James,

I am not going to try to change your mind about having a way to
change the train id.  And (this may sound cold) I am not
interested
in debatting your reasons for you wanting this ability.  So the
following
information is for thoses that are faced with a simular problem.

Software and hardware only does three things.  It works, it
doesn't work,
or it does something unexpected.  Once it works, it will never
get any
better, it can only stop working or give unexpected results.  So
by
NOT installing pieces of hardware or software, it will NEVER go
bad.

Clive Sinclare had a way of designing things.  He designed the
little
Z80 computer that sold for $100 while the rest of the industry
was
charging $2,000 for their computers.  Any way, his way was to
design the thing and get it working.  Then he would start pulling
out
parts until it stopped working.  After putting that part back in,
he would
continue by pulling out other parts.

By not putting something into the mix, there is no need to pull
it out
later.  And after the thing works, you will see things that are
not
necessary or that can be done a simpler and better way.

But one has to be carefull that the hardware and software is
designed
so that possible features can be added easily.  The question "Why
can't
we...?" can become a command from hell if one does not keep a eye
on the future.

While trying not to be negative about the ideas that have been
sugested,
and the one you have selected, I am going to describe some of the
difficulties you may expierence with this method.

It requires software and hardware.  The software for setting the
id on pic reset and on "track data grounded"  Also you mentioned
having the pic use one of the pins (normally a output for a
light)
to send its current id serially.  This is not trival software.
The hardware, while simple, will require board space and will
require adjustment at some point after it is set and painted
shocking pink.
This adhustment may not be necessary for a day or thirty years,
but it
can't get any better, it can only go bad.

You describe the method for setting the id. It looks like this
would involve
removing the train from the track. dissasembly so you can get to
the board,
and hooking the PC serial line to a pin.  And then setting the
id.  If there is
any question about a train's id, as when things don't work like
you thank
they should, the same procedure will be necessary to confirm the
id of
one or more trains.

While I do not know your reason for wanting to assign a specific
id
to a specific train, one reason may be to have the id's reflect
the
engine type, number or line.  All good reasons.

Keeping this in mind, I make a suggestion.  As you add controls
to
the trains, program each pic with a unique ADDRESS.  Please
notice
I said ADDRESS.  Then in the PC control program, have a routine
that
will assign a specific train id to a specific pic ADDRESS.

Now here comes something that is a little off topic but is
important.
As people we number things starting with 1 and go to n.  1, 2,
3....n
Computers number things starting with zero,  and go to n-1   0,
1, 2....n-1
When the first person thing is refered to as #1 and the computer
knows
it as unit 0, confusion, the people kind, will occure.  So either
use the same
numbers as the computer uses, and you don't have to use number
zero,
or use some completly different method of refering to the people
things.
Such as bird names  robin, red bird, wren ...... owl

It seems like for trains it would be approate to refer to them
using their
engine number.  And let the big, all powerful and knowing
Personal
Computer translate between the engine number (id) and the pic
ADDRESS.

Submitted for your viewing pleasure,
Bill C.           bill@cornutt.com

>For the locomotive model driver chip, I'd use a trim-pot.  A
multi-turn
>pot would be good.  Using the presumably 5v regulated supply on
the
>circuit would be the most stable, without having to add a
voltage
>reference.  I'd run the A/D in the way you described, and have
it read
>at two times;
>
>1)  at reset,
>2)  on seeing a continually grounded track data input line.
>
>While the track data is grounded, I would have the chip read the
value,
>then convert it to RS-232 serial 2400 baud and emit it on one of
the
>pins used for controlling the locomotive lights.  Continually.
>
>So the procedure for setting the identification becomes;
>
>a)  ground the track data input line,
>b)  connect a PC to the RS-232 serial output,
>c)  apply power to the chip,
>d)  watching the ID displayed on the PC, adjust the trim-pot to
be
>    midway between the thresholds of the desired ID,
>e)  apply significant other's nail polish to trim-pot.
>
>Some temperature stablisation may be required.  ;-}