I don't see objectionable grain but I see slight decrease of speed with TMX in D-76Ad 1:1 at 20C. With this stuff I'm happy if I get better result with HP5+. I'm trying something else for TMX anyway.(posted 8321 days ago)I don't know where 1.8 came from. Maybe molar weight ratio. The XTOL patent also states that the ratio of ascorbate and superadditive reagent should be pretty high, at least 20:1 if I remember correctly. This is where Gainer's formula is at. But some reported ascorbate is more strongly superadditive with metol than hydroquinone. My ratio is smaller than that, and yours between. I think this parameter may need tweaking.
Your solution is probably around half a pH unit higher than XTOL, at least before you soak film in it. I don't exactly know how XTOL achieves finer grain than D-76 with several films, but many think lower pH gives one reason. XTOL patent specifically prefers a suitably formulated borate buffer solution. We kinda know what it is. So pH is one thing I am tweaking now.
Buffered low pH and the property that dilute solution pushes better is kinda common with Microphen. So I'm suspecting some of XTOL's marketed advantage came from buffered low pH to which ascorbate adds something. Microphen or ID-68 might have been a better starting point for this project, I might think, though I use metol.
There has been a criticism about adding sulfite to Rodinal as a panacea because all it does may be just lower the pH profile a bit or regenerates reductants in a different way, or something totally different from solvent effect. Ascorbate might also be the case at least in part when added to Rodinal-type developer. Gainer's common salt experiment is interesting, but I wonder if it works equally well on fast films and T-grains, which may give some clue in further speculation.
Gainer's formula also contains less reductants and more base. I don't know if this is a result of image quality tuning or his cost calculation (probably both) but I think D-76Ad 1+3 or 1+4 for T-MAX and Delta may suggest something.
Just an ascii text dump of what I am thinking about. Either way, I know formulating a developer isn't that easy like make one up and it works great. Just as hard as making a really great blend of coffee :-)