Making a figure-8 with an omni and a cardioid capsule.

[k brown]

As was done with the AKG CK6, combining an omni and a cardioid out of phase, one gets a figure-8 pattern.

Anyone ever DIY'd sich a thing?

Curious if it has sonic advantages over the back-to-back cardioids approach.

 

[Barry Hufker]

I'm not sure that's right. An omni and bidirectional can give you cardioid, depending upon how the signals are mixed. An omni and a cardioid should just give you omni, sub-cardioid, cardioid depending upon how the signals are mixed. You can get bidirectional from two cardioids back-to-back.

 

[k brown]

I'm not sure that's right. An omni and bidirectional can give you cardioid, depending upon how the signals are mixed. An omni and a cardioid should just give you omni, sub-cardioid, cardioid depending upon how the signals are mixed. You can get bidirectional from two cardioids back-to-back.

The proof is right there in the photo: the AKG CK6 contains a CK2 omni capsule and a CK1 cardiod capsule. The switch selects either capsule, or the two mixed out of phase.

You get sub-cardioid when omni and card are mixed IN phase.

 

[abbey road d enfer]

I'm not sure that's right.

I'm not sure either, but considering Card= 1/2(Omni+Fig8), one could write
2Card-Omni=Fig8
Processing is not only reversing polarity, it should also apply gain of 2 on the cardioid.

 

[Barry Hufker]

I concede the point. I've learned something new!

 

[k brown]

I'm not sure either, but considering Card= 1/2(Omni+Fig8), one could write
2Card-Omni=Fig8
Processing is not only reversing polarity, it should also apply gain of 2 on the cardioid.

Yep - my tests with a pair of Behringer B-5s show this to be true; deepest 90 degree null is with card up 6dB.

Raise the card further, and you get hypercardioid.

 

[k brown]

First tests seem to show that for a given total output level, the combo has 6dB more self noise than a single mic.

 

[RCAguy]

Mixing a coincident cardioid mic and an omni results in a fig8 polar pattern. Mix the omni out of polarity with equal sensitivity. A "downside" is anomalies at shorter wavelengths compared to the space between two small diaphragm condensers (SDC), beginning above 9kHz for 19mm (3/4in) center to center (An LDC mic breaks down on its own above 9kHz). Mixing coincident mics equally reduces noise, as the correlated signal gain is 6dB while the uncorrelated noise gains 3dB, for an improved SNR of 3dB.


The math (ideal) is (Acr + Bcr*cosC) - Aom + B*cosC) where the coefficients (A, B) are each mic's (pressure, velocity) and C is the source angle off axis. The coefficient-only shorthand for a cardioid (cr) minus an omni (om) is (1, 1) - (1, 0) = (0, 1), which is a fig8. Note that native fig8 mics are single diaphragms having, if SDC e.g. a Schoeps MK8 or CCM8, fewer HF anomalies, but not the improvement in SNR.

 

[k brown]

. . . dancing to the algo-rhythm.

 

[RCAguy]

Any two coincident microphones with different polar patterns can be mixed to algebraically add to any other pattern. The math can be fun making a cardioid from a fig8 and an omni, a fig8 from two cardioids, or a hypercardioid from a fig8 and an omni. Here’s how using SDC mics…

The polar pattern for a "1st order cosine" microphone output level is L = A + BcosѲ where A is the non-directional (pressure) coefficient, B is the directional (velocity) coefficient, A+B adding to 1.0, and Ѳ is the incident angle (in 3D). So a cardioid, expressed as L = 0.5 + 0.5cosѲ, can be made by mixing a coincident -6dB attenuated fig8 (A=0, B=0.5) and an equally attenuated omni (A=0.5, B=0). On axis, an ideal cardioid's output (rotated about its axis in 3D) is 1.0; at 90deg off axis it is 0.5 (-6dB); in back it is 0.0 (-∞dB).

In this way, you can make a fig8 using two cardioids aimed opposite as L1-L2 = 0.5 + 0.5cosѲ -0.5 + 0.5cosѲ, where the opposite mic's cos coefficient is minus a minus, or a plus, adding to 0.0 + 1.0cosѲ, the expression for a fig8. (A popular application is M-S using three cardioids, the S mixed from two cardioids aimed left & right mixed equally out of phase for a fig8.) Or make a variable hypercardioid (A=0.25 to 0.33, B=0.75 to 0.67) by mixing e.g. about -3dB a fig8 (A=0, B=0.71) and a -9dB attenuated omni (A=0.29, B=0).


Other combinations prorate A & B to arrive at the coefficients of any desired pattern: a wide cardioid (A=0.75 to 0.67, B=0.25 to 0.33) can be approximated by mixing a -9dB attenuated fig8 (A=0, B=0.29) and a -3dB attenuated omni (A=0.71, B=0). Understanding this reveals some "audio malpractice:" A typical host-guest interview with two cardioids close together is the same as a single omni. And many recordists do not envision when aiming a directional mic or combination above that its polar pattern is actually 3D, spun about its axis.

[corrected]
Mixing a coincident cardioid mic and an omni results in a fig8 polar pattern. Mix the omni out of polarity with half the cardioid's sensitivity. A slight "downside" is anomalies at shorter wavelengths compared to the space between two small diaphragm condensers (SDC), beginning above 9kHz for 19mm (3/4in) center to center (A LDC mic pair breaks down above 4kHz).

The math (ideal) is (Acr + Bcr*cosC) – Aom + Bom*cosC) where the coefficients (A, B) are each mic's (pressure, velocity) and C is the source angle from axis. The coefficient-only (A,B) shorthand for a cardioid (cr) at +6dB minus an omni (om) is (1, 1) – (1, 0) = (0, 1), which is a fig8.

 

[RCAguy]

As was done with the AKG CK6, combining an omni and a cardioid out of phase, one gets a figure-8 pattern.

Anyone ever DIY'd sich a thing?

Curious if it has sonic advantages over the back-to-back cardioids approach.

Mixing a coincident cardioid mic and an omni results in a fig8 polar pattern. Mix the omni out of polarity at cardioid gain of 6dB, after matching on-axis sensitivities. A possible "downside" is anomalies at shorter wavelengths from the spacing between two small diaphragm condensers (SDC), beginning above 9kHz for 19mm (3/4in) units. LDC mic pairs break down above 4kHz.

The math (ideal) is 2(Acr + Bcr*cosC) - Aom + B*cosC) where the coefficients (A, B) are each mic's (pressure, velocity). The coefficient-only shorthand for a cardioid (cr) raised 6dB minus an omni (om) is (1, 1) - (1, 0) = (0, 1), which is a fig8. Single diaphragm velocity-only fig8 mics, Schoeps MK8 or CCM8, have insignificant HF anomalies.

 

[k brown]

The HF anomolies are of little practical concern, as the vast popularity of Mid Side mic'ing shows.

 

[kingkorg]

Simplest thing that comes to mind is to test the concept using two end adress mics bound together. I would go for two that have identical frequency response, i don't think you would get desired effect if responses are different.

 

[k brown]

Simplest thing that comes to mind is to test the concept using two end adress mics bound together. I would go for two that have identical frequency response, i don't think you would get desired effect if responses are different.

I tested with two Behringer B-5 bound together; 90 degree null wasn't perfect at highest frequencies, but about the same as I've heard on Sony's top-line MS mics. Dropping the hi EQ on the card mic at the mixer improves the null considerably.

Since most cards have quite even response at 90 degrees, but very dipped at 180, this idea is most useful for the 90 degree rejection of the figure-8 pattern, not for also picking up sources with the rear lobe (like they did in old radio days, with an actor/singer on both side of a ribbon).

I'll have a chance to try this out on an orchestra rehearsal in a couple of weeks.

 

[RCAguy]

I'm not sure either, but considering Card= 1/2(Omni+Fig8), one could write
2Card-Omni=Fig8
Processing is not only reversing polarity, it should also apply gain of 2 on the cardioid.

[corrected]
Not quite the appropriate math - please see my posts, which in your terms would be more like Fig8 = Card – 1/2*Omni.

[BTW I like your sign-off statements.]

 

[RCAguy]

I previously posted some math hoping some in this conversation would find it models what's possible, and saves a lot of trial & error.

To the OP’s specific question, I’ve used extensively - after understanding the math - every combination of two or more mics from a cardioid-omni pair to M-S to dual M-S surround to OCT surround to discrete Ambisonics. A cardioid at +6dB minus an omni produces a good fig8 polar pattern - sum and select polarity using a mixer with compensation to match on-axis sensitivities. (Compared to SDCs insignificant HF anomalies above ~9kHz, LDCs begin above 4kHz, so can be troublesome.)

 

[k brown]

Somehow this notification 7/6/24 from the OP k brown has disappeared…

OP - “Not sure what the point of all this 'mathing around' is. A cardiod mic coincident with an omni, then mixed with the omni down 6dB, and you get figure 8. All I asked is if anyone has built a DIY mic using this idea. Further, if such a mic had sep outputs for the two capsules, they could be combined in phase or out of phase to achieve every possible polar pattern, but with the advantages of an SDC. Though I realize this is a pretty LDC oriented place .”

RCAguy - Sorry if the OP isn’t interested mathematical modeling - I posted it hoping some in this conversation would, as it leads to knowledge of what's possible, and saves a lot of trial & error.

To his specific question, I’ve used extensively - after understanding the math - every combination of two or more mics from a cardioid-omni pair to M-S to dual M-S surround to OCT surround to discrete Ambisonics. Cardioid minus omni works IF, after compensation to match on-axis sensitivities, you sum and select polarity using a mixer with neither nominally at +6dB - it works. Compared to SDCs, HF anomalies using LDCs begin above 4kHz, so are troublesome.

I deleted it myself because I felt it was peevish and unnecessary.

 

[k brown]

DIY-wise, I'm imagining something like the Josephsn C700A, but with an SDC cardioid in place of the LDC capsule.

The C700A has a five pin XLR, and comes with a breakout y-cable for processing the two outputs individually.

I really should have titled the thread "Omni plus cardioid for any polar pattern.", because if built with separate outs, like the Josephson, it can do any polar pattern; it's just much easier and cheaper to build with cardioid and omni SDC capsules, than with an omni SD and a figure-8 LDC.

With the two outputs mixed in phase it can go card > subcard > open card > omni.
With them out of phase, card > supercard > hypercard > figure-8.

Theoretically, the advantage I see over back-to-back cardioids is that the spacing between the two cards introduces anomalies that vertically-aligned, front-facing capsules wouldn't have. Also much easier to build than something like the Neumann KM 86, with that diffuser disc between the spaced capsules.

 

[RCAguy]

I deleted it myself because I felt it was peevish and unnecessary.

Ah, I suspected a computer glitch. I’ve deleted quips myself, and I’ll take down the quote. But I’ll leave my point that even relatively simple math modeling can open up possibilities and save trial & error. My difficulty is explaining it clearly.

 

[RCAguy]

DIY-wise, I'm imagining something like the Josephsn C700A, but with an SDC cardioid in place of the LDC capsule.

The C700A has a five pin XLR, and comes with a breakout y-cable for processing the two outputs individually.

I really should have titled the thread "Omni plus cardioid for any polar pattern.", because if built with separate outs, like the Josephson, it can do any polar pattern; it's just much easier and cheaper to build with cardioid and omni SDC capsules, than with an omni SD and a figure-8 LDC.

With the two outputs mixed in phase it can go card > subcard > open card > omni.
With them out of phase, card > supercard > hypercard > figure-8.

Theoretically, the advantage I see over back-to-back cardioids is that the spacing between the two cards introduces anomalies that vertically-aligned, front-facing capsules wouldn't have. Also much easier to build than something like the Neumann KM 86, with that diffuser disc between the spaced capsules.

Besides any difference in capsule diameters (affecting HF >4kHz for an LDC), there's the matter of proximity. For many proximity-compensated cardioids, bass will roll off while the omni remains flat. The fix is to compensate the compensation before summing. It's a bigger deal with an omni and a fig8, e.g. a Schoeps MK8 or cCM8 is compensated flat at 12in (30cm), but rolls off below 200Hz at greater distance.