This is interesting news. Were your new ones 74491 (2012-) while your older ones were 74490 (last made 2011)?
So that suggests there is a change in the circuitry to the system for the 2016 switches, where a short no longer is the right solution, but an open circuit is.
Did you test the switches once removed - have they changed? In that are they now the opposite type of the old ( normally open NO to normally closed NC)?

If Märklin redesigned the cut-off circuit so that the opposite switch is now employed, I wonder if that oxidation buildup under the lower power is still a problem, or just pushes the time to get a failure out further in time of usage.

Shorting the old switches (to prevent their cut-off feature) would imply the switches were normally closed (so shorting always keeps them 'closed') and thus they never signal to power off the coil. Travel of the coil switch would 'open' the switch at the extent of travel, cutting the power to the coil. Once oxides built up enough on the switch contacts they would no longer attain an electrical closed state and would thus always prevent the power to the coil.

If what you found is indeed the new 2016 design, that now one needs an 'normally open' switch - where removing the switch ensures it stays open all the time - then the coil switch movement would 'close' the switch at the end of travel, and that would disconnect power from the coil. So the closing of the switch shuts off power, and thus shorting that style would disable the function of the coils. In this design the same build up of oxide would lead to the switch, when it fails, to no longer electrically close (it would effectively fail open). In that failure state, you would get the same effect as if you had removed the switches.

So it seems, if there has been a redesign of the coil cutoff switches in the Märklin rail switches from Normally Closed (NC), to Normally Open (NO) - the the failure over time from oxide build up would cause it to fail in a way that didn't prevent it working in digital systems (as it would fail as if you had removed the switches), where-as the analog usage (from Martin's video) would still burn off the oxide from the greater level of arcing, and should not fail either.

Seems to me, if the design change your indicated has happened, then the newer switches no longer need to be modified as they will fail safe. But I'd have expected Märklin to change the part numbers for this change.

Can someone confirm?

---

Ok, I did some more research (I really should work through some research BEFORE posting, my bad). and found a Modelbahn blog entry related to the C-Track switches and the update.
http://mobatechnikblog.b...-weichenantrieb-von.html


So the nature of the switches hadn't changed but the usage of other electrical components and the design had - the result of which is, according to the blog article, that there would no longer be enough voltage to produce any amount of arcing at all on the switch contacts. So this should then prevent any accumulation of oxides formed by arcing (that might be burned off by substantial AC switching arcing, but not burned off by lessor digital switching arcing).

So the change doesn't make the track switch fail-safe, it just makes it less arcing and thus less oxidation build up.

From the blog, the two circuit diagrams:

Old C-Track 74491 circuit (pre 2015?)

When you short circuit the switch, you short circuit the varistor, so the ability to power the coil is always there - power on the input powers the coil.

New C-Track 74491 circuit (modified early 2015?)

Here, shorting the switch takes the capacitor out of the loop and leaves you with the varistor parallel to the coil(inductor) (unlike the old circuit where you'd have the full control voltage applied to the coil only); so I'm guessing something about the supplied voltage on the switch inputs (the voltage of which is highly dependant on what/how that switch input signal is provided ) results in a switch that stalls out in some un-switchable state.