There are a few more aspects to the question posed by Phil Dayson
concerning designing around major transistor failures.  In addition to
designing circuits conservatively so that, in the event of transistor
breakdown, one isn't faced with disastrous fires, the software designer
can include sub-routines which can recognize problems and head them off
before fires result.  One commercial designer employs a method wherein
the microprocessor scans the note-on data and automatically turns off
any solenoid which has been on for more than 12 seconds.

Of course, this safeguard will not avoid transistor breakdowns caused
by dead shorts across the transistor leads.  For example, in the
Pianocorder System which uses 170 Volts DC on its driver transistors,
spider webs, in the presence of high humidity, can short out the output
transistors.  This condition can char the affected transistors and can
burn holes in the circuit board proper.  To date, I have yet to hear of
any Pianocorder System causing actual fire damage to a piano itself.

On the other hand, one of the very active installers of a competitive
system told me of five serious fires in an earlier design.  One of the
interim fixes used by that company (and since discontinued) consisted
of including output fuses on each output transistor.

Still another competitive system's earlier design was subject to
microprocessor glitches which would freeze up the entire system and
cause the solenoids' plastic liners to physically melt.  This
phenomenon "glued" the solenoid actuators and rendered them useless.

Regards,
Bob Baker

 [ The resistance of the copper wire of the solenoid coil rises quite
 [ measurably with the temperature.  It shouldn't be difficult to add
 [ a coil resistance sensor circuit, to monitor temperature, to the
 [ already complex driver circuitry in solenoid reproducing pianos.
 [ (Maybe it's already there in new models!)  -- Robbie