Topic: Omnis to figure of eight

[Gutbucket]

Okay, so it works in principle although we haven't yet determined how well and if good enough for our purposes.  Can better establish that after some tests.  Until then, I'll go ahead and jump right to the "ideas for use" driving my interest in this.

I've posted about my wish for a truly miniature fig-8, ideally PIP/low-voltage powered.  I'd use it as the Side channel in a Mid/Side pair with a DPA miniature omni, cardioid, or supercard Mid, probably with both mics mounted in one of the hard-rubber DPA boundary-mount disks, modified to accept the Side mic.  I could use two miniature DPA cardioids to form the fig-8, and for that to fit in the boundary mount I'd probably need to remove the short interference tube grids from the DPA miniature cardioids, which I suspect wouldn't cause a problem.  Alternately as this thread explores, I might be able to use two DPA omnis to form the fig-8.  Both options would be small enough, but neither is attractive due to the increased channel channel count needed for the pair - to do it that way I'd need 3 channels total for the Mid/Side pair, one for the Mid and two for the Side channel.  The clever scheme DPA mentions requires only 2 recorder channels, but only works into a balanced input. It uses two DAD6001 XLR adapters or the equivalent, one for each mic, connected in parallel through a passive summing adapter with pins 2 and 3 swapped on one of the inputs to achieve a differential sum of the two, into a single balanced mic input on the recorder. 

That requires three mics into two balanced recorder channels for a Mid/Side stereo pair.  But I've an idea that would use only two omnis into two unbalanced recorded channels-

If placing two miniature omnis on either side of a small baffle, say about the size of a typical fig-8 diaphragm is able to produce a fig-8 Side channel good enough for what I want to do, I may be able to reduce the channel count needed for a M/S pair back down to just 2 AND use a small recorder with unbalanced inputs.  To do that the two omnis would be recorded to individual channels. Afterward on the computer, those two channels get summed with polarity inverted on one of them to create the Side channel.  But in addition to that, both channels could also be summed without inverting polarity to create the Mid channel.  With appropriate gain adjustment I'd then have both Mid and Side channels.

Conceptually that works best if a small baffle (typical fig-8 diaphragm size would be acceptable) between coincident omnis is able to create a fig-8 which is good enough.  If some spacing between the pair is required either instead of, or in addition to the small baffle in order for to form a sufficiently good fig-8, that complicates the summed Mid channel pattern.  Can probably get away with a little bit of spacing but not too much.

Taking that where I would..
I would use this to record anywhere from one to four Mid/Side pairs.  The simplest two-pair / four-channel application would use them in place of a typical pair of mics worn on either side of the head.  Doing so would provide for the choice of any desired polar pattern afterward.  In addition, if choosing a pattern more directional than omni, separate forward and rearward facing patterns can be produced.  Same principle of operation as the Schoeps KFM360 system.

I like recording using four baffled/boundary-mounted 4060/4061 omnis facing in the four cardinal directions.  I've written a bit about doing that here at TS, but not nearly as much as I do about other, open taping techniques.  The ultimate goal is to extend each of those four omnis into a Mid/Side pair.  Would require eight recorded channels rather than four, but allow for M/S manipulations that would provide increased control over stereo image by working the front-facing and rear-facing M/S pairs, and even better control over front/back direct/reverberant balance and overall sensitivity pattern of the entire array.  I currently have some welcome control over that via how much rear-facing omni is added to the mix, but this would add front/back directivity control via the side-facing pairs as well (KFM360 like).

Although I prefer using a baffle larger than my head for this, all of the mics would easily fit into a hat.  If a bit of spacing works instead of, or in addition to a fig-8 baffle that's small enough, each M/S pair wouldn't be much thicker than a single 406x.  All the mics could even fit into a hat band or a sweat band, early Mark Knopfler style.

I have some testing to do when I can get around to it.

[kuba e]

I think the low mid rolloff corner is a function of the phase differences used to shape pattern in any directional mic. 

^
Yes, it's details concerning this about which I'm curious. Specifically, what explains the multiple octave region (something like 4 to 7 octaves) of relatively flat response that lies above the low-mid roll-off corner and extends up to the mics upper frequency limit where pattern and response are no longer well behaved?

Some of that is probably the result of using the influence of diaphragm/housing geometry/size on high frequency response/directionality.  Might use that to flatten the dipole peak and extend it a bit before it starts rolling back down aggressively into the first combing notch. But I can't imagine that alone flattening the peak enough to extend flat response across that many octaves.

Classic dipole response curve-

It is possible that I did not understand the discussion correctly. if so, please skip it. The theoretical response of fig. 8 rises +6db per an octave before comb filtering as Gutbucket shows in the picture. Microphone designers compensate this by the design of diaphragm - by a self resonance of diaphragm at low frequencies. This resonance compensate the response reduction in the lower octaves. The resonance also determine the lower limit of the frequency range of fig. 8. The upper limit is determined by comb filtering and this is determined by the size of the membrane. All is explained very nicely in a book by Martin Geoff which Gutbucket has pointed us many years ago.
Chapter - Pressure Gradient Transducers:
https://www.tonmeister.ca/main/textbook/intro_to_sound_recordingch7.html#x27-4590006.7.7

Book - Introduction to Sound Recording by Martin Geoff:
https://www.tonmeister.ca/main/textbook/index.html

By the way, it is very interesting topic. It never occurred to me to create fig. 8 from two omni. I would automatically reject it, saying it is not possible.

[DSatz]

Gutbucket, you wrote:

> What I didn't understand about directional microphones (mentioned in my post yesterday) is how the phase offset between arrival at the the front and backside of the diaphragm is somehow kept the same throughout the usable frequency range of the microphone.

For a long time this used to baffle me, too. Then I realized that it's much simpler to think in terms of transit time. Consider a wavefront arriving from behind a cardioid microphone. Regardless of the frequency, it will take as long to travel { through the acoustic labyrinth and the backplate to the back of the diaphragm } as it takes to travel { around the capsule to the front of the diaphragm }. Since the timing is the same either way, then regardless of frequency, so is the signal phase on arrival. Thus the (roughly) equal forces on both sides of the diaphragm cancel each other to a great extent.

It's never perfect because of the multiplicity of paths and the fact that the diaphragm is a surface and not a point, but ~20 dB in the broad midrange seems to be good enough for most applications. In good supercardioid, hypercardioid, and (especially) figure-8 microphones, the null can be even broader and deeper.

That said, the part of a front-incident waveform that gets into the rear sound inlet of a cardioid microphone reaches the back of the diaphragm after two delays: the transit time due to the extra path length plus the effect of the internal "delay line" behind the backplate. For that reason the impulse response of a cardioid is never quite as good as the impulse response of a good pressure transducer, all other things being equal.

--best regards

P.S.: to my mind the DPA posting seems simply erroneous. A kind of noise-canceling microphone for close speech pickup can be made that way (inverting the polarity of one omni), but not a figure-8 unless there's something that I'm really not getting.

[Gutbucket]

Gutbucket, you wrote:

> What I didn't understand about directional microphones (mentioned in my post yesterday) is how the phase offset between arrival at the the front and backside of the diaphragm is somehow kept the same throughout the usable frequency range of the microphone.

For a long time this used to baffle me, too. Then I realized that it's much simpler to think in terms of transit time. Consider a wavefront arriving from behind a cardioid microphone. Regardless of the frequency, it will take as long to travel { through the acoustic labyrinth and the backplate to the back of the diaphragm } as it takes to travel { around the capsule to the front of the diaphragm }. Since the timing is the same either way, then regardless of frequency, so is the signal phase on arrival. Thus the (roughly) equal forces on both sides of the diaphragm cancel each other to a great extent.[..snip]

Thanks for that insight.  Makes sense to me when considered that way.  Its a great help.

[Gutbucket]

I'm now realizing that generating a figure-8 sensitivity pattern from an array of omni sensors (in the initial case just two, but could of course include more) is rudimentary example of beam-forming.


It may help to think of this in that way.  Extrapolating to the intended application..

[snip..] I would use this to record anywhere from one to four Mid/Side pairs.  The simplest two-pair / four-channel application would use them in place of a typical pair of mics worn on either side of the head.  Doing so would provide for the choice of any desired polar pattern afterward.  In addition, if choosing a pattern more directional than omni, separate forward and rearward facing patterns can be produced.  Same principle of operation as the Schoeps KFM360 system.

I like recording using four baffled/boundary-mounted 4060/4061 omnis facing in the four cardinal directions.  I've written a bit about doing that here at TS, but not nearly as much as I do about other, open taping techniques.  The ultimate goal is to extend each of those four omnis into a Mid/Side pair.  Would require eight recorded channels rather than four, but allow for M/S manipulations that would provide increased control over stereo image by working the front-facing and rear-facing M/S pairs, and even better control over front/back direct/reverberant balance and overall sensitivity pattern of the entire array.  I currently have some welcome control over that via how much rear-facing omni is added to the mix, but this would add front/back directivity control via the side-facing pairs as well (KFM360 like).

Although I prefer using a baffle larger than my head for this, all of the mics would easily fit into a hat.  If a bit of spacing works instead of, or in addition to a fig-8 baffle that's small enough, each M/S pair wouldn't be much thicker than a single 406x.  All the mics could even fit into a hat band or a sweat band, early Mark Knopfler style.[/i]

As described above, I've envisioned four Mid/Sid pairs distributed around a circular array. This thread explores the possibility of an alternate way of forming the figure-8 pattern needed for the Side channel of each M/S pair using omnis, making for a reasonable way of achieving that.

An even distribution of omnidirectional sensor elements arranged in a circular array is a recognized and well understood beamforming arrangement.  Okay, so 8 omnis evenly spaced around our Knofler-esque headband.  Apply beam-forming techniques to derive directional patterns.  That works, but is the result up to our standards of quality?

Now thinking the four M/S pairs arrangement, each consisting of two closely-spaced omnis (perhaps with a small diaphragm-like baffle between them) may represent a further optimization for the specific beam orientations of interest.  Rather than 8 omnis evenly spaced around the circle, we now have four evenly-distributed sets of two.  Which sacrifices the isotropic distribution of elements for improved quality in specific beam orientations of interest.. perhaps.

Thinking out loud here.

It would certainly work if we replace each omni pair with a pair of cardioids arranged back to back, although the cardioids will need good response 90-degrees off axis to produce a good omni Mid sum). That's 8 miniature cardioids rather than 8 omnis.

[kuba e]

Thank you to all. I hope that I have understood the discussion well. This discussion helped me solve my old question about ambisonic microphones which are made of omni capsules. I couldn't imagine how to create a directional pattern in ambisonic only with omni capsules. Now, I see that it is teoreticaly possible to make fig 8 with the two omnis.

It still leads me to another, but small question. Directional microphones solve the frequency +6db slope by acoustic and mechanically using resonances, dampings and labyrinths. Ambisonic microphones with omni capsules (and fig 8 of two omni) can deal with 6db slope electronically, but this must significantly add the noise. I can also imagine that they could solve it by complex mathematic models in ambisonic processing.

[Gutbucket]

^ Three methods for making an ambisonic mic:

1) A "native array" of three figure-8 mics and one omni, crammed as close-together as possible.
2) A tetrahedral array of four cardioids, crammed as close together as possible.
3) An array of omnis specifically spaced (most often across the surface of a sphere, solid or open).

The first two methods attempt coincidence. Ideally if possible all mic elements involved would inhabit the same point in space.  That's physically impossible but typically a tetrahedral array is more uniform and can be made smaller than a native array.  Mid/Side is a simpler 2-channel sub-category of this approach.

The last one using all omni elements requires that there is some spacing between elements.  It leverages the phase differences produced by such an arrangement and would not work if all the elements were coincident with each other.  There are design trade-offs in optimizing the radius of the sphere.  This approach is closely related to beam-forming.  The proposed generation of a fig-8 pattern using two omnis is a simpler two-channel sub-category of this approach.

[Gutbucket]

Examples of each-

Native array (no Z channel)-


Tetrahedral arrays (also includes the 8channel 2nd order OctoMic - credit to Len for the photo)-


Omnis distributed on the surface of a sphere-


Omnis distributed in open space-


The last above is a 3d measurement mic. A more advanced implication, but directly related to the simpler goal of this thread. Trinnov used to offer a horizontal plane only ambisonic recording system using an array of something like 8 DPA 4006 omnis arranged in sort of a pointy boomerang/StarTrek medallion shape, which was intended for music recording.  It was a very expensive high end system, so would seem that the goal can be achieved with good audio quality.

[Gutbucket]

Would seem there is at least one real-world commercial application of using two omnis to form a pseudo fig-8 output.  It appears that the stereo version of the Instamic lavalier micrphone (company was recently acquired by Zoom, current version is Instamic Pro C), which is capable of Mid/Side, derives it's Side channel by way of a pair of omni MEMS mounted on opposite sides of its very small lavalier housing.  I'm awaiting good examples of it's stereo output so as to assess quality, and would love to see actual polars, however..

If sufficiently good, four of the Instamic stereo units might achieve the 4xM/S array I mentioned above in earlier posts, and do so wirelessly.   I almost committed to purchasing 3 or 4 of them a week ago to try it out. But decided to hold off for the time being, awaiting the posting of a few more examples of its M/S stereo output.

Link to the TS thread on the Instamic Pro C- https://taperssection.com/index.php?topic=208310.msg2433003#msg2433003
tl;dr- Other than how it's Side channel is derived, which is the primary point of interest for this thread, the interesting and seemingly unique aspect of Instamic is that audio is/can-be recorded locally to each individual unit, thus ensuring quality and robustness, while being controlled and clock sync'd wirelessly via a phone app.  In that way the audio need not be subjected to compression and wireless transmission, but clock-sync between individual units needs to be tight enough across multiple units so as to achieve phase-synchrony sufficient for stereo using multiple units.  Afterward, the individual channels from each unit are each imported to the DAW and sync'd via Time Code.  There are a few examples of using multiple units in mono mode to produce a stereo mix, which seem to be sync'd closely enough.