PDP-11/45

I'd been looking for an older PDP-11 (with blinking lights and switches) for some time now with little success. Once in a while one would show up on e-bay, but when I did bid on one, it seemed like I was always outbid at the last minute (in one case I lost the bidding in the last 6 seconds!).

Not one to give up, I started looking for other sources for PDP-11's. After doing a search on Google, I found the classic computer mailing list (on Classic Computing ). There was a reference to someone who was selling a PDP-11/45. Figuring my luck couldn't be that good (it had to have already been sold after all the e-mail indicating that it was for sale was about a week old), I e-mailed the guy on the off chance that it was still for sale. To my amazement it was still available. What luck!

Then the reality set in. He wanted more than I really wanted to spend (all right *alot* more) for something I really didn't know what I was going to do with (other than I *had* to have it). After much thought and soul searching (aka rationalization) I went to the bank, got a money order and sent it overnight. Then the horror really set in. I had to have 2 19" racks totalling 1000 pounds shipped half way across the country. Eeek! Oh well, in for a penney, in for a pound (all puns intended).

With much anticipation the 11/45 was crated and shipped. It arrived on 12/11/02. Here it is with all of the packing material off but still strapped down to the pallet.

The '45 is pretty well configured:

128KW (256KB) of DRAM
FP unit
BA-11 expansion chassis
2 RL02 10MB drives
lots of DL11 async serial controllers
a couple of DR-11's of different flavors
an AVIV tape controller

The front panel needs a bit of cleaning up, but appears to be in pretty good shape (after replacing some broken switch covers).

The bad news was all of the cables had been cut. Including the BC-11A that connected the BA-11 to the 11/45. *sigh* On a positive note, I have been able to find many replacements on the web so they've been ordered and should arrive soon.

The other challenge was trying to figure out what type of power was needed by the 861-C power controller. All of the '45 documentation that I have indicated that there were two inputs:

120v 2 phase
220v single phase

I'd never heard of the first but I was planning for the worst since the 861-C indicated 120v 2 pole on its front. But upon further examination of the plug on the end of the cable and finding some more documentation, the 861-C requires *only* 120v at 24A (the plug is a 120v 30A plug). My plan is to not use the 861-C initially (until I make sure that the CPU works to some extent) and then figure out where and how to wire it up.

12/14/02 Update

I spent the afternoon getting the racks off of the pallet. They were held down with metal straps and blocked in place by 2x4's that were nailed to the pallet. It took a while, but the racks are finally free to be rolled around in the garage (its temporary home).

I've stripped off all of the cut cables and taken an inventory of the cards in the 11/45 and the BA-11. Here's the result:

Enclosure	Quantity	Card	Description
11/45	1	M787	Line Frequency Clock
	1	M8114	11/45 fraction high order data path module
	1	M8115	11/45 fraction low order data path module
	1	M8112	11/45 floating point ROM control module
	1	M8113	11/45 floating point exponent data paths module
	1	M8100	11/45 data path module
	1	M8101	11/45 general register address module
	1	M8102	11/45 instruction register control module
	1	M8103	11/45 ROM address control module
	1	M8104	11/45 processor data an unibus registers module
	1	M8105	11/45 timing and misc control module
	1	M8106	11/45 unibus control module
	1	M8108	11/45 segmentation status register module (KT11-C)
	1	M8107	11/45 segmentation address paths module
	1	M8109	11/45 timing generator module
	2	M7800	DL-11 Async interface
	1	M9200	unibus jumper
	1	M9202	unibus connector
	1	M792-YD	Bootstrap loader
	2	??	64KW Unibus MOS memory card
BA-11	5	M7800	DL-11 Async interface
	1	M7860	DR11-C General Device Interface
	1	M7762	RL11 RL01/02 disk drive controller
	1	??	Tape Controller
	1	M8716	DR11-W General Purpose DMA parallel interface
	1	??	DMA parallel interface

I plan on not using the BA-11. All of the cards I need can be put into the 11/45 CPU cabinet (the only card I'll be moving over from the BA-11 is the M7762 disk controller). However, since the 11/45 has only one system unit (4 unibus slots), I'll be taking the double width system unit (9 unibus slots) out of the BA-11 and installing it in the 11/45 eventually. Here's a picture of the 11/45 internals now.

12/28/02 Update

Some friends came over and helped with relocating the 11/45 and RL02 drives in the H960 cabinet. This involved lowering the 11/45 in the H960 by 10.5". Prior to doing that we had to move the H742 power supplies lower. This gave me enough room in the H960 to put in the second RL02 drive at a reasonable height.

Still haven't applied power yet. Maybe during the coming week.

1/1/03 Update

Well I finally applied power. I discovered that both H742's need to be powered up. Once I figured that out, I tested all of the lamps and replaced the ones that were burned out. Right now it doesn't do very much. I'm not sure what the problem is, because the console doesn't do anything. Some of the lights light, but I can't do a load address. Moving positions of the two knobs on the right of the panel does change which lights are lit, but that's about the most I can get out of it right now.

1/2/03 Update

RTFM. I hadn't realized that rotary switches on the front panel were more than two position. Once I had them set properly (like CONSPHY) the 11/45 seemed to be much happier. Previously I was getting ADDRERR and PAUSE illuminated on the front panel. Now, I can toggle in a program and it will execute. It's still early, but this is great progress. I'll see what else I can do later today.

1/5/03 Update

Argh! It's acting weird again. I don't think it was entirely the switch settings. Sometimes it seems to work and others it looks like it's trying to run a program (even with the halt switch down). Very strange. Resetting it seems to help sometimes.

1/17/03 Update

After several days of head scratching I finally decided to get back to basics and check the various supply voltages (why I didn't do this before is anybody's guess). The H7441 in slot B of the top H742 was running at 4.4v. Hmm, could this be it? I adjusted it up to 5.05v and the 11/45 seems to be much happier. It's running programs much better now. It seems rock solid. I'll check it out again in a day or so and make sure all is happy.

Right now there's only a single system unit in the 11/45 chassis. I plan on putting in a dual system unit to allow me to put some I/O in the 11/45 chassis. The 2 memory cards take up all of the available hex slots in the single system unit, so I can't put in any interesting controllers until I have more slots. The problem is that since the 11/45 is s/n 242 (very early) it has an "old" style power distribution board. None of the DD11's I have plug directly into that distribution board (DEC fixed this on systems with s/n > 2000). So I have to go and wire up a harness to plug into the power distribution board that the DD-11 can plug into.

1/18/03 Update

I took the M7800 SLC out of the 11/45 chassis and made sure of its settings. The XMIT and RCVR were set at something other than 9600, so I changed those settings. I entered the little program that I'd used on the 11/10 (with the SLC's addresses changed) and saw that I was getting output to the terminal.

Now it was time to see if it could execute someone else's code. I removed the M930 terminator that was at the far end of the UNIBUS and replaced it with an M9301 bootstrap/terminator. After some messing about trying to determine what address I should get it to run at (773000 is the right answer) I got it to print out 4 sets of 6 octal numbers and then a '$' on the next line. Hmm. I was sort of expecting an '@'.

Figuring that there had to be something wrong with the M9301, I tried an M9312 (yes I set all of the W jumpers properly). That proved to be pretty pointless. Nothing I would do would get me to a prompt of any sort. I tried 765144 (which is what the reference manual said). I tried 773000. Nothing useful. Oh well. Back to the M9301.

After looking over the M9312 documentation, I decided that it was worth a try to see if the '$' prompt was in fact the prompt for the console emulator. It turns out that it was and I could use L, D, E and S to set the load address, deposit values into memory, examine them and start a program running. Wow! This is starting to act like a real computer!

1/20/03 Update

I just took some pictures of the trouble some power supplies. Note that in the bottom power supply there is only one regulator. That is to supply -15v to system units 2 and 3. The other regulators would be for the bipolar or MOS memory which isn't present in my 11/45.

1/27/03 Update

I built the wiring harnesses necessary to hook a DD-11DK (9 slot backplane) into the 11/45 processor cabinet. I had to scavange bits from another harness to get some of the connectors. I had to replace the mate-n-lock connectors on the DD-11 with another set of Molex connectors since I couldn't find the mates for the connectors that were already there.

It was a tight fit but everything is hooked up with power and the backplane is now at the end of the 11/45's UNIBUS. So I now have plenty of slots to add in peripherals. At this point I'm not stressing the H744 +5v 25A regulator that supplies the system units. But if I start needing more, I can swap the H744 for an H7441 which provides 32A.

Here are a couple of shots of the 11/45 now with the DD-11 DK fit in. Notice on the wirewrap side the jumble of power cables by the DD-11 DK. That's why DEC repositioned the power distribution board to the top next to the fans in systems with s/n > 2000.

2/13/03 Update

I'd been putting this off, but finally over the past few days I put an RL11 controller into the 11/45. All of the cable were hooked up and the bulkhead connector was attached to the H960 cabinet. Things started looking good. Both drives powered up and neither had their fault light on. I again tried the M9312 without any success. I really wanted it to work because I have RL01/RL02 bootstrap ROMs for the M9312. The only other bootstrap board I have is the M9301 and it does not support the RL drives.

Not having any luck I decided to go for broke and toggled in the bootstrap for VTserver (it's a really cool program). It downloaded the standalone copy program. For the source I entered vt(0,0,0) and for the destination I entered rl(0,0,1) which would theoretically load BSD 2.9 onto the RL02 pack.

Well it remained theoretical. I quickly got an error. Hmm. Seems like there's a problem with it reading/writing the drive. Hmm. Looking at the value for the status code it seems like it can't find headers from the drive. OK. I haven't ever run two drives together. Maybe it's a problem with the cabling so I remove drive 1 and retry. OK. This time the error is different. This time the error seems to be that the drive is not sensing write current from the head. This is bad. Damn. Down to one "good" drive.

As much as I want to continue, it's getting late so I'll tackle this tomorrow.

2/14/03 Update

I recabled the drives (and swapped the drive IDs) so that the drive I "think" works is the only one cabled up. I again toggle in the program to download the standalone copy program for VTserver. All goes well. Now for the moment of truth. I enter vt(0,0,0) and rl(0,0,1) as before. Cool! It's now downloading an image over the serial port (from my laptop) and writing it to the RL02 drive. Then I realize this is going to take awhile. So I go back to my home office and get the laptop's power brick. To load the 8.7MB image of BSD 2.9 and write it to the RL02 drive took about 3.5 hours with the serial port running at 9600bps (I can't wait to see my power bill). What an accomplishment. I still don't have a bootstrap for the 11/45 but hey, I have BSD on an RL02 pack!

After sending some frantic e-mails, I came up with an RL bootstrap here . I'm replicating it here as you can't have too many copies.

I was now ready to have my 11/45 run an OS. Back to the "machine room". I powered it up and was half way through toggling in the bootstrap when the 11/45 got weird. Uh-oh. This time it looked really bad. Fiddled around with it for a bit when that sinking feeling came back that I'd seen this before, except this was worse...much worse. So I pulled the 11/45 forward out of the rack and started probing voltages. +5v on the A regulator looked OK. I was having a really hard time getting a reading from the +5v on the B regulator. Hmm. It's not that hard to get the probes to make a connection on the backplane. Uh-oh. There is no voltage comming from the B regulator. This is the same regulator that was reading 4.4v. So I grab another H7441 regulator and I swap it out. Then I notice the burns on the PCB of the old regulator. It got cooked. I don't think it liked being powered on for almost 4 hours. With the new regulator, the 11/45 was happy once more.

I decided to try something a bit simple to start so I put an RT-11 pack into the drive. I toggled in the bootstrap and hit start. On the terminal I saw RT-11 booting up! Yes! It's running an OS. Well now it's really late so it's time to give all of this stuff a rest. I'll see if I can't figure out how to boot BSD tomorrow.

2/20/03 Update

I still haven't booted BSD on it yet but on a hunch I took the M792-YD (diode ROM bootstrap loader) and checked its contents. It is 64 bytes (32 words by 16 bits) of diodes. By looking at the board and writing down the bit patterns as to what diodes were present (1) and those that weren't present (0) I could read out the contents of the ROM (don't even have to apply power). The contents are here.

There was also a notation written on the board (in permanent marker) 772700 which would seem to indicate its address. I wasn't in a mood to do any serious debugging of an RL02 drive, so I put the M792 into the 11/45's chassis. Since there were a couple of words that looked like it might have something to do with an RL02 bootstrap (the default address of the RL11 (M7762) is 177400), I decided that I'd try it out. I'd been getting real tired of toggling in the RL02 bootstrap to boot up the system. I entered 772700 into the console switch register, hit load address and then start. What do you know! It booted from the RL02 drive!

I left the M9301 at the end of the UNIBUS with its ROMs enabled. There isn't an address conflict and if I ever want to use the console emulator, all I have to do is load 773000 into the console switch register.

2/21/03 Update

I finally got the second RL02 drive working. I tried replacing the r/w head pre-amp with no success. I finally took the main logic board from the drive that had had the head crash and put it in the drive that was cabled up to the 11/45. Success! I now have two fully functional RL02 drives.

2/22/03 Update

Today was very busy. A friend came over and we "racked" a bunch of my systems. But on the 11/45 front I booted BSD 2.9 and Unix v7. Although both booted, neither was a roaring success. I realized that BSD needed 2 RL02 packs to contain everything to run and I only had the root pack. It booted ok, but I wasn't able to do very much since much (most?) of the system was missing. It did however put on a pretty spectacular display on the front panel when it was doing something. This is cool!

Unix v7 I was only able to run in single user mode. It was a copy of a pack that someone else had. It did boot, however I could not get it to recognize that the terminal could generate lower-case. I may go back and use vtserver to load up a pack with a known v7 distribution on it and try that one. Like BSD, it does produce an impressive display on the front panel when ever it did anything.

6/14/03 Update

I had been trying off and on for several months now to get the TS-11 9-track tape drive to work. It was very discoraging because I couldn't even get the drive to do something simple like loading a tape. I really wanted to have a working 9-track tape drive since that would allow me to create tapes on a PC (using a SCSI 9-track tape drive) that I could then read on the '11s. This would make creating boot images and such so much easier.

I was offered a TU80 and controller that were supposedly working, so I jumped at the opportunity. My only disappointment is that the TU80 is a "top loader" so it doesn't look like a proper tape drive. But functionality first!

I'd had the TU80 for a couple of weeks now, but hadn't had the opportunity to hook it up and test it out until now. I tried using RT-11. I was a bit disappointed that there didn't seem to be any support for a TU80. After some searching on the web (Google is your friend). I discovered that the TU80 and controller "looked" like a TS-11. OK, I know RT-11 supports that.

I loaded a tape into the drive and hit the load button. It loaded the tape successfully and then I put the drive on-line. Then I tried something simple:

copy *.sys ms0:

Nothing happend. Hmm. There are two ribbon cables that go from the drive to the controller. Since there are no distinguishing markings on either cable or the connectors on the controller, I tried switching them. After repowering everything back up, rebooting RT-11 and re-threading the tape, I tried the command again.

Hey! This time it was doing something. The tape was spinning. It was moving forwards and backwards. Cool! Over to look at the console terminal. It was writing one file at a time, but for each file it had to rewind the tape and start from the beginning. Uh oh! This was going to take a while. But the drive continued to spin tape forwards and backwards. After a while (OK, I admit it, I got bored watching the tape after a while and went to do something else for about half an hour) it completed. I tried:

dir ms0:

The tape started spinning and as it came across each file on the tape, it would print information about it on the terminal. Wow! I now had a working 9-track tape drive on a PDP-11!