PDA

View Full Version : PDP-11/23 adventure begins



matu
November 17th, 2009, 05:01 AM
I found the remains of a PDP-11/23 system in the attic... I was co-owner of a PDP-11/03 around 1985 and fell in love with its architecture, so I just had to drag this machine down and see if I could get it running again. Apparently it will take some work, but this seems like a pretty typical system so there are lots of online resources out there to help.

Pictures and notes can be found at http://www.physics.purdue.edu/~mjones/pdp-11/

The diagnostic LED's indicate that the system is alive and passes basic memory tests, but I need to acquire is something like a DLV11-J module so I can actually get a prompt.

I also expect that I'll have to learn where to acquire some 8 inch floppies for writing boot images...

cosam
November 17th, 2009, 05:40 AM
Not bad going for the "remains" of a system! Looks like the SLU is all you'd need to get going. I'm sure someone here or on the classiccmp list (http://www.classiccmp.org/lists.html) can set you up with one (I have a number of them, although I'm on the wrong side of the pond for you).

The current order of cards is a little puzzling. You'd normally have comms stuff (like the SLU, if it's been removed) near the top, right after the memory. You'll probably want to shuffle the rest down a slot when you get one.

Do you have any RL02 disk packs at all, maybe lurking in the attic? There could even still be a couple in the drives. They need the power on to open them, but you don't want them spinning up just yet as the head lock may still be in place. It would be safer just to undo the four screws on the top cover and take a peek.

matu
November 17th, 2009, 05:50 AM
There are indeed two RL02 disk packs in the drives - I took pictures but haven't posted them yet. The upper one is labeled with the name of current colleague of mine and the lower one is unlabeled. Both drives always have the fault light on, but I won't worry about that until I can initialize the controller.

pontus
November 17th, 2009, 07:37 AM
Nice setup. Makes me want to go working on mine.

Darshevo
November 17th, 2009, 07:46 AM
Can't wait to see your progress. I have one of these myself I have been waiting until the snow flies to work on.

-Lance

Crawford
November 17th, 2009, 11:59 AM
Matu,

You can also boot from a serial port (SLU) from a PC emulating a tape drive. That's what I did when I built my 11/53.

As far as disks, the M7555 boards are pretty cheap, and can drive MFM drives (which last forever), and 5 1/4 floppies that are still around. You could use this as a way to get started...

Good luck with the project! Let us know how it progresses .

- Crawford

Lou - N2MIY
November 17th, 2009, 03:36 PM
Very nice machine there. The DLV11 should be easy to find. They are on ebay quite often. I strongly suggest the DLV11-J M8043, it has 4 SLUs and was what was in your machine originally. I suggest one SLU for the console, one for LS: (serial line printer), and one for DD: (PC emulated TU58 ) eventually. Since you are familiar with Will's work, you've got to have seen his TU58 emulator.

Your machine has a BDV11 bootstrap terminator board. The BDV11 has bootstraps for DY: (your RX02) and DL: (your RL02s). If you want to add an RQDX3 M7555 (device DU: ) as Crawford suggests, you will need to upgrade the roms on the BDV11 to look like an 11/23+ M8189. I have done this and it does work. If you ever come to want to do this, let me know.

This machine built from parts is very similar to yours except that I have an RQDX3 instead of RLV11 (as mentioned above by Crawford): http://www.vintage-computer.com/vcforum/album.php?albumid=37 In the album are pictures of the disk drive chassis I built for 5.25" floppies and MFM hard disks.

Actually I have an RLV11 and RL01s, but they have problems. Long story.

I see the H19 terminal on top of the rack. Does it work?

As you can tell, we are all more than willing to help you get this going.

See, the good machine are all in attics in academia!!!!!! Go back to the attic and look for the pdp-12 that might be up there. If you find a TU56 you should let us know.

Lou

PS. Obviously I can make bootable RX02s for you if it comes to that.

matu
November 18th, 2009, 08:25 AM
I've ordered a DLV11-J... maybe I'll get it before the end of the week.

The H19 terminal does work, sort of. It seems that the vertical refresh timing is off, so the bottom row of text is in the middle of the screen, but scrolls off the top to re-emerge at the bottom. There is a second H19 terminal that has a different issue - it displays two copies of every line, as if the interlace signal is screwed up. I don't know if these are common issues with these terminals. If they have a well known fix I would be inclined to open one of them up. In any case, my preference would be to connect to the serial port on a Linux box.

No PDP-12's up there but instead I found http://www.physics.purdue.edu/~mjones/pdp-11/IMG_2907.JPG, but it's just not the same.

Regarding PDP-11's in academia, and especially in physics, they were extremely well suited for data acquisition in nuclear and high energy physics experiments. The experiment that *almost* discovered the charm quark in 1968 at Brookhaven National Lab used one of the few PDP-6's ever built (http://www.physics.purdue.edu/~mjones/pdp-11/pdp6.jpg). I have no idea what the big round radar screen was for.

I haven't given up on finding a tape drive... it may be around somewhere because there hundreds of 1/2 inch magnetic tape in boxes all over the building from experiments at the HRS experiment which ran at the Stanford Linear Accelerator Center in the early 1980's.

Lou - N2MIY
November 18th, 2009, 04:53 PM
Well, although it's not a pdp-12, it's still not bad. I hope you checked the attic thoroughly to make really sure there wasn't a TU56 up there.

Your H19's problem does not sound like the one that mine had. They are pretty easy to work on when you feel so inclined.

Yes, the 11s certainly were popular in academia. Three of mine are from schools. I was fortunate enough to be in grad school when they were being sent to the dumpster. I know that one of my machines was used with a CAMAC crate, also in some high energy physics experiment. Others had GPIB interfaces in them (which I really like to use with HP DAQs I have) and were from an EE department. Another had a set of nice Data Translation A-D and D-A cards that were high speed and 12 bit resolution.

Since being out of grad school, I just don't come across anything so complete anymore. If you like dec stuff, you should make it known among your colleagues - all kinds of goodies will come out of the woodwork.

The giant screen on the 6 was the point-plot display. I have seen photos of such a screen in action, but connected to a KA10.

Parting memory - I remember seeing a VAX 8800 at the PPPL surplus sale. Didn't have room in my apartment though.

Lou

tradde
November 18th, 2009, 05:05 PM
Well, although it's not a pdp-12, it's still not bad. I hope you checked the attic thoroughly to make really sure there wasn't a TU56 up there.

Your H19's problem does not sound like the one that mine had. They are pretty easy to work on when you feel so inclined.

Yes, the 11s certainly were popular in academia. Three of mine are from schools. I was fortunate enough to be in grad school when they were being sent to the dumpster. I know that one of my machines was used with a CAMAC crate, also in some high energy physics experiment. Others had GPIB interfaces in them (which I really like to use with HP DAQs I have) and were from an EE department. Another had a set of nice Data Translation A-D and D-A cards that were high speed and 12 bit resolution.

Since being out of grad school, I just don't come across anything so complete anymore. If you like dec stuff, you should make it known among your colleagues - all kinds of goodies will come out of the woodwork.

The giant screen on the 6 was the point-plot display. I have seen photos of such a screen in action, but connected to a KA10.

Parting memory - I remember seeing a VAX 8800 at the PPPL surplus sale. Didn't have room in my apartment though.

Lou

I can do you one better: I have used a point plot display on a KA-10. Wonderful device and lots of fun. The 11s are fun too. I have one myself. There is so much software for
them. And many devices are supported.

pontus
November 19th, 2009, 12:40 AM
No PDP-12's up there but instead I found http://www.physics.purdue.edu/~mjones/pdp-11/IMG_2907.JPG, but it's just not the same.



That is one neat looking IMSAI, it must be a custom made "frame" around the front panel. I think I can read the word "plotter" on there, so I guess a custom plotter interface is in the machine. The cromemco is pretty cool as well.

The IMSAI is a sought after system, guard it well :)

matu
November 20th, 2009, 11:22 AM
The Z19 console is working again...

http://www.physics.purdue.edu/~jones105/pdp-11/index.html#z19

matu
December 3rd, 2009, 08:17 AM
I just got into the locked part of the attic... indeed I found not one, but two tape drives! One is a DEC TU45 (http://www.columbia.edu/acis/history/tu45.html) and the other is something called an Aviv Corp GCR Multiport Tape System. Both are buried behind a lot of stuff, so it will be some time before I can move them. Besides... I still don't have my DLV11-J yet...

nige the hippy
December 3rd, 2009, 09:06 AM
There's an M7555 on ebay -
Item number: 270488261549

Lou - N2MIY
December 3rd, 2009, 03:03 PM
Very good on the TU45! There must have been some big iron around there then. The TU45 is a massbus tape drive. You'll need a unibus pdp-11 with an RH11. You should ask your colleagues about what happened to the machine that the TU45 was connected to.

Lou

matu
December 4th, 2009, 12:34 PM
With any luck I may have a DLV11-J module next week some time. However, I'm a little confused about where to put it. Recall that this is a PDP-11/23 with an 18-bit KDF11-A double width processor and not a PDP-11/23+ which would come with a quad width 22-bit processor board. The inside of the box has sticker that says "P/N: OBA11-N, CS REV 139-C, ECO 006" in messy handwriting (so maybe it's an M, not an N), but it is a little unclear what the part number for the 4x9 backplane is without taking it apart. Would it be probably be a H9273-A? What appears to be resistors are soldered into W2 and W3 but I can't read the color codes - could these be low impedance jumpers?

Anyway, the real question is which slot I can put the DLV11-J in... Is this a serpentine backplane and can I put the serial line unit in slot 4(CD) (ie, where slot 1(AB) is the KDF11-A) which would be beside, and possibly upstream, of the RXV21 module which currently sits in slot 4(AB)?

Otherwise I think I would have to move all the modules down a slot, which almost certainly wasn't the original configuration. The 'ba11nug1.pdf' file doesn't help much because it refers to 11/03-L, making no reference to the 11/23.

Any suggestions?

Lou - N2MIY
December 4th, 2009, 02:50 PM
Your machine has AB/CD on all slots. Your RLV11 uses the CD interconnect. Everything else goes in the AB slots (the left hand slot.) So, put your DLV11-J in the first open slot on the left. This is not a serpentine backplane.

Lou

matu
December 4th, 2009, 04:04 PM
Is the order an issue? Are there any pitfalls in having the priority of the console interrupt lower than the other devices?

tradde
December 4th, 2009, 04:57 PM
What is this serpentine backplane I keep hearing about? I never heard of it before.

matu
December 5th, 2009, 02:23 AM
See, for example, http://www.dunnington.u-net.com/public/PDP-11/QBus_chassis, which also suggests that I probably have a H9273-A straight backplane.

Lou - N2MIY
December 5th, 2009, 06:20 AM
Matt,

I would think that the least important device, from an interrupt priority standpoint would be the console. Among the most important would be disk controllers.

I usually put the memory closest to the processor, followed by disk controllers, followed by other stuff (A/D, D/A, HPIB) interfaces. I put the SLU interfaces at the end. Of course, in your backplane, the very last slot has to be the BDV11.

Yes, you have an H9273A. I too have the same BA11-N chassis.

Lou

pontus
December 6th, 2009, 02:15 AM
Your find got me inspired to work on my own PDP-11/23 which has been standing for a while untested and with missing cards. I too have the BA11N, KDF11 and MSV11. So, given the cards and how they are placed in the backplane, I think I can safely say that you have a H9273 backplane (There should be a sticker on top of the backplane), which is straight. So if you just keep the cards in the order they were in to begin with you should be fine. If you have problems starting it, you can remove cards to isolate the problem.

I have spare DLV11 cards and probably a cable too. I didn't say anything before since I've been unable to test them. But yesterday I got my 11 working, and now I can :)

So if your other source of DLV11 doesn't work out, maybe we can work something out. I'm generally not that interested in money and would rather make a trade for something interesting. Send me a PM

(notice that I'm in Sweden).

matu
December 12th, 2009, 09:55 AM
Okay - my DLV11-J was shipped by ox cart but eventually arrived. It had been configured
so that Ch3 was not the console, and had Ch0 at 9600 baud and Ch1..3 at 1200 baud.
But with my trusty wire wrap tool I restored it to the factory defaults, threw together a cable and got into ODT:
@1000/000000 112701
@1002/000000 52
@1004/000000 105737
@1006/000000 177564
@1010/000000 1774
@1012/000000 110137
@1014/000000 177566
@1016/000000
@1000G*
001020
@
Yep... it works!

pontus
December 12th, 2009, 03:58 PM
Okay - my DLV11-J was shipped by ox cart but eventually arrived. It had been configured
so that Ch3 was not the console, and had Ch0 at 9600 baud and Ch1..3 at 1200 baud.
But with my trusty wire wrap tool I restored it to the factory defaults, threw together a cable and got into ODT:
@1000/000000 112701
@1002/000000 52
@1004/000000 105737
@1006/000000 177564
@1010/000000 1774
@1012/000000 110137
@1014/000000 177566
@1016/000000
@1000G*
001020
@
Yep... it works!

Nice, hmm I should learn to use ODT properly, doesn't look that hard.

matu
December 14th, 2009, 05:57 AM
The KDF11/LSI11 processors have such a flexible instruction set and addressing modes that it isn't hard writing assembly code, which is easier to read than what gets programmed in ODT. The original source code is at http://www.physics.purdue.edu/~mjones/pdp-11/#DLV11.

matu
December 15th, 2009, 09:58 AM
I'm having trouble with the RX02 drives - I can initialize the controller but when try to issue a read or format command, I get back definitive error code 120: "A PREAMBLE COULD NOT BE FOUND". The RX01/RX02 pocket service guide suggests that this could be due to the diskette, R/W electronics the drive, the controller, the power supply or the interface module, which doesn't really narrow it down much. I did verify that both drives are spinning and the head engages when I issue commands. The left drive makes a rattling noise when I close the door but the right drive sounds fine. I think the controller is fine because I can fill and empty the buffer without any problems and all the status bits come back as expected. The +5, -5, +9.5 and +24 volt supplies are all okay.

Does anyone know what the most common problems with these drives might be? Presumably there are some optical components that should be easy to check... or is it true that RX02 drives need pre-formatted media from DEC? My spare 8" floppies are from IBM...

Lou - N2MIY
December 15th, 2009, 04:02 PM
The RX02 can format floppies. It was for the RX50 that DEC wanted you to buy their floppies (and so they told you an RX50 drive could not format floppies. However, this is not quite true, because a Rainbow can format RX50s.)

Under RT-11, use the FORMAT command.

Are you trying to read a known formatted floppy, or are you starting with a blank?

Have you cleaned the heads? They may be filthy. Also, make sure that the head load arm foam is still intact. You should also run the XXDP diagnostics on these drives to see what they have to say.

Lou

matu
December 15th, 2009, 04:35 PM
Well, I haven't actually booted RT11 - I was planning on entering a program in ODT to write an RT11 boot image to a floppy, but haven't got that far. Here's what I've got:


@1000g
RXV21 format.
0. Initial state.
RX2CS = 004000
RX2ES = 000000
1. Initialize RXV21.
RX2CS = 004040
RX2ES = 000004
3. Issue read error code.
RX2CS = 004040
RX2ES = 000000
BUFFER[] = 000000 000000 000400 006400
4. Issue format command.
RX2CS = 104440
RX2ES = 110000
5. Issue read error code.
RX2CS = 004040
RX2ES = 000000
BUFFER[] = 000120 000000 000400 177440
002372
@

The '120' is the authoratative error code for A PREAMBLE COULD NOT BE FOUND.
The disks I have were formatted 30 years ago, but I expected to be able to read something at least even without formatting them.

Lou - N2MIY
December 15th, 2009, 05:40 PM
If you think there could be something good on those floppies, don't reformat them. 30 years is no problem, they should still be readable.

You might have an easier time using Will's TU58.exe and a bootable RT-11 TU58 image. Once you have RT-11 booted off the emulated TU58, you can try to look at the floppies (and read them if they are RT-11 format), and then also make your own, new bootable RX02 system disks. The boot roms on your BDV-11 have the TU58 (device DD: ) bootstrap.

Did you have or find any bootable RL02 packs?

Lou

cosam
December 16th, 2009, 12:54 AM
Yes, I'd avoid wiping the floppies just yet if you can help it. The TU58 emulator is a great tool to get a minimal system running and can even be used to run XXDP diagnostics (I've done this on my '23+ (http://www.cosam.org/computers/dec/pdp11-23/20080322.html)).

For reading and writing disk images, you may want to look into VTServer (ftp://minnie.tuhs.org/pub/PDP-11/Vtserver/) which can be used to pump data over a serial line.

matu
December 16th, 2009, 12:39 PM
Well, here's another possibility... the RX02 is a single-sided drive and the disks I have are double-sided. That means that the index hole is in a slightly different location and indeed, I don't see any index pulses from the MC1414 comparator that is driven by the photosensor for the index hole. The photosensor works because I see the output when I insert or remove a disk.

matu
December 16th, 2009, 01:10 PM
I cut a new index hole in the right place for a single-sided disk. Now I get pulses from the index comparator and after I set the RX INIT bit in the CSR I now see that the DRV RDY bit is set in the status register (it wasn't before). So that's progress... I can format track 0 in single or double density, and read it back again in the appropriate density, but the format function still never gets past track 0...

matu
December 17th, 2009, 09:35 AM
I've learned more about RX02 drives than I ever knew I wanted to know and found out that there is a lot I took for granted many years ago. Indeed RX02 drives do not format their media. Although they could, the firmware in the M7744 controller board doesn't. Track 0 is special, but the other tracks need to have preformatted RX02 preamble data on the tracks in order to do anything useful, and the 'Set Media Density' command needs to find the preamble before it will write the density information to the header and zero the data on the sectors. These issues seem to be covered fairly well in http://www.classiccmp.org.

Anyway, I'm able to read any track I want on some bona fide RX01 formatted disks I borrowed. Making more RX02 disks from unformatted single sided media seems to be an adventure in itself, but at least it is well documented.

Lou - N2MIY
December 17th, 2009, 03:24 PM
The DSD440, which was a non-dec RX02 clone can format RX02s without being attached to a computer. It has a sophisticated controller with various on board diagnostics and includes a formatting function. I have two of these (one working, one waiting for me to get around to repairing it.)

If you are really in a bind, I can mail you a bootable RT-11 RX02.

I would like to hear more about your technique for making a new sector hole at the single sided location. I have quite a few brand new DS disks that I want to use as RX02s. I know where the SS sector hole is supposed to go, but I would want to make the new holes in some kind of clean way and without scratching the media. If I make two sets of new holes, I can make these into flippies.

I have (when desparate) opened the jacket of a hard sectored floppy, taken it out of the jacket, and covered all but one of the sector holes to make a soft sectored floppy. That was a royal pain. I have thought about taking the media out of these DS disks, then punching a new sector holes, then putting the media back.

Lou

matu
December 18th, 2009, 01:46 AM
Making the SS sector hole was an act of desparation, but here's what I did... I had a head cleaning disk that had both index holes, so I used that as a template, marking the new hole location with a Jiffy Marker. Then I stuck two pieces of kapton tape together and slid it from the hub hole under where the index hole would go so as not to damage the surface of the disk when I cut the holes on each side with a razor blade. I think this is actually quite safe - there is no data recorded on that part of the disk so as long as there is no mechanical damage and the disk still turns and doesn't have a nick that snags on the linder then everything should be fine. This could get tedious if you have a whole stack of disks but with practice I'm sure you could do 20-30 in an hour.

Here's my plan for making my own boot disk... I compiled SimH from http://simh.trailing-edge.com and can boot RT11 using one of the archived disk images. After making my way though the RT11 system generation manual I'll create a simulated RX02 boot image, and then copy the sectors of this to a real RX02 disk and try to boot with that.

matu
December 25th, 2009, 07:07 PM
The details are long, but here's basically what I did...
1. Write an assembler for the KDF-11AA... this was just a quick hack to produce data that can be written to memory using ODT. With C++ strings and standard template libraries this isn't that difficult (I call it "Hacro").
2. Write a program to format a floppy disk and download hex data sent over the console to a buffer, then write the buffer to the appropriate sector... http://www.physics.purdue.edu/~mjones/pdp-11/rx02_write.lst
3. Install the pdp11 simulator from SimH on a machine running Linux. Create an RX02 boot image from one of the RL images. This contained a rather limited set of files as it was only a proof of principle. You can verify that this image boots on an RX02 drive using the emulator.
4. Dump the disk image to the real floppy in the RX02 drive. This takes a while at 9600 baud.
5. Here we go:


@173000g
28
START? DY0

RT-11SJ V05.00
?KMON-F-File not found DK:STARTS.COM

.DIR

RT11SJ.SYS 73P 12-Mar-83 SWAP .SYS 26P 12-Mar-83
TT .SYS 2P 12-Mar-83 DY .SYS 4P 12-Mar-83
TECO .SAV 50P 12-Mar-83 DIR .SAV 19P 12-Mar-83
MACRO .SAV 60P 12-Mar-83 LINK .SAV 49P 12-Mar-83
TT .MAC 7P 12-Mar-83 DY .MAC 22P 12-Mar-83

10 Files, 312 Blocks
662 Free blocks

.R TECO
*IBASICALLY WORKS...$$
*B$$
*V$$
BASICALLY WORKS...*


So it more or less works... next I need to build a useful boot disk and see if I can figure out what's up with the RL02 drives.

jackrubin
December 25th, 2009, 07:22 PM
Whew, nice work! Must be pretty quiet in West Lafayette on winter break -