RE: CBM900 hard disk timeout

From: Martin Hoffmann-Vetter <>
Date: Fri, 22 Aug 2014 14:51:59 +0200
Message-ID: <003e01cfbe07$dd4fb270$>
Hello Michau,

> Controller/drive initialization failed

> But what is really interesting, and what I have completely overlooked
> the first time, is that when I select drive #0, the first error
> message I receive is actually not a Timeout, but a Read error (on
> block 0x00000002, whatever that means).

So you mean the message above is a "drive initialization failed" and not a "controller initialization failed". This is a good idea!

> Which leads me to this possible theory: maybe the drive is OK, and the
> controller is OK, but the drive is simply not formatted? Which
> explains the first read error.

If i remember correctly the data lines from controller and drives are differential lines with a terminiator. Did you checked this termininator on both sides?

> Then the computer tries do "something"
> with the drive, which results in timeouts

If the data lines are broken or the disk isn't formated, no marker can read. This can issue a timeout.

> So, now the big question is: how to format the drive. From what I read
> about MFM drive controllers, each of them can have a different
> low-level format, not interchangeable with other controllers.

Yes that's true.

> The
> controller here is WD1003-CMD, with some custom DMA channel
> interface. Would it use the same format as a PC-compatible ISA
> WD1003-something controller?

All hard disk controllers have tolerance, so they are not 100% interchangeable. In the 80, when a ship data with hard disks, i put the passed controller with the hard disk.


> W dniu 2014-08-09 20:15, Michal Pleban napisal(a):
>> Hello!
>> I am back home, and I tried to fix the broken C900. Three RAM chips
>> needed replacement, I exchaged them for new ones and the computer
>> again boots normally. 
>> I disassembled the hard drive and there were no fuses inside, but
>> after I rotated the drive by hand a few times and reseated all the
>> connectors it started to spin again. After I assaembled it back it
>> stopped spinning, so I repeated the process and it now spins
>> properly. It looks like some kind of mechanical problem.
>> Anyway, the computer is still not able to communicate with the drive.
>> All operations result in timeout. I tried these solution:
>> * Select all four "drive select" DIP switches.
>> * Use another data connector on the controller.
>> * Play with the undocumented first DIP switch.
>> * Play with the "Set drive type" option in BIOS; apparently the
>> option 2 is for 20MB disk.
>> * Use "Park drive 0" and "Park drive 1" commands from BIOS.
>> All I get is only a bunch of timeouts on every oepration (seek,
>> reset drive etc.) 
>> Here are the theories I came up with:
>> * The controller is dead. That doesn't seem to be the case, as
>> removing the controller gives additional error message
>> "Controller/drive initialization failure" which is not present when
>> the controller is inserted. So there must be at least some
>> communication between the computer and the controller. 
>> * The controller is partially dead. Maybe exchanging the ASIC chips
>> with another WD1003 would help in such case, but all WD1003 PC
>> controllers I found on eBay have the chips soldered. I wasn;t able
>> to find a WD1003-CMD anywhere for sale, only WD1003-WA? ISA cards.
>> * The drive is partially dead. I guess timeout issues would rather
>> indicate errors on the PCB and not the drive mechanics. On the other
>> hand, the drive is able to perform self-test and flash the LED, so at
>> least it must be partially working. Maybe some chips related to the
>> MFM interface are dead. Exchaning the PCB with another Miniscribe
>> 3425 mught help here. If MikeS' offer to do that still stands, I
>> would be glad to take it. 
>> * The computer is able only to communicate with select types of MFM
>> disks and the 3425 is not one of them. This seems unlikely, as the
>> WD1003 controller is pretty standard ans there should be no reason
>> for some special kinds of disks to generate timeouts with it.
>> Any other ideas would be welcome :-)
>> Regards,
>> Michau.
>>        Message was sent through the cbm-hackers mailing list
>        Message was sent through the cbm-hackers mailing list

       Message was sent through the cbm-hackers mailing list
Received on 2014-08-22 13:00:42

Archive generated by hypermail 2.2.0.