Here's the best comparison of DPA mics you're ever going to hear. I recommend you download the FLACs and give a good listen on monitors and headphones:
http://www.dpamicrophones.com/da/MikrofonUniversitet/StereoTechniques/stereo-recordings.aspx
For me, the ranking is clear:
1. 4006
2. 4060 (but not by much)
3. 2006
This page is what totally sold me on wanting a pair of 4060s.
Very cool resource, thanks for posting that link! I wish there was a decca-tree recording made with 4060 as well as 4006. IMO the change of configuration with the addition of the third mic is a significantly larger difference than the move from 4060 to the 4006 in the A-B array. Considering that one can purchase four 4060 for the cost of a pair of 2006, and probably 7 for the cost of a pair of 4006, I think that's a pretty significant indication of how choice of microphone configuration is more important (after placement), especially if one is willing to use more than two channels to record, and especially if one is not averse to using EQ.
After listening and comparing all the samples, I did two things which some of you may find interesting to try yourselves:
1) I opened each of the omni files in a separate instances of the player (I was using VLC) and used the built-in graphic EQ to adjust each one to whatever sounded most-pleasing. I didn't do this in reference of one to another, trying to make each sound identical, but instead just made each sound "as right as possible" on its own in isolation. At that point I listened and compared them to each other again, switching back and forth. Then made some minor corrections to make them as similar as possible. Pretty interesting.
2) I tried a few simple "live combinations", by playing one of the omni files and one of the caridoid files simultaneously. Thankfully that wasn't difficult to do as all these samples were recorded concurrently, and have exactly the same start point and length. It would be simple to do so by playing them back from a multi-channel audio editor, but I did so using the same VLC players as before by starting one, then starting the other, then going back the first (which was playing slightly ahead in time) and making a couple rapid double-clicks of the pause/play button until the two played in sync without audible delay or phasing. I used to synchronize playback from two R-09s this way years ago before I owned any digital multichannel recorders, and can usually sync them up in just a few seconds. With the files replay looped, two of them remained in perfect sync for about an hour, while I was out finishing some night gardening in back.
That was very interesting and confirmed my experience with regards to choice of configuration for a cardioid pair used in the center between spaced omnis. I've posted pedantically about why I think an X/Y pair of cardioids is a better choice for combination with spaced omnis than a near-spaced pair, so I won't go into all that here. I will say that the 4011s in ORTF didn't sound bad combined with the omnis, but the addition of the omnis (or conversely addition of the ORFT to the omis) didn't change my impression of either listened to in isolation as radically as the combination of the X/Y 4011's and omnis. With X/Y and omnis in combination, I hear a more significant difference - to my ear is a larger improvement over either alone. Both seem to be contributing what they do best, without trying to do the job of the other or getting in each other's way.
I then opened a third instance to compare the 4006 deca tree sample against the combined 4006 spaced pair + X/Y 4011 and prefer the imaging sharpness and drier center image clarity of the later. I wish there was a photo of the mic setup for these recordings, as the position relationship between the omnis and ORTF and X/Y pairs is not indicated - we don't know if the cards were positioned centrally between the omni pair (probably were) or if they were placed in the same plane rather than positioned forward of the omnis like a decca tree, or behind them as they probably would be when used as pairs alone placed optimally with regards to critical distance (however, in this test such fore/back spacing variability is probably of less significance than the minor differences in timing between the two files introduced by my quick, 'good-enough', sync of them).
I don't care for X/Y much on it's own, as it's pretty boring stereo, and if I was only running a pair of cardioids I'd choose ORTF instead, but X/Y works very well in combination with wide omnis, and I'd choose that (or just a single center mic*) over ORTF + omnis.
*I'd also love to hear another decca tree sample which substitutes a single forward facing 4011 as the center mic instead of a 4006. My guess is that I'd prefer that over the 3-omni version, although perhaps not as much as the omnis + X/Y version. (If using an editor for playback, one could easily sum the X/Y pair to simulate a single forward-facing center mic, although that would produces a virtual subcardioid rather than a cardioid) However, I refuse to commit to that personal preference until I can actually make such a listening assessment myself. I plan to do just this kind of experimentation using one of the Naiant figure-8s I've just received, coincidently mounted with a single cardioid center, in combination with spaced omnis. I can then directly compare the omnis and single center cardioid, with the omnis and same center cardioid expanded into a M/S pair, with the ability to dial in various amounts of center stereo width by introducing more or less 8. A M/S center is equivalent to an X/Y cardioid center with regards to phase of the center pair and "recording and combination of signals from 3 separate points in space rather than 4" in the mixdown to 2-channel stereo.