After a discussion with Roy who was extremely helpful and a few tests I think I am a little closer to understanding what's happening here. Roy mentioned that he had not played around with ABW for some time but that he had found with QTR that OS-X always
did a profile conversion to Generic Gray Gamma 2.2 Profile. This is obviously a significant, and unhelpful, intrusion into the prior workflow and the fact that it isn't known to the user is very annoying.
I printed a step wedge 3 times to test this with ABW: (1) untagged, (2) converted
to Gen GG 2.2 in PS (perceptual intent and BPC) and (3) assigned
to Gen GG 2.2 in PS. All the prints were using ABW's default settings for "neutral" and, of course, Printer Manages Colors. (1) and (2) produce basically the same output (some printer/ink drying variance but very similar). With (3), whereby the K values are not altered by the tagging, produced a rather different L* output (most notably, unsurprisingly, in the darker regions).
Now hopefully the guys at Epson know that this is going on and that the receipt of a file tagged Generic Gray Gamma 2.2 doesn't upset their ABW driver or they've adjusted the OS-X driver to accommodate for this constant. If anyone out there can shed some light on this point it would be great.
If the above is correct then some relatively modest workflow adjustments are required to both print targets for QTR ICC profiling and final images. (Presumably this can be applied by Eric to his workflow as well.) In OS-X, we need to assign
the Generic Gray Gamma 2.2 profile to the step wedge prior to printing with our preferred ABW settings (assuming we want to profile with neat steps of K). Read the step wedge and generate the profile as before. For image printing, prior to sending the image to print we need to convert
the image to the QTR ICC profile and then assign
Generic Gray Gamma 2.2. Then we can print with Printer Manages Colors and the ABW print settings associated with the QTR ICC profile we used.
I would greatly appreciate it if others could confirm the tests I did.
Ernst got it very close here
but no one progressed that discussion in the earlier thread. Pity.