After running more tests, I've come up with the following modification to D316 which gives a tighter toe with Delta-400. The 1-litre formula:
Propylene glycol ....................... 10 g (yes, measured in grams)
Sodium metaborate 4 mol ........ 1.5 g
Ascorbic acid ............................ 3.5 g
Phenidone ................................ 0.05 g
Use 14.7 g/L of concentrate with 50 g/L of sodium sulfite. pH = 8.18.
I'm calling this formula "Trial-130119", as that's the date of the first mixing (yesterday). This is also more concentrated than D316. It is very close to something I mixed 4 months ago, so it should not crystallize, but we'll know for sure in a couple of weeks. Note that I raised the sulfite from 45 to 50 g. That was solely to raise pH, but I might drop it back to 45 g. The time for Delta-400 is the same as D316 (17 min), and we'll see how times for other films are affected.
Grain is still a bit finer than Xtol, despite Xtol having the advantage of lower contrast. The scanner was manually focussed on the middle wedge and contrast adjusted to be equal:
Xtol: 13-20-Xtol.jpg Trial-130119: 13-00-Con119-50g.jpg
And the curves. Note that the toe is identical to Xtol:
As with D316, highlights are stronger than Xtol.
Fog is higher than Xtol. I noticed D316 produced a bit higher fog with a few films. Trial-130119 behaves the same. My experiments indicate that the fog is likely due to sodium metaborate mixed into propylene glycol, which Gerald Koch said produces esters. Could those esters boost fog? With both D316 and Trial-130119, Delta-400 has 0.12 more fog than Xtol, which is a lot. Fortunately, the scans show that grain is not affected. Maybe add a restrainer anyway?
So why am I fooling with this after burning over 50 rolls testing D316? Because I'm a perfectionist! Those tests showed that D316 had slightly softer toes with a few films, but Delta-400 already has a somewhat soft toe, and I am loath to make it any softer. Delta-400 is an important film, so I want my dev to work well with it.