Jim McGrath jmcgrath at
Fri Mar 3 14:58:39 PST 1995

        Reply to:   RE>SCAM for IDE?

Cable Select (CSEL) is plug and play for IDE drives.  All Quantum drives
implement it, and have done so for years (I will not speak directly for
others, but believe they have done the same).  The cables are very simple
to implement and have no negative effect on non-csel systems.  HP has
inplemented this for years in their line of products - many other OEMs
have as well.

We might want to create a marketing pressure to have people do this - i.e. use
CSEL drives and cables, and call you system plug and play IDE.  Similarly
drives and cables sold to end users could use the same lable - as long as
everyone buys P&P devices and cables, things work fine.

I would be strongly opposed to any new sort of P&P for IDE.  It will take
years before we do not have to worry about older devices that do not implement
it (a price we had to pay in SCSI, but not in IDE).


Date: 3/3/95 6:29 AM
To: Jim McGrath
From: JEFFE at
                               - - Mail - - 
         March  2, 1995        7:27pm   MAIL IS -
   FROM: Jeff Epstein                    Viewed Ntfy Urgent    
     TO: SCSI Reflector                  
         ATA Reflector                   
SUBJECT: SCAM for IDE?                   
   COPY: Jeff Epstein

It has been said many times by some people in the IDE industry that,
"IDE just works!  Only one jumper ... master or slave."
I recall WinHEC-93 as the time this was the IDE gospel.

But here we are in 1995 and we see a plethora of different ways to set
jumpers on all the "IDE just works" drives.  Some examples are:

    "master without slave"
    "master with slave"
    "slave" with master
    "no jumpers at all means ?????"
    "disable cache"

Of course all this is very clear to everyone on the ATA WG committee, but not
so clear to anyone else.  Then to add insult to this terminology, the drive
may or may not have a label on it indicating where the jumpers are.
Or the physical orientation of the jumpers (if you can find them) never
matches the label.

Is IDE really Plug and Play?  I'll bet that anyone who has ever tried to
put a 2nd, 3rd or 4th IDE drive on their PC knows its definitely NOT Plug
and Play.  Then with the 3rd and 4th drive, you get to wrestle with the I/O
port assignments, "primary vs secondary".  Heaven forbid if you need a
3rd or 4th channel ..... and on and on and on.  Now you're up to a whopping 8

With SCSI, on the other hand, we invented SCAM, a method to have an automatic
SCSI ID assignment protocol.  It goes a very long way to eliminate ID
conflicts on a single bus.  After all, we're also dealing with lots more
devices in SCSI, not just two as in IDE.  It even works with legacy SCSI
devices ... as long as their own hard ID jumpers can be found and changed.
But thats no worse than it was before scam arrived.

Is it time for IDE to "scam" itself?  I think so.  The debates in the ATA
Working Group meetings on "Device-1 only" configurations etc indicated to me
that IDE drive configurations are not as automatic as you think.  Truth in
labelling on the drive would go a long way too.  Perhaps the Microsoft
Windows 95 logo (PnP compliance) requirements should include:

1.  IDE drives must clearly label the Master and Slave jumpers.
2.  IDE drives jumper implementations must all be the same:
    The ATAPI spec is a good start, but woefully inadequate.
    What good is "csel".  Who makes cables like this?
    Why do some drives work different with the "master" jumper installed
    than they do with no jumpers installed?

I'm sure you've all heard me belly-ache before:

SCSI is tired of being a 2nd class citizen!

But thats a different topic; I promise not to bore you further.

Jeff Epstein
Future Domain Corp.

More information about the T10 mailing list