Re: a quick scsi question
While thinking in terms of absorbing is probably not too bad an idea,
it is also misleading. The terminating resistors do indeed absorb
energy (as does any shunting resistance). The SCSI bus terminating
resistance is the same as the "Thin-net" coaxial cable ethernet
termination.
The "problem" is that when an electrical pulse is imposed upon a
"transmission line" (the wire that the bus is of), the pulse will
reflect (or bounce back) from the ends of the cable unless the cable
"appears" to be of infinite length (which is what the termination
does).
The reflected pulses themselves might "bounce" from the opposite
end if the cable is not terminated properly there either. The problem
is that these reflected pulses arrive at various points along the cable
at times later than the original pulses (determined by the "travel time"
along the cable). They can (and do) interfere with valid pulses present
at any point on the cable and of course there is no way for hardware
to determine which pulses are valid and which are reflected.
The matter is a bit more complex than this as the signal level at any
point along the cable is the algebraic sum of the instantaneous pulse
intensities of all of the pulses present at that instant (and the pulses
reflect either "in phase" (with respect to polarity) or "out of phase"
depending upon whether the termination impedance (resistance) is higher
or lower than the cable's impedance.
On Wed, Aug 04, 1999 at 04:39:26PM -0500, Paul Miller wrote:
> Aaron Solochek wrote:
> >
> > This is not debian specific, but I figure someone on this list will know
> > the answer. Now I know you are supposed to terminate the end of each
> > scsi chain, but whats the difference between any of the plugs on the
> > chain? its just a ribbon cable, so wouldn't it be ok if you terminate
> > _any_ device on the chain, so long as you only terminate one? I don't
> > see how there is any electrical difference between the plugs.
> >
> As long as you terminate the last device on the chain as not to starve
> other devices further down. It is generally practiced to have the
> termination found at the end of the cable (either the cable or device
> attached), but I don't see why you could not termintate in the middle.
>
> Remember that the terminator is absorbing the signal so that it will not
> bounce back on the cable to cause interferance. If you were to terminate
> as such
>
> SCSI card <----> Terminated Tape Drive <----> UnTerminated SCSI HD
>
> The hard drive could not communicate with the SCSI card in your computer
> as the Tape Drive would absorb the signal going both ways.
>
> Hope this helps,
>
> --
> Paul Miller
> pmiller@jove.acs.unt.edu
Reply to: