Passive Non-Derailing: Works Great on Good Rolling Stock, but...
It's hard to find a good blow-by-blow photo series of a derailment in progress. Such derailments, though, typically happen so fast that it's difficult to determine exactly where on the layout the train went wrong. With the help of today's modern equipment, you can get much closer because you can run your trains much slower. After slowing down to the point where I could isolate the switch pictured below on the flatcars in this picture series, I took the attached stop-action series to try and clarify exactly where and how the derailment occurs.
Keep in mind Ross Custom Switches and Z-Stuff do not recommend doing what I'm doing here; the switches are supposed to be thrown to the correct position before the train enters. They still come close to working, and it's only on rolling stock that, well, "needs work" that I have the troubles shown below.
DZ-1000 switch with old-style square-bend spring. It's got a little tension now, as it pulls the points toward us.
Loco has no problem pushing the points. Note the flex on the spring.
Tender also has no problem. Its wheels have slightly narroawer flanges, so there is slightly less flex on the spring, but it's still got a lot of flex showing.
Kit-built 28' flatcar with home-made undercarriage weighted to NMRA-standard 12 oz on MTH arch-bar trucks get through without derailing. You can see the wheel is riding up on its flange, but it doesn't actually pop out. The spring flex is still clearly evident.
Same car with kit-built trucks did not fare as well. These are brass trucks from SMR with NWSL code 172 wheelsets. I have not mastered their assembly, and that I'm sure is part of the problem. There is a bit less flex on the spring than the MTH truck had, but it's very close.
The switch machine's spring seems to be winning this battle against the truck's wheels, as the spring flex appears to be a bit less and the wheel's flanges are riding higher and higher.
Below: the front wheelset finally clears the switch points. The near flange is riding on the railhead; the centering action of the coupler is not enough to pull the wheelset back toward the center of the rails.
Without the added resistance of the front wheelset on the points, the DZ1000's spring is strong enough to pull the rear wheelset up onto the railhead in the next 1/8" of forward rolling.
And finally, the wheels are just plain off-the-track. Bummer.
It's worth noting that the MTH trucks had the advantage of 17% more pulling load (it's pulling 3 cars, while the SMR truck behind it is only pulling 2 1/2 cars). Also the home-made link-and-pin adapter might not be entirely on-center, so the SMR truck may not get the same tendency to center as the MTH trucks did.