HO MDC 2-6-0 derailing problem and solution

Folks:

After running with mostly trainset-grade equipment for some time, and fixing a few trackwork problems (talgo-truck trainset equipment with horn-hook couplers is great for finding those), I brought out some of my better stuff and ran an MDC Roundhouse 2-6-0 with some old-time MDC reefers and some scratchbuilt flatcars. It ran well, but for one thing.

Like clockwork, the loco’s lead truck would derail and lead the whole engine off the rails at one of my switches, as shown here:

This is a Snap-Switch, the only one I’d bought new for this layout. The loco didn’t derail on the really really old brass ones, no no, nor on the presumably inferior Life-Like Clone-Switches, but on the shiny one. Clearly this was the work of Decepticon saboteurs. Time to go bug-squashing.

I ran the loco slowly over the switch, and saw that the outer pony wheel was dropping inside the rail, upon which the inner would climb the rail, and bingo, derail. Wheel gauge was fine, wheel tread width was too narrow. Ugh.

Measuring the switch, I found wide track gauge at the points. The older Atlas and Lifelike switches, incidentally, were in gauge.

So I tried adjusting the wheel gauge a bit wider - still within tolerance, but pressed against the outer side of the gauge notches, not centered as I’d had them before. I also squeezed a sliver of wood between the outside of the stock rail and the raised lump of plastic that the Atlas switches have in the throw bar area, at the tie ends, to narrow the gauge there.

This eased the rail-climbing problem, but now the wheels wanted to climb out near the center of the stamped points. Arcing the diverging point rail to smooth the curve (they tend to be reverse-curved from the factory) eased the problem here, but after some extended running I found that occasional derailments still happened. Those narrow

Good post, if I may say so. Careful observation, analysis, trials, changes, and then reporting the outcomes. [tup]