"TE Chea" wrote in message ...
Win98se / ME will report "This device is working properly." how
ever IDE devices are connected, but IDE controllers have 2 limits
never pointed out by hdd / cdd / ddd / mboard /chipset makers

Per www.pcguide.com/ref/hdd/if/ide/confCable-c.html
"Using the middle connector and leaving the end connector
unattached... leaves part of the cable "dangling". This is called a stub
and creates much worse electrical characteristics on the cable, due
to reflections from the unterminated ends of the cable wires. It is
not recommended."
I found my Toshiba cdd & LG ddd ( using cable select, & assigned
by Award bios as 2nd channel's slave device ) though can play any
vcd / dvd just as well, cannot pass BCM diagnostic ( blue screen of
"fatal error" appears in Win98se ) unless I use the end connector of
this 80 wire cable to connect this drive ( i.e. bios assigned as master
device ) to my ( VIA ) kt266a 's IDE controller.

This web page had ( in 1998 ) warned that hdd & cdd must not share
the same IDE channel, I found this true : my Quantum hdd ( using
ALi Aladdin IV 's IDE controller ) in 1998 used to make all kinds
of errors while sharing a channel with my Toshiba cdd, until I moved
this cdd onto a separate channel.

Users beware.



The spec is pretty simple. Any lone drive or other lone ide device goes on
the end of the ide ribbon cable. Jumpering as master or slave is not of any
consequence. Cable select, master will always be on the end of the cable.
You cannot cable select a slave and leave the stub. If a second ide device
is added to the cable, the master must go on the end, and the slave in the
middle.

There were some cases in the pre ATA66 spec where one could have a stub
without performance loss or loss of data. Usually a cdrom.

Also not well documented is the low-level ide communication difference
between Maxtor and Western Digital drives. In some cases, you can't get
these to work in the same PC, even on different ide ribbon cables.
--
Jonny