Great question. The TL;DR is to go as wide as you are able to in practical terms. Two feet is reasonable if you can push the center mic forward somewhat in front of the other two. The wider you can go the less the center one needs to be pushed forward, which is helpful in multiple ways.
If an extended answer is welcome, read on..
I understand the more you point (directional) mics forward, you need to increase the distance of the mics to maintain stereo qualities.
^
This angle / spacing relationship (in combination with directional pattern) is the core stereo relationship between a stereo pair of mics.. and by extension it applies to arrays of more than two microphones as well.
Very generally, a minimum of twice the spacing appropriate for an L/R pair is the starting point for L/C/R triplet spacing. That kind of makes sense as that makes the distance between any two adjacent microphones the same as it was before. Problem is the angle between those two adjacent mics is halved, so ideally more spacing is required to compensate for the narrower angle. The other thing driving the need for increased spacing is the angle between mics that you need to achieve PAS.
I’m really starting to prefer PAS method and I will be doing this with the outside pair. The on-axis response, when pointing your mics forward more, is much more preferable to my ears than leaving the mics pointed outward in a standard ORTF and getting more off-axis sound hitting your capsules.
^
This, in my experience is spot on. Problem is that from a typical recording position, PAS typically equates to a rather narrow angle between mics, which ideally benefits from increased spacing between them to compensate. Because of that things get wide fast. Check out the
>>Improved PAS table<<, also linked in my signature line. It's useful as an easy way to determine the optimal spacing between a pair of microphones when Pointed At Stacks, based on the PAS angle from the recording position and mic pattern. Yet that's two mics, not three. Three takes more spacing, but it illustrates the trend.
The Neumann Recording Tools app and Schoeps Image Assistant EmRR mentions are useful tools and work for analyzing a stereo pair of mics. These tools are primarily about predicting image width and stereo image position accuracy, which while nice to have is not the most critical thing to a good recording. But its a good start. These apps can be a bit wonky and tricky to interpret, so don't feel like mastering or fully understanding them is necessary. The Image Assistant [
http://ima.schoeps.de/] is able to analyze a 3-mic array in addition to a stereo pair (not sure about the Neumann app). To do that it will be easiest to start with the Decca tree preset and modify the mic patterns to emulate those of your mics, then modify mic angle (labeled as Epsilon, equating to
half the total angle between L/R mics), horizontal spacing, and center forward spacing to reasonable values while watching the graph display. Read the total image width (the recording angle within which imaging should be accurate) across the bottom, where the diagonal lines curve back to horizontal at lower left and upper right. Best optimization will form a straight diagonal line running from lower left to upper right across the display, meeting up across the center without too much variation. Play with it and try to get the two diagonals to line up (usually takes adjusting forward spacing as well as L/R spacing).
Other resources:
I've attached a couple PDFs. The first covers 3-microphone configurations, except using omnis instead of subcards. You can use a bit less spacing with subcards in place of the omnis if you angle them. The second covers 4-microphone configurations using an X/Y or M/S pair in the center in place of the single mic. Consider doing that if you can. Page two shows the most compact version, which works in challenging rooms and replaces the omnis with supercardioids. That can work with subcards too. Space them a bit more if you can, but don't worry too much if you can't. (These are new revised versions of the old OMT booklet linked in my signature which needs updating)
Michael William's MMAD (Multi Microphone Array Design) website. He's the "Stereo Zoom" author, and much of the stuff above is based on his work. Start here:
https://www.mmad.info/MAD/Ch_n_cov.htm. Similar to the apps, that website all about achieving seamless imaging between microphone pairs, in this case the multiple mic pairs of an array. It's a hyperlinked decision matrix which links to printable PDFs showing different microphone array setups. The 3-mic configuration section was never completed, but the 5 mic section works and starts with the L/C/R front portion, so just choose any configuration angle for the rear-facing mics, and it won't matter if you choose Front Sound Stage Coverage or Surround Sound. It only uses a single mic pattern for all mic positions however, so choose either cardioid or hypocardioid (subcard). Then compare choose between approximate PAS angles. It can be useful to see these known-good arrangements even if you can't achieve them simply as reference points.