[net.micro.pc] Reset Button Obtained

keithe@teklabs.UUCP (Keith Ericson) (12/15/83)

I got the hint I needed to get the hardware reset button installed in
my IBM pc (older version).  (My thanks to David Trentham at USC-ISIB)

	I tried two approaches: one didn't work, the other did (does).

	For both methods the "pwr good" wire from the power supply to
pin 1 of connector P1 must be disconnected; that is, removed from the
housing connecting to the system board.
	In the method that didn't work I wired a 1 microfarad capacitor
from pin 11 of U11 - the 8284A clock generator - to ground, and a 4.7
kilohm resistor from the Vcc (of a nearby ic pin) to the same pin 11 of
IC 11.  (The idea here is that the capacitor will hold the pin at a
logic low until the power supply has time to charge it - through the
4.7k resistor - to a logic high.) This didn't work because the nearby
Vcc wasn't high enough to get pin 11 up to a high enough level to be
recognized as a logic high: the computer was "stuck in RESET".

	The next attempt - the one that works - was to remove the 4.7k
and replace it with a 1k that runs between the "pwr good" wire
(disconnected earlier) and pin 11 of ic 11.  I left the 1 uF cap in
place 'cause I (a) didn't think it would hurt and (b) didn't want to
put a slodering iron in there any more than I had to.  I then put a
normally-open momentary pushbutton switch in the knock-out hole in the
back panel.  (I got a switch out of company stock that ALMOST filled the
hole.  A couple of 1/2" washers took care of the rest.) One side of the
switch goes to pin 11 of ic 11, the other side goes to ground.  (This
discharges out the capacitor (asserting RESET) which then gets
re-charged through the 1k resistor from the "pwr good" line.)

	Now the machine powers up just fine, and when it does a
nosedive into oblivion - and the Control-Alt-Delete won't recover - I
can hit my reset button and get back to work.

	The next step will be to modify the bootup ROM so as to bypass
the (presumably content-destructive) RAM memory-check procedure.

keith ericson at teklabs
(keithe.tek@rand-relay)