Gear depthing issues

By filing the anchor pallets and escapement wheel, I managed to get the escapement mechanism working well with the escapement wheel, anchor, and fourth wheel.  The next step is adding the third wheel.  This seems to cause problems.  It will run smoothly for a while and then suddenly the power transfer to the escapement stops abruptly.

After carefully tracing what was happening, (mostly by watching and poking at the mechanism when it stopped) I've concluded that the problem is with the meshing of the third wheel and the fourth wheel pinion.  Some of the third wheel teeth seem to jam on the pinion.

So I went through a very extensive (multiple hours) exercise of trying to file teeth on both the pinion and the third wheel.  Some of the third wheel teeth a bit narrower than they should be, and this seems to be where the most trouble occurs.

However, I may have overdone the filing, especially because I now think that the problem is that the gears are improperly depthed.  It seems like the centers are spaced too far apart -- probably no more than about 1/16" and possibly much less. I can reproduce the symptoms in the idealized drawings as shown below.

The trailing leaf of the pinion usually jams, as also shown in the diagram.  There should be (and usually is) a gap that allows the leaf to move into position.  Every so often, it fouls.

Probably, the right way to fix this is to remove the bearings, redrill the bearing holes (filling the inevitable gaps) to proceed.  This seems fraught with difficulty, so I'd like to avoid moving the bearings unless there's no other reasonable option.

So I'm going to try to cut a new fourth wheel pinion that has extended addenda.  This should improve the engagement as the diagram below shows:

Notice the addenda are extended by about the error in depthing -- around 1/16" -- though I can file it back later if needed.  This solution does have a drawback, in that the pinion engages before the centerline.  There's likely some increased sliding friction and power loss as the force is not tangential.  But it seems worth a try.