision made capsule and good front to back symmetry. That's because the figure 8 pattern is produced by cancellation. The output represents the difference ins sound pressure between both diaphragms, which makes it a near perfect pressure gradient transducer. What differentiates it from a perfect pressure gradient transducer would measure the difference in sound pressure on both sides of the same diaphragm. Yes, there is a small distance between both diaphragms. Let's say it's 6 mm (it's often a bit less). That would translate to a time difference of 0,02 ms. Not much, is it?

The difficulty with fig 8 is frequency response, i.e. getting good treble response. That difficulty is basically the same for "natural" figure-8 capsules such as ribbons and multipattern mics in figure-8 mode. And as has been said, multipattern condensers are usually optimized for cardioid. Thats especially true for most classic capsules. But there are newer capsule designs such as the Neumann K89 (U 89, TLM 170) and K107 (TLM 107) which offer a more linear figure-8 frequency response better than most ribbons. Here's a video with the TLM 107 in figure-8 mode:



Omni: Yes, technically speaking, the omni setting on multipattern mics is less satisfying. Frequency response wouldn't be a problem - if the mic was optimized for omni, which it never is. Still, there are some who do a pretty good job. Again, its the not-so-popular newer capsule designs such as the K89 and K107. Technically, the capsule output represents the sound pressure to both sides, so it is a near perfect pressure transducer (and as such has almost no proximity effect). But the large diaphragm puts limits to the off-axis response, so the pattern narrows considerably in the top octave. Another aspect that's usually undesirable: The top octave becomes directional to front and back. The rear lobe has no practical use. A single diaphragm omni becomes directional only to the front, so becomes less sensitive to noises from the rear (e.g. caughs and sneezes from the audience).

Still, although multipatterns in omni mode are less than perfect, some engineers put them to excellent use. The late great Al Schmitt preferred using U67s in omni mode. And the results sounded fantastic!
 
Sorry to be pedantic, but a tree of mics with MS at center is not a Decca Tree - it's just a mic tree.
I did not say it was a Decca tree, only the same mics in the same position as Wallace is a true Decca tree, but the geometry of the setup is based on Wallace's findings, and most people in the biz will recognize not only the physical layout, but also understand the theory of how it works. It is also possible to use an X-Y set of figure 8 coincident in the point position, but that lacks the ability to control the point coincident image width, only the time domain rear flanks. With the M-S in front, I can use it solo first, to establish the best coincident holographic image, and then raise the rears to flesh out the time domain spacial cues, and if needed tweek the M-S a bit. It is an amazingly powerful setup, allowing a wide range of stages and room characters to be adjusted for, and still get a great immersive sound field with excellent focus of individual source points within that larger stage. Note also I am using an omni in the M-S, not a front facing cardioid, a Josephson C617set. The 8 is the Sennheiser RF capsule MKH-30 single pattern, which is very 'fast' and has excellent transient response, as well as decent noise floor.
 
Last edited:
Hello Rossi, thank you for chiming in. Your comments are more than appreciated.

Before i continue, i want to stress this one more time. Please understand that there are exceptions, and if you like using your favorite LDC in figure of eight (or omni) there is nothing wrong with that.

In the light of some of the current posts regarding components and their inherent sound, that IMHO fall in the audiofoolery category, i wanted to bring this topic up, as most are unaware of these effects, which are both audible, and measurable. However, this falls in very nitpicky category, and is questionable if many would care.

Even though the rear diaphragm is electrically inactive in cardioid mode, it is still acoustically active.
I didn't mention this in order to simplify things. Yes, it is active as a mechanical part of the delay/damping network, and is blocking some of the high end, however it is not active in the same sense as the front diaphragm in cardioid. It is not electrically active.

It's easier to understand what's really happening, if you don't think about polarization voltages but think in terms of sum and difference. Cardioid is the electrical signal of the front diaphragm, Omni is front + rear diaphragm (in phase), Fig 8 is the front - rear diaphragm (anti-phase).
I agree 100%. But this is in idealized conditions, both diaphragms have rarely identical electronic paths, and the balance is often disturbed by many factors.

Yes, there is a small distance between both diaphragms. Let's say it's 6 mm (it's often a bit less). That would translate to a time difference of 0,02 ms. Not much, is it?
Not much at all, but it is present. There is none in ribbons, which was my point.

I agree with everything you said, i just wanted to clarify my standpoint, in case anyone got it wrong.

I once again thank you for your contribution, keep 'em coming!
 
Oh, I'm sure it made something ...
I used one mic for mid and one for side in terms of the signal, but of course it didn't actually make a literal mid and literal side recording :ROFLMAO: with some intense finessing, it ended up being the cleanest way to create a stable stereo image of the target that sounded OK in the (terrible) recording environment, and it was coherent enough that it didn't make my neural network reverb removal spit it back out. It took a while, but I did manage to get something usable out of it.
 
Last edited:
Not much at all, but it is present. There is none in ribbons, which was my point.

(bold italic emphasis above is mine)

OK. So I'm going to chime in with something that is half nit-picking and half trying to increase my understanding.

Sure, there is circa 6mm between diaphragms in a dual membrane LDC and only the ribbon material thickness (small number of microns) between the "diaphragms" of a single-ribbon ribbon mic.

But isn't it the acoustic path difference that is important? In the diagram you posted earlier, kingkorg, the circa 18mm path from front to back is even labelled. Isn't this partly why magnet motor structures are shaped like this to be as small as possible either side of the ribbon? With an LDC the acoustic path difference is of the order of the capsule diameter, right?

I'm trying to picture wavefronts coming in at different angles of incidence in my head and think about pressure gradients, diffraction effects, etc.
 
Question, bear with me (as I’m not a physicist)—

The two faces of a ribbon are essentially 100% inversely correlated in their movement, since they’re one piece. Correct?

To put it another way, a given positive deflection on one side results in an essentially-perfectly-equal negative deflection on the other. This is what makes it a velocity sensor… right?

But if we have two separate diaphragms, spaced some distance apart on either side of a shred damping air mass (read: spring), wouldn’t this behavior be expected to change?

I’m underqualified to model this (and I know how dangerous “common sense” can be to scientific understanding), but maybe someone here can clarify.

Obviously the air mass inside might conceivably impact the acoustic relationship between the two diaphragms. But any real-world mechanical properties (e.g. tension; stiffness) of the diaphragm material (and the manner in which it’s affixed and mechanically terminated) could conceivably be different in one membrane versus another, even if just by manufacturing tolerances.

Happy to have someone more-knowledgeable set me right if I’m on the wrong track
 
Sure, path length in a ribbon, along with outer case properties/blockages/reflections have a huge bearing on sound and response. True with condensers too, shouldn't be overlooked. Many of the off axis response properties are the case and total path picture. One of the best looking off axis response graphs I've noticed recently is the EV RE-16! And they sound it!
 
Right! Path length is important for sure. But I'm also interested in the spring compliance of the air mass inside a dual-diaphragm capsule, how this might decouple (or otherwise decorrelate) the behavior of the two diaphragms mounted on either end, and how manufacturing tolerances might also render the two diaphragms non-identical in their mechanical behavior (and the significance thereof).

More broadly, I've come to feel that the off-axis behavior of a mic is the biggest thing that separates a great mic from an "okay" mic. It's not crucial in every application, but when it matters it matters a lot.

It's one thing for a mic to sound okay when addressed directly, but for the off-axis leakage to still sound sensible and coherent is a much rarer accomplishment
 
Is off-axis sound, the difference of a great mic vs an okay mic or is it simply the difference of target user? I mean, is it that 90% of the time, users just need cardioid and 90% of the time, off-axis sound doesn’t matter, as there is no other sounds happening off-axis? So really, is it just a demand difference?
 
If one encounters variable recording scenarios, one needs to know which mics and patterns to pick, it's success/failure level. I am rarely single tracking, there's always bleed. I went from live punk rock to a big band concert recording in the last month; it matters!
 
For Symphonic recording, I combine Wallaces 3 point 'Decca Tree' with M-S in the point position instead of a single mic. For M-S I mount a Sennheiser MKH-30 Fig8, capsule coincident just under a Josephson C617set omni. The rear flanks in the triangle are also Josephson C617set omni mics. A heavily modified 4 channel Millenia Media Systems HV-3D preamp includes 4 raw outputs, and an inboard M-S matrix with L and R mixer, giving a stereo out, direct to 2 track DSD. Mix is accomplished by the main input level controls, resulting in absolute minimum equipment in the signal path. I absolutely prefer the sound of dedicated pattern mics for purist sound in far-field recordings. For near field vocals, I use dedicated single pattern cardioids by Neumann and others, again non adjustable, not chosen for absolute linearity, but for their specific colorations. I have only 1 multi pattern mic, a 1940s era Western Electric 639a/b ribbon/dynamic, but I always use it in one single mode, never combined.

Thank you for the interesting information. I have a few questions. What DSD device do you use for recording, do you do the final mix of all microphones in software or do you have a dedicated stand alone mixer for that, and have you had experience with B&K microphones for recording classic music? Thanks!
 
If one encounters variable recording scenarios, one needs to know which mics and patterns to pick, it's success/failure level. I am rarely single tracking, there's always bleed. I went from live punk rock to a big band concert recording in the last month; it matters!
Of course! I’m also one who has always encountered variable recording situations over the last 25 years (I’ve never even created or used a template in PT), but I’m also not necessarily 90% of users.
 
....The two faces of a ribbon are essentially 100% inversely correlated in their movement, since they’re one piece. Correct?
To put it another way, a given positive deflection on one side results in an essentially-perfectly-equal negative deflection on the other. This is what makes it a velocity sensor… right?
But if we have two separate diaphragms, spaced some distance apart on either side of a shred damping air mass (read: spring), wouldn’t this behavior be expected to change?
I’m underqualified to model this (and I know how dangerous “common sense” can be to scientific understanding), but maybe someone here can clarify.
Obviously the air mass inside might conceivably impact the acoustic relationship between the two diaphragms. But any real-world mechanical properties (e.g. tension; stiffness) of the diaphragm material (and the manner in which it’s affixed and mechanically terminated) could conceivably be different in one membrane versus another, even if just by manufacturing tolerances...

That's a very good question, and I think it's been analyzed far more in complex speaker systems.
An example is a speaker with one active (bass) driver and one passive. In one frequency range, the passive membrane moves in phase with the active one, in another it moves almost out of phase, and in the third it does not move at all.
A similar thing happens with the transfer of vibrations from the rear to the front diaphragm of the microphone. It depends on several factors. On the one hand, it is the mass and elasticity of the air and the front membrane itself, on the other hand, it is the reactive characteristics of the tunnels and passages through which the vibration passes. It's not very simple.
 
Let's not forget there are "true" single diaphragm condensers for Figure 8. And of course those sound different than a ribbon. So that would be a better reference when comparing a multipattern condenser in figure 8 mode.

The treble response of a (good) condenser in figure 8 (native or multipattern) tends to be be better than a ribbon's. Because the acoustic path from one side of the capsule to the other is often smaller. A ribbon will always have magnets on either side, and some sort of construction that holds the magnets in place.

Of course that depends very much on the capsule design. The Neumann TLM 107 has an outstanding figure 8 response (as exemplified in the video above). And it has a lot to do with how the k107 capsule is constructed: Unlike a k67 it has no rings to hold down the diaphragms; they're glued. That's also true for the good old M7, but the M7 was conceived as a cardioid capsule (which was a new thing in 1932!); it was only later that multipattern options were discovered and explored. Being a much more modern design, the K107 was constructed to have an even response in all patterns i.e with a similarly balanced response for both the pressure and the pressure gradient component.

But there were older designs, too, that were pretty good in this regard. I have two KM88 microphones (small diaphragm multipattern) that measure almost identical in cardioid, omni and Figure 8. Or the Siemens SM204 I had on loan once. It had a fairly linear CK12 with only a moderate treble boost. The frequency plots for cardioid, omni and figure 8 were impressively similar. And if I remember AKG's patent description correctly, that's an aspect they consciously addressed.

Bottom line: Yes, many multipattern mics are optimized for cardioid. But there multipattern condensers that perform very well in other patterns, too. It's just that those more universal types tend to be less popular than LD condensers with a more distinct sound character. Which could mean that many users don't care too much about off-axis response.
 
Is off-axis sound, the difference of a great mic vs an okay mic or is it simply the difference of target user? I mean, is it that 90% of the time, users just need cardioid and 90% of the time, off-axis sound doesn’t matter, as there is no other sounds happening off-axis? So really, is it just a demand difference?

I think it matters in more situations than just the obvious. Unless recording in an anechoic chamber, there are always other sounds happening off-axis.

In the adapted spaces (read: sub-optimal acoustic environments) where inexpensive condensers are often used (i.e. "home recording"), reflections and room resonances can easily become significant with louder sources (trumpet, for instance).
 
Back
Top