[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: high speed scsi trm pwr mod...



Sandra Warnken <warnken@mother.com> wrote:

> You need to have term power at the end of the chain, not on each device.

You are confusing "termination power" with "termination".

A SCSI bus is supposed to be terminated at both ends.  In the case of
the Apple II, this means that the bus should be terminated at the
controller and at the last drive.  (For a computer which has internal
SCSI devices, such as a Mac, the last internal device and last external
device should be terminated.  I think that the Mac also automatically
terminates the bus at the motherboard if either the internal or external
connector is unused.)

The purpose of the termination is to prevent signal reflections at the
ends of each signal wire.  In effect, the terminators provide a
"dampening" effect, swallowing signal pulses that reach the end of the
wire.  They also provide a clean "idle line" condition by holding the
bus at a reasonably well defined voltage.

With the simple "passive" type of terminator, this basically means that
each signal wire has two pairs of resistors connected to it, one pair at
each end of the wire.  One resistor goes to ground and the other goes to
the TERMPWR line, which is supposed to provide power for these
resistors.

An active terminator provides more reliable termination than a passive
one, but it still needs the TERMPWR signal to get power.

At least one device on the SCSI chain must provide a +5V signal to the
TERMPWR line.  It doesn't matter which device does this, and it doesn't
matter if multiple devices do this (the one with the highest voltage
output will "win" and get to supply termination power).  It also doesn't
matter where the device(s) supply termination power are physically
located in the chain. 

Each device that supplied termination power is supposed to use a diode
and fuse arrangement so that the TERMPWR signal doesn't try to supply
power back into the device (and to protect against electrical faults).


The original Apple SCSI card is not terminated on the card, which means
that if you have multiple devices connected, you should have a
pass-through terminator between the SCSI card and the first device, plus
a normal terminator after the last device.  (Alternatively, internal
termination can be used on the first and last device.)  If you only have
one device connected, it should have internal termination, or external
termination after the device.

The Apple High-Speed and RamFast SCSI cards _are_ terminated, so you
should only have one external terminator, connected after the last
device on the chain (or use internal termination on that device).


Neither Apple SCSI card originally provided power to the TERMPWR line.
Late in the production run, Apple did a tack-on modification to the
high-speed card, adding a diode, which allows the card to provide power
to the TERMPWR line.  The original card could be modified in the same
way.  The RamFast SCSI card can supply termination power (I think it has
a DIP switch or jumper to select whether or not it does this).

My personal feeling is that it is not a good idea for the Apple II to
provide termination power, because it can draw quite a lot of current -
almost the entire budget for the +5V rail going to the slots in the
worst case.  My preference is to ensure that at least one of the SCSI
devices connected to the card is providing termination power.  This
avoids placing a heavy load on the Apple II power supply.


If termination is not set up correctly, you can get all sorts of weird
effects.  For example, if nothing is providing termination power then
the terminators acts as signal dampers, and are likely to cause
communication errors over the bus.  You may find that the bus works by
removing the termination, though this can cause interference due to
signal reflections.

SCSI can get very weird with badly behaved devices, cables or
connectors, and you may find that you can only get a particular
combination of devices to work by arranging them in a specific order
and/or breaking the termination rules (e.g. adding extra terminators in
strange places, or disconnecting one or both terminators).

-- 
David Empson
dempson@actrix.gen.nz
Snail mail: P O Box 27-103, Wellington, New Zealand