Unibone Wrong Slots (Woops) & Bus Termination Confusion

[Ashlin Inwood]

A little bit of background, I have an 11/40 in somewhat functional condition, it's a work in progress. The system is complete but I have no spare parts.

Anyway, today I finished my Unibone and promptly put it in the wrong slot and blew up the bus drivers. I am very much kicking myself about that. It was just the Unibone and bus terminators, so no serious harm was done but still not good.

I have been working on getting the Unibone back to a working state (Note that I have NEVER confirmed that it does work). I have had mixed results with the latch test depending on my terminator configuration.

I have access to a M9300 and a M930. I have the M9300 set for end of processor bus termination (maybe I'm not 100% sure). I have the  M930 at the start of the bus and  M9300 at the end of the bus.

When running the latch test I get this outcome:

Both M9300 and  M930 installed

Buslatch 0 will pass only a few hundred tests, sometimes a million.
Buslatch 1 will always fail on the first test.
Buslatchs 2, 3, 4, 5, and 6 are testing fine.
Buslatch 7 will always fail on the first test. (It's dead)

Only M930 installed
Buslatch 0 will pass
Buslatch 1 will pass
Buslatchs 2, 3, 4, 5, and 6 are testing fine.
Buslatch 7 will always fail on the first test. (It's dead)

Note that all drives test OK in a breadboard (quick and dirty check, chips may still be marginal).

So there are three problems I have to work out (possibly the same problem).
1. Why does the inclusion of the M9300 negatively affect Buslatchs 0 & 1
2. what is happening with Buslatch 7? (Are U33, U44, U35 and BBB damaged)
3. How do the 16 8641N drives map to 8 buslatch? (I assume in pairs, adjacent chips?)

 Also, what does this mean? It's always the same structure.
0) buslatch[0] = 0x1d (bits = 0x1f, R/W bits = 0x1f)
1) buslatch[1] = 0x20 (bits = 0x7f, R/W bits = 0x7f)
2) buslatch[2] = 0xca (bits = 0xff)
3) buslatch[3] = 0xb4 (bits = 0xff)
4) buslatch[4] = 0x27 (bits = 0x3f, R/W bits = 0x3f)
5) buslatch[5] = 0x93 (bits = 0xff)
6) buslatch[6] = 0x6c (bits = 0xff)
7) buslatch[7] = 0x04 (bits = 0x3f, R/W bits = 0x3f)


M9300 Setup

[Jay Jaeger]

The M9300 clearly has active circuitry, so I'm not surprised that it would confuse the basic Unibone tests like that.

From the webpage:  " One terminator is enough, but it must be passive: No Boot ROM or M9302. M9302 drives SACK active LOW, if any of the BG*IN or NPGIN is unconnected or HIGH, resulting in a loopback test error on latch 1."

So, that same exclusion would presumably apply to the M9300.  

My PDP-11/40 does not have an M9300 module, nor is it mentioned in my system manual.   A look at the PDP11 Bus Handbook on bitsavers indicates it is a Unibus B terminator - that name seems to imply it is for the B UNIBUS in an 11/45.

My PDP-11/40 has a M981 jumper/terminator (essentially a combination of an M930 and a M920 jumper) that connects the processor backplane to the first DD11, and then an M9312 at the end of the UNIBUS.

My PDP-11/45 probably does have an M9300 (and I have two in my spares boxes), but I don't have a card location chart for my 11/45 to I don't know its full bus configuration.  (Strange, that - I've had the system for more than 20 years, not sure why I never made a card location chart for it.  But I will be making one in the coming weeks, not that that would help you.)  I also don't seem to have any documentation on the M9300, but probably have a schematic.  But looking at the PDP-11/45 maintenance reference manual, it show Unibus B terminator in slot 28.

If anything, for a running configuration, I'd expect you would put a M981 jumper/processor terminator to connect the processor backplane to the first DD11 and then if you don't have an M9302 or M9312, the M9300 at the end of the bus with *all* of those jumpers removed - but even then it might confuse the Unibone tests.

For the Unibone acceptance tests I think I'd remove the M981 (if you are testing in the DD11 next to the processor) and the M9300 and just use a single M930 on the backplane you have the Unibone in to run the tests and also disconnect all the inter backplane UNIBUS jumpers and pull all the boards on that backplane.

The Unibone schematics will tell you what chips drive what, or you could just measure resistance to pins on the driver chips.  ;)

JRJ

[Jay Jaeger]

On this part I mis-spoke:  "the M9300 at the end of the bus with *all* of those jumpers removed - but even then it might confuse the Unibone tests."

I should have said "the M9300 at the end of the bus with ONLY the end of bus jumper removed" (if that makes it work like an M930)  But I am not sure even that makes sense, without looking at the schematic.

JRJ