MicroVAX II - TK50 problem

Ethan Dicks ethan.dicks at gmail.com
Thu Mar 10 17:51:10 CST 2005


On Thu, 10 Mar 2005 14:18:59 -0700, willisjo at zianet.com
<willisjo at zianet.com> wrote:
> I did know about the serpentine weirdness. It caused me a great deal of
> grief in getting the system to pass its self tests in the beginning, before
> I stumbled (quite accidentally) across hamster's digital$resources
> page on Qbus. However, I think you may have hit on something. The RQDX3 is
> not at the end of the bus. If memory serves (I'll verify this
> after I get home from work today), the backplane arrangement is something
> like this:
>      A        B        C        D
> +-------------------------------------+
> |                CPU                  |
> +-------------------------------------+
> |                Memory               |
> +-------------------------------------+
> |     TQK50*       |                  |
> +-------------------------------------+
> |     M9047        |                  |
> +-------------------------------------+
> |     RQDX3*       |                  |
> +-------------------------------------+
> |                DHV11                |
> +-------------------------------------+
> |                ???? **              |
> +-------------------------------------+

I commented elsewhere, but if you literally have this arrangement (one
memory card), you aren't passing grant on the right side of slots 4
and 5.  If this "memory picture" is not entirely accurate, you might
still have a grant issue.

The RQDX3 does not have to be the last card on the bus (unlike the
RQDX1, which _does_), but it was customary to put it there, partially
due to the cable which comes up from the bottom of the card cage and
might cause interference issues with some other option cards' cables. 
Also, it has a command queue, etc., sufficiently long that it won't
starve at the end of the bus, so it's safest to put it there.

Normally, communications cards and tape cards are considered
time-critical because of the nature of tape moving past heads and bits
passing down the wire cause extra work if they have to be re-done. 
Disk interfaces of this era are sufficently complex that they don't
need to be at the front of the bus.  Note that this was more of an
issue in the days before the uVAX-II when peripherals had shorter
internal buffers and needed to be prioritized or they might suffer bus
starvation.

It was also common to put an M9047 grant card in the AB part of slot 3
as a place-holder for a memory option later.  That prevented having to
completely rejigger all the cards, which might be required if one had
a dual-height option (like a DEQNA or TQK50) in slot 3.  Given that
Qbus grant cards were in no way as common as Unibus grant cards, it
wasn't a universal practice, but it's still a good idea if one has a
spare grant card lying around.

Hope this clears up some of the voodoo of Qbus configuration.

-ethan



More information about the cctech mailing list