Unibus / Qbus bus drivers and receivers

From: Michael Davidson <michael_davidson_at_pacbell.net>
Date: Fri Jun 29 11:01:37 2001

Jerome Fine wrote:
>
> I have been following this thread since your first post yesterday.
> What you seem to have omitted is the target interface. Maybe
> I am missing something, but is your target a serial port like a
> DL(V)11 or a disk interface - possibly raw SCSI (or now likely
> MSCP may be allowed since I understand that the patents
> expire this year)?

No, you aren't missing anything ;-)

I didn't mention what the target interface was - it is, in fact, IDE
- and the thread on that topic (from some time last year, I think) is
how I found this mailing list when I did a search to see if anyone
had already done this.

Interestingly, I think that I had already been over a lot of the
ground that was covered in that thread before I discovered it.

My first idea was to use a DRV11-W which I already have and to build
a suitable interface board to go between it and the IDE bus. Once I
looked at this more closely I came to the conclusion that this was
probably more trouble than it was worth and that it might actually
be easier to just build a (non-DMA) Qbus to IDE interface from scratch.

Looking at this idea in more detail it became apparent that it would
need almost no additional interface logic beyond what was provided in
a standard set of DEC bus interface chips (DC003, DC004, DC005) or
their equivalent implemented in discrete bus transceivera and logic
cloned from a typical DEC Qbus interface.

All you need to do is to add 16 bits worth of tristate bus transceivers
to drive the IDE data bus, provide a couple of latches for the low order
address bits, and generate the IDE bus DIOR and DIOW read and write strobes
at the appropriate times. (Fortunately the Qbus data setup and data
hold times at the slave device are sufficiently generous that it looks
as if it should be very easy to generate DIOR and DIOW from BDIN and
BDOUT gated with the appropriate decoded addresses)

> From my point of view, the other side of the fence is the software
> device drivers that will be required. If the Unibus/Qbus module
> emulates an existing DEC board, then there will already be software
> available to use it.

The intention is that this will be an almost direct mapping of the
IDE interface, which will appear as 8 word addressable registers
corresponding to the IDE command register block.

Since I am mainly interested in running old UNIX systems the idea
is to add support for this new IDE interface to the Supnik emulator
running on Linux and then do the necessary driver development work
there under the emultor. Once that is done you just copy the emulator
disk image onto an IDE drive and move it across to the real hardware.
 
> In the past there has been some talk of a Qbus host adapter for
> both SCSI and IDE drivers which emulates MSCP. I don't know
> if Mentec acquired the rights to the MSCP patents, but if not
> then Compaq is not likely to care at this point and in any case
> if the patents for MSCP have expired, then there won't be a problem.

I'm not really interested in MSCP - I only ever worked on fairly
early low-end Qbus and Unibus systems all of which pre-dated MSCP
so I don't know much about it, but it seems like overkill for what
I want.
 
> By the way, if IDE drives are used, the idea was to place the drives
> right on the board since the 18" cable length limitation for a PC
> environment is likely to be exceeded in a Qbus enclosure.

Enclosure??? Who said anything about an enclosure? ;-)
(Well, actually one of my Qbus backplanes *is* in a BA11 box ...)

I agree that the 18" IDE cable length makes things tight, but I can't
see myself putting a drive on the board itself since I was hoping
to get this whole thing on a dual height card and I just don't see
there being enough room even for a PC card form factor drive.
(of course I suppose I could just use a Compact Flash memory card ...)

I should probably add that another motivation for doing this would
be to hook up an LS120 drive to give me some removeable storage
that's more convenient than my dual RX02's.

> Just some thoughts which may be worth the cost of my time to
> write this?

Thanks, I appreciate the information and advice.

Michael
Received on Fri Jun 29 2001 - 11:01:37 BST