mark@promark.UUCP (Mark J. DeFilippis) (02/19/90)
It has taken over 6 months, but the "SCO - TWO CONTROLLERS" problem has been
resolved and I think much of my findings can benefit people, hence this posting.
The following has been found to work together:
Machines tried:
SYSTEM CPU SPEED
ALR 386/2 16 mhz
ALR Flexcache 386/25 25 mhz
Hauppauge 386 16 mhz
Hauppauge 386 20 mhz
JDR Microdevices 386SX 16 mhz
SETUP:
Primary Controller Secondary Controller
master file 36 37
AT Bus interrupt 14 15
Controller combinations that worked fine:
M - Miniscribe m - MFM e - ESDI
S - Seagate r - RLL
Controller Drive 0 Drive 1
Primary WD-1006V-MM1 Mm-6085 Mm-6085
Secondary WD-1006V-MM1 Mm-6085||Sm-ST251 NONE
Primary WD-1006V-MM1 Mm-6085 Mm-6085
Secondary WD-1006-WAH Mm-6085||Sm-ST251 NONE
Primary WD-1006V-MM1 Mm-6085 Mm-6085
Secondary ACB-2372-C Sr-ST27R-1 Sr-ST277R-1
Primary WD-1006-WAH Mm-6085 Mm-6085
Secondary ACB-2372-C Sr-ST27R-1 None
Primary WD-1007 Me-3180E
Secondary ACB-2372-C Sr-ST27R-1 None
The original problem of drive Re-Cal was solved purely by trial and error.
It seems in ALL of the above machines and combinations if the 34 pin cable
on the secondary controller was a STRAIGHT cable with no twist (which in
that case drive 0 is DS1, drive 1 is DS2) there was ALWAYS a re-cal problem.
You could write to it, read from it, but in the middle of the read the drive
would Re-calibrate and seek at 1/2 step rate to cyl 0. This was unacceptable.
If a cable with a twist for drive 0 was used and all drives on secondary
controller set to DS2, the Re-cal problem went away. I don't have an AT
Technical ref. Manual (Someone borrowed it...), so I don't know what pins
these are, hence I could not analyze the reason.
It seems only a cable with a twist will yield good results. (Those of you
out there with a secondary controller, if you are running a cable with no
twist I will be very surprised, send me Email please as I am attempting
to gather as much data as possible. Many people wrote to me stating that
some combinations of controllers seem to work together and some don't.
Originally this seemed to be the case, but after noting the "twisted cable"
phenomena, I have been able to get the above to work together when I previously
could not. Some said they were trying to get two WD controllers to work
together for months with no success, One person wrote and said he had two
WD-1006's working together. It turned out later on that indeed he had
a 34-pin control cable with a twist. Check if your cable has a twist.
About the ACB-2372B. It seems that controller is no longer available.
The new version ACB-2372C does not have the Interrupt Jumper (J13 on the old
2372B)! (So much for "If it aint broke don't fix it", even if it is slow :-)).
However, The Gold fingers for D6 and D7 are still there. This is
not the case with the WD-1006 boards. 1006-VMM1 boards are a _REAL_ hassle
to modify for interrupt. The ACB-2372C also does not have pin J9 which
disables the floppy. When I called Adaptec they told me "There was absolutely
no room for the trace to allow the alternate interrupt line". The fact that
the pins to disable the floppy were left off was an oversight, but the
traces still run to where the jumper pins should be. So, to set up this
controller you should low level format the drive. Then change the controller.
Solder across pin J9 to disable the floppy if desired. Then solder
finger D6 and D7 together. This changes the interrupt from 14 to 15. You
do not have to use interrupt 15, but it is the most convenient since
the fingers are adjacent. I then used a razor blade to lift the
D7 gold finger right off the board once I confirmed it worked. Slap a jumper
on J2-1 to configure the secondary controller at the H170 port address.
Don't forget to change the NDISK kernel parameter and relink the kernel.
NDISK specifies the maximum number of Hard Drives. Then just execute
per the release notes:
mkdev HD 0 1
Again, thanks to all those who have provided their configuration information
and aided in my attempts to get two controllers to work. I hope that this
twisted cable phenomena will help others who may be having the same problems.
--
Mark J. DeFilippis
SA @ Department of Mathematics and Computer Science
Adelphi University, Garden City, NY 11530 (516) 663-1170
UUCP: philabs!sbcs!bnlux0!adelphi!markd