Perhaps hard-wiring four motors together magnifies the effects of uneven track/roadbed and voltage variations?
I have the same issue, the draft (stretching) and buff (compressive forces) on my trains in my compact 28-inch radius, 3-1/2 turn octagonal helix. Remember, the compensated grade is steeper than designed due to the effects of curvature, my infrastructure isn't perfectly even, and the 1:1 railroads deal with the same competing forces within a train. I speed-matched my locos which are all KA-equipped, but even with load sharing some diesels (Atlas, Intermountain, Bowser) just MU better than others (early P2K). As with the prototype, train handling matters, so operators can't just leave their train to descend the helix and must pay attention to changes in current and voltage and max downgrade speed. As it should be.
I can't speak to long passenger trains, but you may want to see if you can shim any low spots/lower high spots in the track.