I did once see a picture of a 'proper' plate-and-ramp diff for a model car, probably a 1/8th. It only appeared to have one plate each side though, so it had ridiculously shallow ramps angles - maybe 15-20°.
In theory, a ball diff provides more torque-dependent locking than a gear diff, because for the ball to rotate it has to slide inside the hole in the 'cage', whereas a gear diff's planets rotate on small pins.
In practice, I once measured a pan car ball diff at about 1.4:1 torque bias ratio (changing how tight it was only affected the initial 'preload' torque, and made almost no difference to the TBR under load) - but 1:1 'open' gear diffs also produce about 1.4:1, so it would be interesting to measure a typical model gear diff the same way I did on the pan car.
Having sealed diffs so you can fill them with thick oil/grease seems like an excellent solution to me - that way you get some speed-dependent locking, exactly the same as a 1:1 viscous coupling (VC) diff.
Torsen diffs provide torque-dependent locking (not speed dependent, Sparrow) and 1:1 ones can do 4:1, maybe a bit more. Presumably a Torsen with thick grease then also adds some speed-dependent locking...
As mentioned though, it's difficult to design something effective in 1/10th size, which is probably why we use spools and one-ways a lot - basically a spool is a diff with infinite TBR, and a one-way is the same but only in one direction (a bit like different drive and coast ramps in a plate-and-ramp diff).
I haven't seen the ATD but it sounds like it effectively gives adjustable preload - but it would be nice be use actual TBR as a setup tool as well.
Careful now, you've got me started on diffs...
