[rickloader]

Having rebuilt the test route, here is a sequence of pics, showing a timetable $transfer. Both the signal heads in the junction signals are type "LSWRStopBranchCallon" having a callon aspect.
PIC1

In the 1st pic, all is well. The active train has Callon, is in trainahd mode, and the signal is displaying the Callon aspect. The track monitor has just lost the "restricted" indication as the loco is passing the signal
Pic2
The active train has $transferred coaches from the passive train, and "has the road" reversing towards the camera.
The following train has entered the loop, reversed and is held at the normal signal as the active train has claimed the route. (ok so far)
The junction signal (with callon) is still showing clear on both heads.
The junction signal nearest the camera is showing the correct aspect for the diverging path of the active train.
Pic3
The active train is now on the diverging line behind the camera.
The following train has a path for the straight route, no callon. Both signals have cleared, and the previous have not reset to stop.

Attached thumbnail(s)

• 
• 
• 

[Jovet]

rickloader, on 04 November 2017 - 09:21 PM, said:

I was misled by getting the .tdb out of sync. Either I did not save my deleted signal, or MSTS RE didn`t. I have now trashed the test route! Can you believe, with my main route backed up on different media and computers, I don`t have a backup of my test route?
The signal type is, of course written to the .tdb, so changing signaltypes alters the saved route. Whereas just changing the scripting (same signaltype) can be applied to an existing save of a route.

Yes, sounds like you broke some rules (see pages 30-34).

You could have manually edited the TDB files to reflect the updated SignalType names for the placed signals after you changed the SignalShape entry. But that is a pain if there are many placed signals affected. The safest course is to never modify a SignalShape's SignalSubObj entries in the sigcfg.dat for placed signals. (But you can add on new entries with no hassle, as long as the existing indexes are not touched and you stay under the 32 limit, and if they're signal heads they must be OPTIONAL and not DEFAULT.)

It's also should be possible to recover your test route in such a situation by doing a TDB rebuild.

Sounds like you got it all sorted, though.