Here's
the machine disassembled on my office floor. It is apart because it has
three problems: - A failing Miniscribe HDD
- A non-working floppy
- A ram bus test error at E0000
Replacing the Floppy Drive
Here's
a picture of the logic board for a TEAC FD235GH floppy disk drive.
Floppy disk drives have evolved to be $10 devices that ship with a
minimal configuration. Lots of pins are disconnected, and the drive is
soldered for DS1 (i.e. the "second" drive in a system) to eliminate the
need for twisted floppy cables in the PC.
Our job is to restructure the drive so that it lines up with a 1988 floppy configuration.
Here are the details on the 9100a floppy interface.
Unfortunately, there are two problems. Floppy disk drives have not had a /ready line for a long time. And the /CHANGE signal is the wrong sense for the 9100a, which expects the signal to be a CHANGE signal.
The solution is to hack the drive and use drive select to also fire up the motor and drive the ready line. The CHANGE line has to be connected to ground, which can cause problems. Here's what I did:
Open floppy drive
Cut trace to pin 34
Change resistor jumper to DS0
Join pins 12; 16; 34
Jumper pin 14 to ground
I have validated this by formatting and copying 720KB floppies. 1.44MB disks result in CRC errors - wrong bit rates.
The system now boots from the system floppies. But there's a catch. Somebody decided to implement the disk change line to help me switch floppies. So the stupid thing thinks that disk#2 is still disk #1. No idea why they would have done this. The fix was to pull the drive connector, faking a change signal.