<<I take it this problem is automatically taken care of if the track power is "frog-routed," I. e. the track leading from a turnout thrown against you is dead? That seems to be our situation., If a train is pulled up too close to a turnout thrown against it, it just dies.>>
Actually, if I am interpreting your note correctly, the situation you describe is the exact situation that caused the problem.
Let’s create an example for discussion purposes:
1) Assume we have a right hand turnout, we’re looking at it from the points toward the frog. The turnout is set for the straight route, and a train is approaching the turnout on the curved route.
2) As we look at the turnout, The left rail is DCC1, the right rail is DCC2. Since the turnout is set for the straight route, the right rail, the frog, and both rails extending from the frog are all DCC2.
3) In the situation I was describing, the gap in the rail emanating from the frog on the diverging (right) route was six inches down the spur.
4) The train slowly approaching from the right arrived at that gap, spanned it, and created a short. The wiring in this area is “suspect”, as it was wired long before we understood how to bullet proof our wiring for DCC. As a result, the circuit breaker didn’t trip (track and wiring _do_ _not_ pass the quarter test), and the wheel sat on the gap and creating an undetected short that slowly headed the wheel and axle. Undiscovered for what ever reason, the engine sat on the gap for an extended period of time, heating and allowing the plastic gear/axle to soften and flow around the bearing, effectively freezing the mechanism once it cooled.
Keeping the gaps between the frog and the fouling point, prevents (in theory) or at least greatly reduces the possibility of this situation and the inevitable results from arising.
With rare exceptions, the real railroads would keep equipment clear of the fouling points, model railroaders should too, and that’s the rational for keeping the gaps inside the fouling points.
One picture would illustrate this easily, but unfortunately I have to result to text in an attempt to illustrate my point. If anyone still has questions, feel free to inquire.