Long winded, but this is the next step I would take in trying to trace down this problem.
I would check the path from track to PCB. Since multiple boards act the same way, this would appear to me to be the next logical area to verify.
I would remove the body so I could access the PCB. I would put the engine on a piece(s) of track that are completely isolated, like a couple pieces of track on a workbench. @gunrunnerjohn identified to you the full wave bridge that the cap you replaced spans. At one end of the cap, the components immediately to the left and right are diodes, whose inputs come from the track. I edited your picture, identifying two of the diodes, and two measurement points that are the ends of the diodes, which I labeled Point A and Point B.
I would us an ohmmeter and put one lead on the Outside Rail (OR) of the track, then measure resistance to the Point A end of one of the diodes and write it down, then from OR to the end of the other diode Point B and write this measurement down.
Now move the ohmmeter lead from the Outside Rail (OR) to the Center Rail (CT). Measure and write down the ohm reading from CT to Point A, and from CT to Point B.
You should end up with two zero ohm readings ( when I write zero ohms, I mean zero ohms or very near zero ohms, like below 10ohms) one zero ohm reading should be to Point A, and the other zero ohm reading should be to Point B, falling into one of the two situations below:
1) OR to Point A and CT to Point B zero ohms
2) CT to Point A and OR to Point B zero ohms
An incorrect situation would be, for example, OR to Point A is zero ohms, but CT to Point B is 200 ohms. Since OR to Point A is zero ohms, CT to Point B must also be zero ohms, and if it isn't, need to find out why not zero (dirty wheels, dirty pick up rollers, wire connection, etc.).
