Unexpected floppy behavior...

[Tony Duell]

> 
> Well, unexpected by me anyway.  I've built a data sep circuit
> from that schematic I asked for advice on earlier.  Works like
> a champ.  I tied the circuit into a piece of ribbon cable so
> that I could pretty much attach any drive I wanted to try with
> it plus still be able to keep my OSI more-or-less unmodified
> and use the original drive.  The drives I am using are a pair
> of Toshiba FDD 5451s.  After testing each drive individually,
> I attached another IDC card edge connector to my setup and
> tried again.  Nothing worked!  Not only did it not work, it
> trashed the diskette I was testing with.  Went back to one
> at a time and everything worked again.  After a good bit of
> trying to figure out what was wrong, I discovered that the
> +5 volt pin in the power connector I was using for the #2 (B)
> drive had pushed out of the nylon plug so that the drive was
> unpowered.  Having either of the drives unpowered on the cable
> caused the other drive to screw up.  The two drives cooperate
> just fine when both have power.  I don't remember ever seeing
> this kind of problem before... I seem to remember having unpowered
> drives hanging off of cables with no ill effects.  I guess that
> some of the signals (write gate for instance) must be getting
> pulled low by the unpowered unit.  Is this normal floppy
> behavior and I'm just remembering wrong?

It would be expected by me if the drive cable terminator was also losing 
power (e.g. if the drive that lost power was the one with the termination 
resistor pack fitted). 

The correct way to drive the signal lines (both on the drives and on the 
controller) is with open-collector drivers, pulled up/terminated by 150 
ohm resistors to +5V at the far end of the cable (outputs from the 
controller), or at the controller (outputs from the drive). If the former 
terminator loses power, then there will almost certainly be a low enough 
resistance from the terminator supply to ground (e.g. through the rest of 
that drive's electronics) to assert some or all of the signals, in 
particular write gate.... Hence the trashed disk.

The use of open-cpllector drives means it should be safe to power down 
individual drives without causing signals to be asserted.

I had a similar occurance with an ST506 hard disk system. This machine (a 
PERQ) used totam pole (or 3-state) drives onto the drive bus, not 
opne-collector. The problem was partly my fault (I was running a somewhat 
Heath Robinson set-up with the DIB (disk interface board) and drive 
running on top of the machine, etc. Anyway, the DIB's power connector 
fell out, the totam-pole drives then provided logic 0 levels to some of 
the drive control lines, and wiped out the only bootable POS G.7 (24 bit 
version) disk that I have... Oh well...

-tony