Hey Thelonious-
I'm breaking up my replies to your post up into three parts. 1st, thoughts on monitoring and the different needs of mixing verses mastering-
I would start with a note on playback being critical on these recordings. Depending on my playback setup, I find the DPA only recording anywhere between very good and almost unlistenable. I pulled out my HD6XX (Massdrop/Senheiser colab that I believe is sonically very similar to the HD650) and the DPA sounded pretty good. However, I do find these headphones rolled off in the treble which makes them very forgiving of poor recordings. Great for general listening but I can't master with them. With my HD800S', or my ~1980 La Scala's (which have horns and are in a small/medium sized room) I find the treble almost unlistenable on the DPA recording. This is despite having rolled off between 3 and 4 DB above 8K DB to try to address the bump from the short grills and then further to make it less harsh to my ears.
You've astute ears. Agreed that the HD600 and HD650 which I generally use for assessing and general listening are somewhat forgiving in a few ways. I find them just about right for general listening, but not the best tool for precise EQ decisions (the HD700s I have on hand are out of operation and in need of repair are better in that regard as they are far more detailed, yet have their own quirks - such as an annoying tinnitus-aggravating 7Khz peak).
I've not used HD800S but would like to try them. I have read that their response has a bit of a peak around 11khz, which is the approximate corner frequency of the peak imparted by the DPA406X short grids. It may be that match up perceptually aggravates the excessive energy of the DPA response in that region when using those 'phones more than it does though others. A similar thing may be happening with the La Scala's, as classic Klipsh horns are well known for being rather energetic in the treble. Two ways to look at that - as feature or bug - and as a feature it perhaps makes for a good tool in highlighting that particular response bump issue in the DPAs so that you might correct it with EQ.
You are no doubt aware that the generally perceived response of various headphones is more or less consistent from listener to listener, yet the more specific perceived response details tend to vary substantially from person to person, particularly in the upper midrange and treble. A consequence of headphones being close-coupled to the ear and the interaction of fit, pina-shape details and ear-canal resonance varying from person to person..
[and (out of order) from later in your post..]The difference in sound from playback matters a lot as the difference between the recordings to me comes down to the EQ primarily, with the 22/41 matrix sounding very polished and pleasant on my systems with high end extension and the DPAs sounding bright and harsh. This is also true of the horns where there is a bite, particularly with the trumpet, that the 22s tame well. I suspect this is partially due to frequency response of the mics and partly due to the trumpet being more off axis from the mic compared to the Omni. I suspect I could EQ the omnis more and get them closer to the profile I want, however, I'm still developing that skill and am also trying to align on a mastering headphone that I trust as "accurate" from which to make adjustments.
I keep meaning to make time to dive more deeply into response correction for headphones as a way of making them more useful for objective tasks like mastering. There are plenty of corrective filter tools for headphones, but most are generic to the headphone model and not specifically tuned to the listeners own HRTF. In my limited experience a number of years back, none of those generic headphone response correction tools I tried were really good enough. It took a personalized calibration to get something really usable. Still want to get around to exploring that more deeply. Maybe they are better now, maybe I just need to delve into a personalized HTRF correction using the right headphones as starting point. Until then, I trust good monitors that have been calibrated for in room response more than headphones when it comes to reliable tonal response details.
Might be a good opportunity to work on EQ skills to make that DPA peak acceptable. Lacking a truly objectual trustworthy headphone response that might entail comparing on both sets of phones along with the awareness of how both are somewhat incorrect in that region only in opposite ways, then shooting for something that sort of splits the difference, or at least makes it bearable on 800S and the La Scalas.
This all gets to the important difference between
relative and
absolute response. Mixing primarily deals with relative response - getting everything to work right together, the relationship of things to other things in the mix itself, and that can be done with monitoring that is less than objective as long as you can identify those relative differences. Mastering is more difficult because it attempts to deal with absolute response - how your mix translates beyond your own monitoring to the outside world. Far more challenging to get really right.