Suspect any limitation in train length will be due to DC voltage drop over long distance (of wire). All the control and protection circuits (door interlock, traction interlock etc...) will come from the rearmost cab relays.
The characteristics of direct current mean you get greater losses as the distance increases. A drop on voltage, might mean relays that should energise/throw do not etc...
There is also only a finite amount of control supply available at the leading end.
As you can imagine, you have a multiple wire control circuit that runs down the length of the train and through autocouplers, this controls train line relays either on each car, or in each non-active cab, or some other method of taking the digital control information from the train lines, usually throwing relays to do so, this passes the control supply demands on to the local car wiring and battery.
You can only control so many of these car (or coding) relays before it's too much for the leading end to handle, as you'd expect, there is a circuit breaker there...
Even in the modern days of PWM control used on Siemens units, you're still limited as to (a) distance that signal can travel and maintain it's fidelity and (b) the power requirements of the remaining coding relays.
Theoretically you could have a train of infinite length if you used something in the uncoupler circuits to 'repeat' the signal in the dead cabs before fidelity is lost, and distribute the train line load along each of the leading cabs, but this is rather complex and introduces significantly more relays, and therefore significantly more issues for people like Newton Heath's finest to work their way around.
