Erik wrote:

Make sure you limit heat generation since your stuff is in the model, it would be embarrassing if you caused the model to catch fire.

One need not to be paranoid (what I'm always...) to imagine, what happens, if a model building made of dry wood and painting catches fire for some reason! This can end up with total catastrophe, with lots of killed people!

Once, I saw a little Christmas tree catching fire. Within a few seconds the whole pomp burned, biting smoke was arround and temperature was unsanely high. Only because a paranoid person (no, not me...) had prepared a bucket of water, worse could be prevented...

Please, Mahmood, be very careful! We would miss your interesting threads...

Erik also wrote:

Also, With 1000 lights on you have 180A and using devices with a Vcesat of 0.5V, that alone generates 90 Watts of heat. Maybe you should consider to equip your 'electronics cavity' with a fan.

Exact! That's why I proposed to drive each lamp by the help of a single transistor instead of Darlington. BD135 is specified to show an Uce of about 0.1V with 180mA collector current. So, typical heat dissipation will only be about 18mW.
But not to forget the 2R7...3R0 resistors! Voltage drop across them is about what Erik stated, namely about 0.5V. And as these resistors are not mounted on a cooling plate, the use of cooling fan seems to be essential.

And, as Steve already mentioned, take care, that current limiting action is NOT invoked with nominal current. Otherwise, Uce will rise! Ok, proposed self diagnosis circuitry will detect this condition if collector voltage rises above 2.0V. But this would make shut-down the whole involved floor...

Maybe it's a good idea to set the current limiting threshold to up to about 2 x 'nominal current'. Then, a 1R8 current sensing resistor should be used. This would decrease dissipated heat per driver and decrease the danger of enroneously invoking current limiting. But then, BD135 MUST be mounted on a good cooling plate, like Michael suggested.

Do also make some tests with oscilloscope, whether current limiting invokes properly. Means, BC817 must be able to detect overcurrent condition fast enough. No glitch current exceeding current limit is allowed to be observed! I think theres' no problem, because BC817 should be much faster than BD135. But maybe I'm wrong...

I would be thrilled, if you could publish a photo of this model building!

Good luck,
Kai