IDE disk problem

[Joel]

[...]
> Cable select is exactly what it says, and is VERY reliable in my
> experience.

I used to repair computers with 5 1/4" floppy drives, and in those days,
CS was an epithet. Been a while since I've messed with mainstream
hardware. I'm going to have a lot of bias to get over if CS really is
reliable these days.

> -Use a cable select cable
> -plug the drive to be used as master to the connection designated as
> master.
> -plug the drive to be used as slave to the connection designated as
> slave.

So, if it's a CS cable, it should say on the connectors something like
"motherboard", "master", and "slave"? If it doesn't say, I can assume
the guy at Pasokon Koubou knew I didn't want a CS cable? Or maybe this
whole business is why he didn't want to recommend using the slave
channel in a multi-drive, multi-boot setup?

I guess I'd better take a look when I get home tonight. I don't want the
drives fighting with the cables on my new cheap Linux/BSD sixOS
multi-booting toybox.

> -jumper both drives as cable select (CS)
> 
> With this config it does not matter if you have 2 drives or one, they
> will always be seen as master or slave depending upon where they are
> connected on the  cable.

That's what they said back in the day. The've had more time to refine
PnP since then, but I read the following and realize that cheap is still
cheap.

> Of course there are a few (VERY FEW) drives that will have problems, but
> I think they are all also very old, and new drives will not have any
> problems being (single/master/slave) on the cable.
> 
> Please note that most new drive cables are UDMA and cable select as
> well. I can't remember when I last saw a new IDE cable that was not
> UDMA.

UDMA? (Wandering off to the usual search engine.) Erk.

<rant topical="no">
Well, pretty soon they'll be talking about separating the IDE disk drive
bus from the main interface bus, and then they'll be talking about LUNs
on IDE and ...

Sure. Re-inventing SCSI. When competition is one dimensional, it sure
produces a lot of history cycles.

Sure would be nice if some helpful company would sell a raw
ATA/IDE-to-SCSI one-drive controller/converter so I could just hang the
cheap, big drive on the SCSI bus. Firewire is an almost reasonable
alternative, but there are physical issues as well as speed -- just like
USB, external firewire devices can get bumped loose.

Yeah, I'm a dinosaur, what can I say?
</rant>

> I may also note that the only time I have ever had a problem with IDE
> drives on a cable select cable has been when the user (customers of
> mine) chose to jumper the drives as master/slave and they were using a
> cable select cable. They got intermittent errors due to the jumpering
> conflicts.  The permanent fix ---- rejumper the drive to CS.
> (Really easy to fix.)

So, if you have a CS cable and you have the master jumpered as master
and the slave jumpered as slave, but have the master on the slave
connector and the slave on the master connector, you should expect data
errors?

--
Joel Rees   <rees at ddcom.co.jp>
digitcom, inc.   株式会社デジコム
Kobe, Japan   +81-78-672-8800
** <http://www.ddcom.co.jp> **