PDA

View Full Version : Qbus Memory cards



Qbus
April 7th, 2011, 10:17 AM
I have seen many newer Qbus style memory boards listed on eBay and assume they are from the last production PDP and early VAX system. They are in the half, one and two Meg size. Will these newer Qbus style boards work in an old PDP-11/23? Also was looking at an M 8067 512 K full width Qbus card and was wondering if that would work with my 11/23? Why is it full width? Have been using two slot 16 K cards and would like to have more RAM.
Ray F.

RSX11M+
April 7th, 2011, 01:08 PM
I think it's mostly a matter of real-estate that they are Quad rather than Dual.
I don't think they require an ABCD slot. [I'd better check though - this if it matters to you]

As I recall, your original boards are non-parity type. The M8067 was only available with parity. [and includes a CSR]

Would it matter if they took the same memory chips as your bad ones?

Lou - N2MIY
April 7th, 2011, 02:51 PM
Ray,

Your 11/23 system has a Q18 backplane. If you want to add more than 256k, you will need to modify the backplane for 22 bit addressing. It can be done, just be aware you will need to do it to put in more memory.

Lou

RSX11M+
April 7th, 2011, 03:33 PM
Ray,

Your 11/23 system has a Q18 backplane. If you want to add more than 256k, you will need to modify the backplane for 22 bit addressing. It can be done, just be aware you will need to do it to put in more memory.

Lou
Lou, did we actually make this determination?

Qbus
April 7th, 2011, 05:03 PM
I have four 8044 DK cards, two working and two bad. Have not located any technical info on them beyond jumpers and switches but looking at them can see they are 32K 18 bit MOS memory to be exact they are M8045/5013128DP1 Was thinking if I can stuff a MSV11 M8067 256K card in the 11/23 system that will free up the other good cards I have for use on the second system I have that has no memory. System two is an old CNC system in a 2 X 12 Micro Backplane. That system came with a Chinese M8631 Ram- Rom card thatís a dead as a rock. Any technical info on the 8044 DK would be helpful but the real question is will a M8067 (MSV11) work in my four slot (ABCD) frame or not? The MSV11 would live in the expansion chassis along with a BDV11 with the CPU, communications, RL-01 controller and RX controller all living up in the main unit. Also note that the MSV11 I am looking at is 256K and not the fully populated 512 K so I am assuming the addressing would be ok, but thatís an assumption!
And as long as we are on the subject of memory anyone have a assembly program that will take a number from R1, write it to a memory location than read it back to R2 and compare them and if the same increment the address by one and do the same thing? If R1 and R2 are not the same halt. This way I can start at different points in the memory and confirm that they are working and if not at least have an idea what bank is bad. I am playing around trying to write something like this but would assume this has all been done years before.
Ray F.

RSX11M+
April 7th, 2011, 06:51 PM
Sorry Q... I didn't mean for Lou and my chatter to leave you hanging.

1) the schematic prints for your MSV11-D M8044 may be found here (http://bitsavers.informatik.uni-stuttgart.de/pdf/dec/qbus/MP00566_MSV11-D_Feb78.pdf) should you want to repair them.
2) the MSV11-P User Guide may be found here (http://bitsavers.informatik.uni-stuttgart.de/pdf/dec/qbus/EK-MSV0P-UG-001_MSV11P_Aug81.pdf)
3) the MSV11-PK [256KB] even though it is a Q22 board, may be used in an 18-bit system if it is the ONLY Memory.

Once again, I recommend it be placed in the same backplane as, and adjacent to, the CPU.

Oh... and the best thing to do with the backplanes to determine if they're Q22 or Q18 is test them with a VOM. I'll post a picture of a modified one for you to see.

Writing a program to do a memory test is a good exercise, but is there any reason you don't want to set the BDV to do the test for you?

UPDATE:


I just checked, and to be used in your ABCD backplane, there are two jumpers [W1 and W2] that MUST be removed from the MSV11-P or damage to other modules may [WILL] result.
Another Point... Are they M8044 or M8045 boards? I'm getting confused again. Anyway, they can't be used in the same system with the M8067 you're contemplating.

Did I answer all your questions?

RSX11M+
April 7th, 2011, 07:43 PM
Here is a document I prepared back in the day, and folded into my VT103 Manual.

It shows a VT103 backplane modification to Q22 [adding A18-A21], and the supporting identification of pins and slots.

Now your backplanes are not the exact same as this. This is ABAB while yours is ABCD. So this diagram can't be used to alter your backplane, however you can use it to identify which pins are AB and which are the extra address bits on those.

Simply test for continuity between those pins [marked in red] on adjacent slots in the AB columns.


Continuity=22Bit backplane
Open=18Bit backplane


Hope this helps more than it confuses. [I can see where it might]

http://96.11.235.30:9080/supplemental/VT103-Q22_modification.jpg

Lou - N2MIY
April 8th, 2011, 02:37 AM
Micronote 5 is my reference for matters related to Q22. I seem to remember that Ray had H9273 backplanes, which stock are Q18 (I have at least three). H9273 is not listed in Micronote 5. The Q22/CD for this chassis would be H9276.

Also, Micronote 5 points out the M8186 muct be etch rev C or higher. This can be spotted by the letter that is stamped into the back of the handle. I have rev A, C, and D M8186s and can attest to the fact that only C and D support Q22.

Lou

RSX11M+
April 8th, 2011, 05:02 AM
Lou's point is well taken here.

1) Adding address lines to make a Q18 backplane into a Q22 won't help you if your CPU does not support Q22.

2) I checked, and if you are going to buy the MSV11-PK [256KB] memory, it will work in your system whether or not it is in a Q22 Backplane, or has a Q22 capable CPU. [provided it is the ONLY memory and jumpers are properly set]

3) Not so with the 512KB version, which cannot be used in a Q18 system at all.

4) As I posted here in a previous anecdote, original documentation for the 11/23-A CPU [M8186] showed it as a Q18 board. I was shocked to discover that my 11/23 was recognizing and using the 4MB memory it was installed with - [B]ALL of it. Upon researching it, I realized DEC had quietly revised the artwork and ICs for the M8186 and that SOME [not all] were Q22 capable. Later on, DEC published this in a "Micronote" which Lou refers to.

5) Lou's method of backplane ID is fine. I came up with the method I use because some of the backplanes I was encountering were Non-DEC, and therefore were not covered by DEC documentation. Hence, testing if the [A18-A21] address lines are connected or open is the only definitive indicator I could come up with. Checking if these unknown backplanes were ABAB or ABCD configuration was another consideration.

6) Undertaking a backplane modification is not advisable if one does not know if their CPU and other modules support Q22.

Qbus
April 8th, 2011, 10:20 AM
Thanks for the information. Will have lots to read and think about this weekend. Am assuming that if I install the 256 K MSV11 it will only need the address range set and thatís it. Understand itís an 18 system and not a 22 but being under 256 k that won't be an issue, yes I will remove the other 8044 DK first and look at my W jumpers. Will be looking at the faulty 8044 cards also, it's just shear speculation but have a feeling the problem may be in the little charge pump used to develop -5 volts so may start by changing out all the capacitors in that circuit. Next problem will be getting Qbus extenders. About the memory diagnostic thing on the BDV11, thought that only will work for the first 28K and nothing beyond? And when I had bad memory cards in the system that did not tell me anything anyway? Is there a registers you have to use to see its results and can it be set to test all system memory????
Ray F.

RSX11M+
April 8th, 2011, 02:02 PM
Am assuming that if I install the 256 K MSV11 it will only need the address range set and thatís it. Understand itís an 18 system and not a 22 but being under 256 k that won't be an issue, yes I will remove the other 8044 DK first and look at my W jumpers.
Just to be sure I was clear - The W1 and W2 I'm talking needing to be removed are on the memory you're buying. If you don't pull them, it will likely do damage other devices using the CD interconnect on your system. [Please read the manual]


About the memory diagnostic thing on the BDV11, thought that only will work for the first 28K and nothing beyond?
Probably this was an error due to your bad memory(s).


And when I had bad memory cards in the system that did not tell me anything anyway? Is there a registers you have to use to see its results and can it be set to test all system memory????...
There are different levels of memory test. I'm not sure which boot roms you have, or if the "Test" is even enabled. It may have only been sizing memory in the mode it was in. This is a very limited non-destructive check to see if the memory is present.


I've pulled the only documents I could find that talk about BDV11 Diagnostic ROM settings from the Microcomputer Products Handbook 1985 and put them at the following links. [I couldn't find an online copy of that handbook anywhere]



Page 1 (http://96.11.235.30:9080/supplemental/BDV11/bdv11-1.pdf)
Page 2-3 (http://96.11.235.30:9080/supplemental/BDV11/bdv11-2.pdf)
Page 4-5 (http://96.11.235.30:9080/supplemental/BDV11/bdv11-3.pdf)
Page 6-7 (http://96.11.235.30:9080/supplemental/BDV11/bdv11-4.pdf)
Page 8 (http://96.11.235.30:9080/supplemental/BDV11/bdv11-5.pdf)


(Sorry I did it quick and dirty so it's one page per file. I need a PDF editor to combine them.)


Update: Here it is combined into a single TIF file (http://96.11.235.30:9080/supplemental/BDV11/bdv11.tif)

The links of particular interest to your question about Diagnostic ROM / BOOT test configurations for the BDV11 are the last two, but I reccommend you download them all to have. Supposedly the ROM formats are compatible with those on the 11/23+. If this is so, there will be upgrades out there [BOOT ROMS] that do some pretty comprehensive stuff at boot time, depending on how the switches are set.

Since I don't know which ROMS you have, this is the only info I have for you at the moment. Perhaps someone else out there will post info on their BOOT ROMs for BDV or 11/23. Anyway, see how it strikes you.

Incidentally, just for future reference - there is a micronote on how to modify the BDV11 to work on a Q22 bus, in case you ever want to step up.

Lou - N2MIY
April 8th, 2011, 05:12 PM
I have replaced the stock roms on my BDV11s with 11/23+ roms. Actually, I replaced them with the Micro-11/23 roms. Yes, it does wonders for troubleshooting! You can plug in a bad memory and it will tell you exactly what address is bad! I have used once to repair an M8044 (without using the logic analyzer!)

The trick to putting 11/23+ roms on a BDV11 is to split the two 64k roms into eight 16k roms. This has been described in at least two places in newsgroups. I merely followed the instructions.

Lou

RSX11M+
April 8th, 2011, 05:28 PM
Thanks Lou.

Could you collect the relevant notes and ROM contents and put em in a ZIP file?

I'd like to archive them.


Did you do the Q22 mod on your BDV board(s)?

Qbus
April 13th, 2011, 11:34 AM
I read the BDV11 book front to back and back to front and donít see anything about setting the size of the memory test, not much about the test at all other than it's turned on by switch 2. Have read somewhere that the test is limited to the first 28K only but have no idea where that was, maybe I imagined it? Perhaps newer ROMs have a better memory test? The bootstrap on my system only reports 28(K) and prompts START? And I have the switches set to respond with a "Y" by booting DL.
Ray F.

RSX11M+
April 13th, 2011, 04:09 PM
I read the BDV11 book front to back and back to front and don’t see anything about setting the size of the memory test, not much about the test at all other than it's turned on by switch 2. Have read somewhere that the test is limited to the first 28K only but have no idea where that was, maybe I imagined it? Perhaps newer ROMs have a better memory test? The bootstrap on my system only reports 28(K) and prompts START? And I have the switches set to respond with a "Y" by booting DL.
Ray F.
Did you verify that A2 is set to perform the memory test [maybe by changing it]?

What you need is a document I have not yet found, on the actual "Boot Code" on the BDV11.

If you remove one of your memory boards, what does it report then? Same or different?

---

What Lou and I have been chatting about is what it would take to get you another set of BOOT ROMS for the BDV. These would have the more "in depth" memory tests from the 11/23+ CPU.

Lou indicates he has done this modification to his BDV11s, and this jives with what others have told me is possible. [I probably have the same version of BDV11 roms you do.]

If Lou can help you locate the instructions and ROM contents, can you burn a new set of ROMs?

If not, I'll take a whack at it, and send them to you based on his instructions. Let's wait for him to check back in.

Lou - N2MIY
April 13th, 2011, 04:36 PM
Ok, then. I dug out my BDV11 notebook. Here is what I did:

First, the roms. My eprom programmer (trusty Needham's PB10) can allow you to load a big eprom image (I usually use .bin) into a buffer, then burn windows into a smaller eprom. So, I took the 64k images of the 23-380E4 and 23-381E4 micro pdp-11 roms (for M8189, KDF11-BG) and burned them into four 16k eproms (2716 or TI 2516), which I called 380-1 through -4 and 381-1 through -4.

The rom images can be found here: http://www.dunnington.u-net.com/public/DECROMs/

Then, the BDV11 needs to be configured to take the eight 2716s. In the Microcomputer Interfaces handbook, I chose what is called configuration F. The complete list of jumper and switches is:

Jumpers
W1 out
W2 in
W3 out
W4 in
W5 out
W6 out
W7 in
W8 out
W9 through W12 don't matter
W13 in

Switches
E21 B1 through B5 all off
E15 A1 through A4 all on
E15 A5, A6 off
E15 A7, A8 on

Then the roms are inserted as follows:

380-1 at E49
380-2 at E45
380-3 at E41
380-4 at E37
381-1 at E54
381-2 at E59
381-3 at E60
381-4 at E55

Helpful Reference Documents:

The Microcomputer Interfaces Handbook (section on BDV11)
Micronote 057 (points out location of jumper W13)
BDV11 print set (for locating jumpers, switches, and chip locations [E numbers])

Hope this helps,

Lou

RSX11M+
April 13th, 2011, 05:12 PM
Thanks for posting Lou. I've archived them and am looking now at the modification.

Just curious, is there a reason you chose '380 BG" and not "453 BJ":


23-453E4, 23-454E4 8189-BJ 2x8K KDF11-BJ (uPDP11/23) bootI was wondering if it was some kind of guidance issue.

Lou - N2MIY
April 14th, 2011, 03:45 PM
I first did this a long time ago. My first 11 was a micro 11/23. It had 380/381 roms. Then, I bought an 11/23 in BA11-N case from university surplus for $5. Before I knew of rom images on the internet, I upgraded the BDV11 in the $5 computer from copies of the roms in my original micro 11/23.

I never tried the 453/454 roms. The 380/381 seem pretty complete! Everything from DL and DX to MU and NQ bootstraps, nice interactive boot menu, nice diagnostics.

Lou

RSX11M+
April 14th, 2011, 03:59 PM
Thanks Lou...

There's nothing like a first-hand testimonial.


I'm looking at a jumper configuration that can use 4 x 2732a. Seems like it should be possible.

I want to test it out, but I can't seem to locate my BDV11 all of a sudden.
I found my MRV11 with [MXV11 boot roms], but that's 28 pin devices.

Anyway, thank you - Let's see what Ray says he needs.

Oh, a question for you Lou... does your BDV11 report K-Bytes, or K-words found on power-up?

Reason is - I wonder if Ray's system is recognizing his MMU? He's not real specific, but I keep getting the sense that he's getting 16-bit address values reported. This would account for the symptom he complains about where the boot never reports more than 28k, because it's in 16-bit mode and it's reporting 28K-Words, [64KB minus I/O space occupied by the BDV?]

Thoughts?

Qbus
April 15th, 2011, 11:36 AM
I have two 8186 cards, one with a FPU and both with MMU chips and get the same results with both so donít think it's not seeing the MMU. All my 8044 are 32K so doesn't have anything small enough to see reports under 28K. The bootstrap version allows you to boot from a RL (DL), RX02 (DY), RX01, TU and RK drive, would assume newer ROMs have more options? Yes if I turn switch two off there is no memory test and no 28K prompt just above the "START" option.
Still think the answer may be a little assembly program that will stop when it gets to a bad memory address than just look at R7 and you know where the problem is. Has anyone ever compiled the diagnostic paper tapes into a form that can be entered from the consul?
Ray F.

RSX11M+
April 15th, 2011, 12:36 PM
I have two 8186 cards, one with a FPU and both with MMU chips and get the same results with both so don’t think it's not seeing the MMU.

If the BDV11 boot roms weren't written to enable the MMU, there would be no difference between CPUs. There would also be no test of memory above 64KB [32KW].

32KW-I/O space = 28KW

This seems to be what your system is reporting. Is the same quantity of memory reported, regardless of whether you have 1 or 2 memories installed?

When the system cannot boot, what is the console prompt?

Does it go to ODT?

How much memory does RT-11 report when it boots?

RSX11M+
April 15th, 2011, 01:34 PM
...The bootstrap version allows you to boot from a RL (DL), RX02 (DY), RX01, TU and RK drive, would assume newer ROMs have more options? Yes if I turn switch two off there is no memory test and no 28K prompt just above the "START" option....

This does sound like an older boot rom set.

It would help me, if you could post a screenshot of the console startup sequence.

Whatever -

I think Lou is making the best suggestion for you - Upgrade the BDV11 to use 11/23 style boot. As he says, he's been able to repair memories using this alone.

Do you have the ability to burn 2716 or 2732 eproms?

Do you have 8 of either of these eproms?

Lou - N2MIY
April 15th, 2011, 05:59 PM
Well, with nothing better do this Friday evening, I dug out a BA11-N chassis and set up a BDV11 with micro 11/23 boot roms to show how it can help troubleshoot memory problems.

I took an MSV11-D M8044 that I had repaired before and pulled the replacement ram out of the socket: http://www.vintage-computer.com/vcforum/album.php?albumid=51&attachmentid=5595

Then, when powering up, the terminal shows this: http://www.vintage-computer.com/vcforum/album.php?albumid=51&attachmentid=5596

You can see that the "bad" rom holds the LSB at the bottom of memory on this board. The base address for this board is 400000. From the MSV11-D print set (MP00566) is it not hard to spot which chip holds the LSB at the start of memory. If a ram was in that socket with one bad bit, it would show up and the address and bit position would be displayed. From the prints, the offending chip can be identified.

The BDV11 is pictured here: http://www.vintage-computer.com/vcforum/album.php?albumid=51&attachmentid=5597

Lou

RSX11M+
April 15th, 2011, 08:13 PM
Great Info Lou, super pics. Thanks.

You have the makings of a good article there.

I guess we're in the same mood tonight. I just took a chance and powered my "spare" BA23 chassis for the first time since 1992? [pulled all the boards first]

A few minutes ago, I tested a minimal system of an 11/23 board, MSV11-LK 128K parity memory, DSD control and DLV11J, using my PC as a terminal. Really glad there was nothing to fix.

Now all I need to do is locate my darn BDV11. It's around here somewhere. As I search, I keep finding other "missing" items instead, so it's not a total loss. LOL

I'd really like to see if I can make a set of 4-2732 roms do the same as your 8-2716s. I'll publish how if I do.

Where the heck is that thing?&!...

I can see it in my mind's eye... I just can't place where. It's maddening.

RSX11M+
April 15th, 2011, 08:39 PM
UPDATE:


Darn. Guess I posted that "Fix" comment too soon...

It was sitting there - powered off but plugged in... Suddenly it let out a huge cloud of the "magic blue smoke", spat sputtered and then I unplugged it. It's fried. Oh well, another thing to do. :(
I guess finding the BDV is a little less important right now.