On 8/30/2017 5:53 AM, Pete Turnbull via cctech wrote:
On 30/08/2017 05:29, Douglas Taylor via cctalk wrote:
I'll send along a picture of the rear of the back plane. I'm getting
the impression I can't do what I want with the old cpu cards, M7270
and M7264.
I had really hoped to be able to put together a simple system to
demonstrate the differences in processing power between the 11/2 cpu,
the 11/23 and the 11/73.
They are all dual width cards and it would have been simple to swap
them out. I think to do it I would need 2 boxes, one with a 16 bit
backplane and the other with a 22 bit backplane.
I don't see why you couldn't do what you want with the BA11-M and a
little work, *providing* the Emulex UC07 controller works in an
LSI-1103 system - and the manual (on Bitsavers) suggests it should.
Section 1.6.3 says "The UC07/08 is compatible with the Q-Bus used on
all LSI-11 ... series computers."
First, you'd need to undo any backplane upgrade that made it 22-bit
instead of 18-bit. BTW, there's no such thing as a 16-bit backplane,
only 18-bit and 22-bit. BDAL17/18 are always bussed, to allow for the
use of parity, even in 16-bit-CPU systems such as an 11/03.
The only reason you need to do this is that the KD11-H and KD11-F
processors put other signals on those lines, which the Emulex (and
other 22-bit devices) won't like and will interfere with.
The soldering you mentioned is almost certainly the extra four bus
lines for the upgrade. It will be on both the B and D fingers of the
backplane, because it's a serpentine backplane with Q-Bus on both
sides. Look for wired connections between BC1, BD1, BE1, BF1 and
between DC1, DD1, DE1, DF1. Check there no other extra connections;
sometimes people added connections for other signals - for example I
have a backplane with the SRUN signal on extra slots for diagnostics
and faultfinding. Also check you don't have an H9270-Q, which is
inherently 22-bit, instead of an H9270. I've never seen one, but
presumably they exist.
See http://www.dunnington.info/public/PDP-11/QBus_chassis for a little
more information.
Next you'd need some sort of bootstrap. What's in the custom EPROMs
on you MXV11-AC might do. Or might not, depending on whether it uses
any 11/23 (KDF-11) specific instructions or diagnostics, and includes
an MSCP bootstrap. The autoboot feature on the UC07 might do
instead. Or might not. You'd have to experiment.
If you do keep the MXV11-AC, you've already got 32KB of memory that
works with any of your 11/03, 11/23, or 11/73 processors, and you have
two DLV11-compatible serial ports. In fact the serial ports are
virtually identical to half of a DLV11-J. Since RT11 rarely has any
use for more than two, you probably don't need any more.
If you keep the MXV11-AC and re-enable the memory, you only want
another 32KB, and maybe not even that. I can't remember if RT11 5.3
will run in 32KB; it probably will, and I'm sure it would if suitably
SYSGENned. I do remember RT11 5.6 either didn't or didn't unless it
was seriously pared down. Don't use anything older than 5.3 because
there are bugs in the MSCP drivers that prevent it working with just
about anything other than RQDX1/2 interfaces.
Or you could probably use the MSV11-P. It works in 18-bit systems,
and should still work in a 16-bit (CPU) system, but obviously you'd
only be using the bottom 64KB. If you want "period" memory to match
the 11/03, you could find an MSV11-DC or -DD to use instead. The -DC
has 32KB to supplement your MXV11-AC; the -DD has 64KB. The -EC and
ED versions are the same boards but with parity circuitry added, which
makes them less common and more expensive, but they'd also do what you
want.
Hope this helps...
Yes, it does help. There are 3 issues that I am trying to resolve:
1. Running in 32kb of memory. If I use the 32kb MXV11 can I run RT11
V5.3? I tried this in SIMH and set the Cpu to 11/03 and memory to 32kb
and it did work. The MXV11 has PROM and is set to boot from it, but it
is not a device boot. This BA11-M was connected to a MicroVaxII and was
set up to answer telephones for the Univ of Wisc. I got this about 15
years ago and I think I looked what was coming across the console line
and I remember DL showing up or something like that. The MXV11-AC is
devilishly tough to setup, all those wire wrap jumpers and I've
misplaced my wire wrap tool.
2. Bootstrap. I transferred the RT11 V5.3 to a DEC 535MB SCSI disk and
was able to boot it using an Alphatronix SCSI controller, it is a Viking
QDO rebadged. It only can see 2 disks at a time, but auto configures on
startup, unlike the UC07. When I say is was able to boot it, I
connected it to a 11/53 CPU in a BA23 box just to test it out. The QDO
doesn't have a native bootstrap so that's why I began thinking about the
UC07. The manual says it has an auto-boot for LSI-11 only, but the
details were few. Someone else pointed out the modern bootstraps want to
address the PSW which doesn't exist for the 11/2, 11/03 cpus.
3. 18 Bit addressing. It appears that the H9270 backplane I have has
been modified by DEC with wire wrap and soldered in connections. I
really, really don't want to undo any of that. I may have to settle for
just running an 11/23, 11/53 and 11/73 cpu in this box.
Doug
Oh and also the usual issues: I haven't powered up this box in maybe 15
years, I haven't tested any of the CPU cards I want to use, and I'm not
sure what I'm really doing.....