so as I look at these "tetra" style mics, I see that the deal is its a cheap form of the soundfield array. Even their new A format mic is basically the same..
but what is the difference between B and A format ?
A-format is the raw feeds from a mic with caps in the 'tetra' form. It is un-corrected and specific to each mic. (imagine the signals output by capsule position- ie: upper front left, lower front right, lower back left, upper back right). In most ambisonic mics (like the Soundfields) it is converted to B-format by the mic itself or its processing box before recording. In Len's mic it is not, and the raw A-format signal is recorded. That allows him to eliminate the processing hardware and do the conversion later digitally. It also means you can not monitor or listen to the recording until it is decoded later.
B-format is the basic universal ambisonic format. This is what you need to 'do something' with the recording like adjust, mix, decode to different stereo or surround formats, etc. Ambisonic 1st order B-format signals are the mathmatical equivalent of and omni and three fig-8 mics, one pointing left, one forward, one up, all of them located in the same point. (W,X,Y,Z) If you were to set up some coincident figure-8's and an omni and recorded the output you would be recording 'native B-format'.
also, the Soundfield MK/ST series, the diaphragm is mounted where the actual capsules would be "screwed in" if you were using pre-bought caps. Does that difference in capsule distance matter ?
Yes it does, and that's one of the reasons that the 'tetra' mic form (and the hassle of dealing with A-fomat) is a popular design choice for an ambisonic mic. The 'tetra' form gets the capsules much closer to truely coincidet than you could by arranging three figure eights and an omni (it's also less expensive to use four cardioid capsules to build the mic). Even so, It's still not close enough, so there is some high frequency correction that is usually done along with the A to B format conversion that helps compensate for the non-coincidence somewhat.
Conceptually, this whole ambisonic thing is an extension of M/S recording. Think of it as starting with a pair of fig-8's arranged for M/S and adding another fig-8 (pointing up) and an omni. The rest is matrix tricks just like decoding M/S.
In this light, consider that the design problems in adpting a 'tetra' style arangement for a ambisonic mic is similar in nature though more complex to those encountered when trying to create a figure-8 pattern by using two opposing cardioid mics and flipping the polarity on one of them.