[5187] in linux-scsi channel archive
Re: Termination questions
daemon@ATHENA.MIT.EDU (Harvey Fishman)
Thu Nov 19 23:29:23 1998
Date: Thu, 19 Nov 1998 11:25:31 -0500 (EST)
From: Harvey Fishman <fishman@panix.com>
To: Andy Poling <andy@globalauctions.com>
cc: Steven Saner <ssaner@southwind.net>, linux-scsi@vger.rutgers.edu
In-Reply-To: <Pine.LNX.3.96.981118133824.13973C-100000@roadrunner.realbig.com>
On Wed, 18 Nov 1998, Andy Poling wrote:
> Sometimes it's a good idea to disable the internal terminator on the device
> and use an external terminator anyway, since you can then use a
> forced-perfect active terminator (in the case of single-ended SCSI). Most
> internal termination seems to be simple passive resistor-pack terminators -
> not as effective as a forced-perfect active terminator.
errr, "forced perfect" and "active" termination are quite different things.
Forced perfect termination is diode snubbing of only the control lines to
limit the amplitude of the ringing. The termination on the data lines is
still resistive and could be either the resistor pair or the totally
equivalent single resistor back to a voltage supply of the correct value.
This was something that was advocated by IBM for a while, but has generally
died the quiet death that it deserves. I have spent a LOT of years
designing electro-mechanical systems, and while the idea of snubbing is
seductive you generally find in practice that it is a disaster since it may
limit the amplitude of the ringing but it GREATLY extends the duration.
So called active termination is really just replacing the resistor pair
with its Thevenin equivalent circuit of a single resistor of value equal to
the two resistors in parallel back to a voltage source whose value is equal
to the TRMPWR value applied to the voltage divider of the resistor pair.
It does involve an active element to generate that voltage at a low level,
but from a black box view point there is absolutely no way to tell them
apart. They are generally used today for two reasons - one, the hype of
calling it an ACTIVE terminator, and second they can be built as an IC that
is smaller and cheaper than the resistor network. And in some situations
they DO give better results than the resistive network. But that is not
magic. There are two reasons for the better performance. The lesser of
the two is that the voltage source has less noise than a possibly degraded
TRMPWR line. But the MAJOR reason for better results is one of impedances.
The classical termination value for the resistor network has been 132 Ohms
which worked quite well with the usual characteristic impedance of a
twisted pair line (such as that specified for external SCSI cables) which
is about 135 Ohms. BUT the characteristic impedance of the vinyl insulated
ribbon used for internal cables is generally lower than 80 Ohms. The value
that is generally used for these active terminators is 107-108 Ohms, which
is a MUCH better compromise between the 135 Ohms of the external cables and
the 75-80 Ohms of the ribbons.
But in the particular case that raised the question, the device as
described (HP SureStor) is usually an external using the twisted pair cable
so that a resistive termination is likely to give better results.
Harvey
----------------------------------------------------------------------------
Harvey Fishman |
fishman@panix.com | A little heresy is good for the soul.
718-258-7276 |
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.rutgers.edu
