I should preface by saying there are two general approaches to our type of recording: capture as much direct sound as possible, or capture the sound and environment as best as possible to duplicate it during playback. An example of the former: stack taping; the latter: ORTF at some reasonable distance. My assumption for this post: we want to achieve the latter, i.e. record the performance in a way that, upon listening with a good playback system, provides us the feeling of "being there".
Check out the link to Williams' Sterephonic Zoom (SZ) in the stickied thread at the top of the forum. It's not the be all end all, but may help provide an understanding of how the configuration impacts the mic array's characteristics, and therefore impacts the results.
Based on Williams' SZ, you should know that running coincident (0 cm spacing, i.e. XY) and reducing the included angle will produce a very wide stereophonic recording angle (SRA). For example, 0 cm with 70º included angle will produce an SRA of approximately ±100º. That's plus or minus 100º from the mic array's center line. So if you're standing and looking down the center line of your mic array towards the sound source, -100º is just behind and over your left shoulder, and +100º is just behind and over your right shoulder: a total span of 200º. From your recording position, does the sound source span 200º? I highly doubt it. The result: using a narrow included angle in a coincident configuration will help reduce reverberant sound (room boom), but you'll end up with a very mono recording because both mics will be picking up nearly the same sound. (FWIW, the only way you'd have a soundstage of anywhere near 200º is basically by running on-stage.)
Personaly, I'd run hypers to help minimize reverberant sound. And a good starting point is to use a configuration with an SRA that's close to the width of the sound source. Here's what I'd do...
- Guesstimate the width of the sound source from the recording position, in degrees. Someone here mentioned long ago that a closed fist at arm's length covers approximately 10º. For the sake of discussion, let's say we've guesstimated the sound source spans approximately 60º from the recording position.
- Using the Williams' SZ graph for hypercardioid mics (pg 10, Figure 10), find the SRA that closely matches the width of the sound source. In this case, an SRA of ±30º closely matches our sound source width of 60º.
- Since I want to minimize reverberant sound ("room boom"), I want to use as small an included angle as is reasonable -- this ensures I pick up more direct sound and less reverberant sound. So, follow the SRA ±30º plot line to a point where the included angle (on the vertical axis) is relatively small. The plot line and graph end at ~50º included angle, so that's the included angle we'll use.
- Where the SRA ±30º plot line hits ~50º included angle on the vertical axis, note the spacing on the horizontal axis (~50 cm). This is the spacing we'll use.
So, what does this tell us? Using hypercardioids, a spread of ~50 cm and included angle of ~50º will produce an SRA of ±30º. This basically will provide complete stereo coverage of our 60º wide sound source -- not far too much stereo coverage (like 0cm, 70º, which would produce a very mono recording), or too little. Alternatively, we could achieve an SRA of ±30º using hypercardioids by spacing ~33 cm wide with a 100º included angle -- meaning we still have complete stereo coverage of our 60º wide sound source...not too much, not too little -- but the wider included angle means we'll pick up more reverberant sound in the process. Which, of course, in this case we want to avoid.
So...that's what I'd try. Again, this is just one of many potential starting points. Results may vary using different microphones due to differences in the polar pattern. Try something, see how you like it, then adjust next time based on your perception of the results.