Date   : Fri, 21 Oct 2005 11:43:30 +0100 (BST)
From   : Pete Turnbull <pete@...>
Subject: Re: Dual Floppies On A Master

On Oct 21 2005,  9:57, Richard Gellman wrote:
> Tim Fardell wrote:

> >Isn't the termination simply a bank of resistors between all the
data
> >lines and ground? If so, wouldn't it be fairly easy to detect the
presence
> >or absence of such termination?
> >
> Thats the cheaper (read: lazier) form of termination, and it works
> *most* of the time. Proper termination actually involves two resistor
> banks, one tieing the lines to ground, and one to +5v, usually at
offset
> resistances. This causes the bus lines to "float" at roughly 2V.

It may be the lazy way but the pullup resistors used are the standard
method for 5.25" and 8" floppies.  The common 220/330 ohm and 180/220
ohm termination networks used elsewhere (eg SCSI) actually set the
lines at around 3V or 3.5V, not 2V -- that would be far too low.  In a
normal 5.25" floppy system, there is supposed to be a 180-ohm pullup to
+5V (not to ground!) on each signal line.  The best termination, of
course, uses a single resistor for each line, connected to a fixed
low-impedance voltage reference of appropriate value.  For example
active SCSI termination uses 100-ohm or 120-ohm resistors (to match the
bus characteristic impedance which typically is between 85 and 110
ohms) tied to a 2.85V supply (less for LVD, of course).

> Termination of floppies is a valid point though. I remember many
5.25"
> drives simply couldn't be "doubled" because the resister pack was
either
> absent, or more frequently hard wired (doubling involving cutting
> resistors on one drive from the board). I'm not sure how modern 3.5"
> floppies terminate, but I'm guessing they don't bother too much about
> auto-detecting on the really recent drives (who wants two 3.5" floppy
> drives on a PC these days?).

They're supposed to have a 1K terminator, or rather a 1K pullup, on
them.  This is not really a terminator in the transmission-line sense,
it's just to ensure the lines tend to float high rather than to some
indeterminate level.  The value is chosen to be high enough that having
several drives on the bus won't overload the bus drivers.  I've never
known a drive have what's sometimes described as auto-termination
(detecting the need and enabling or disbling the local termination).

> If you have the equipment to hand, you can test for termination by
> plugging in power and checking the voltage on one of the input lines.
If
> its terminated, it'll float at around 2v.

No, much higher.  With a simple set of pullups, it'll float at about
5V, and with a termination network, typically at about 3V (DEC
equipment and some others use 3.5V or thereabouts on various busses)

> Something to keep in mind, the floppy connector, unlike say, an IDE
or
> SCSI connector, is not a bus - i.e. it doesn't have "data" lines,
> there's no strobe line, REQ/ACK, etc. Its a *very* low level
interface,
> that basically consists of motor enable (switch on), head step, head
> direction (when the step occurs), track 0 detect, sector 0 detect,
read
> bit stream, write bit stream, write gate (to synchronise writes to
the
> correct points on the disk), side select (:0 and :2 to our DFS
> afficianados), and an access clock to keep the drive timing in line
with
> the computer.

I disagree, it's a perfectly ordinary bus, in the sense used by
electronics engineers.  It's a set of signal/control lines to which you
can attach several devices in parallel, along its length.  It happens
to be an asynchronous bus, and there is no clock on it.

> (Oh, and the lil' red light on the front? Tied to the motor enable
circuit).

That depends on the drive.  Normally it's tied to the select line,
actually, but that may in turn be linked in some way to the "motor on"
line.

-- 
Pete                                           Peter Turnbull
                                               Network Manager
                                               University of York