Cable Select vs. Master/slave settings

[BRUCE STANLEY]

--- Gene Heskett <gene.heskett at verizon.net> wrote:

> On Sunday 07 August 2005 15:53, Tony Nelson wrote:
> >At 3:19 PM -0400 8/7/05, Gene Heskett wrote:
> >>Nice idea Claude, but can you tell us how to tell the difference
> >>between the cables so that we can properly identify them?
> >
> >Just google on "cable select" (with the quotes) and you'll find that
> > a Cable Select cable has pin 28 connected to a ground wire at one
> > device connector (master) and not connected at the other (slave);
> > normally this is done by punching out a little bit of wire 28 (on a
> > 40 wire cable) just past the middle connector, so you'll see a
> > little hole in the cable.  You'd also find that, as normally a
> > single device should be at the end of the cable (slave) and a
> > single device should be master, so using a Cable Select cable with
> > only one device is, umm, problematical.
> >
> >If one must use a Cable Select cable for only one device and it
> > doesn't seem to work quite right at either position, I suppose one
> > could just cut the cable after the middle connector, which would
> > make it the end connector.  Cables are cheap.
> 
> That solution has much to recommend it in terms of signalling 
> integrity.  As a general rule, there should <b>never</b> be any cable 
> beyond the last connected device, and this is particularly dangerous 
> in a scsi environment where as little as 3" of unterminated cable can 
> and will cause data error problems from the open end echo's.  I've 
> been looking at udma100/133 cabling for years trying to figure out 
> just how the hell they get away with the sloppiness and still make it 
> work.  But that sloppiness in termination also restricts the maximum 
> length (18" absolute max) to something thats not always reachable in 
> a big tower case.
> 
> -- 
> Cheers, Gene
 


18" max?

I just orded two new UDMA 66/100/133 ROUND (inclosed) cables for my
system that are 24" long.  They also had 36" lengths..