• Please review our updated Terms and Rules here

Time to repair the *second* pdp8/L

C

czunit

Veteran Member
Joined
Aug 7, 2015
Messages
530
Ok, so with the first pdp8/L of mine "working" I decided it's time to start working on the second one. This one was a bit more of a basket case, I bought both of them together years ago and I think this was the one that was the spares pile for the first one. So it's going to have a lot of.... opportunities.

Initial state is that the system powers up, voltages check out properly, many of the bulbs are blown, and the unit does very weird things. Like for example it would load an address, and examine would work (ish, it would at least increment MA) but dep would not even do that. After a few minutes of running even trying to load an address with the LA switch would set the computer "running" with the lights dim and FETCH/EXECUTE lights lit.

It's got... problems. So since it ain't gonna fix itself, time to start working. On the positive side I do have a somewhat working first 8/L (see the other thread) and I have verified that key components (the M360, M700, and the like) are at least working.
 
First step was to pull the display and replace all the blown bulbs. It was a mess, someone else did a bulb replace long ago and left a lot of flux on pins and broke at least one trace. But that's what isopropyl alcohol is for, and testing each bulb with my meter showed the bad ones pretty quickly. A good bulb reads about 21-24 ohms resistance and will make the continuity tester *beep*. A bad bulb will either read ~100 ohms or 400 ohms (just the transistor I am guessing).

Replacing them consists of using desoldering braid plus flux to remove as much solder as possible, a tweezers to lever the bulb leads vertical (as DEC people bent them down before soldering) and gently pulling on the bulb with air heat at about 290c. Not too much, you don't want to burn the board. Once the old bulb is out clean everything up with Q tips+isopropyl, make the holes clear with the braid, then put in and solder the new bulb. I like to solder one lead a bit, then hold the bulb and heat the connection again to get it to stand up perfectly straight, then do the other lead. Less bending for the next person who works on this in 30 years (and if you're here reading this, HI!)

With the bulbs all replaced I can see that the computer is bats. As I mentioned even hitting load address sets the computer to a run state, so it's going to be fun figuring this out. First step is going to be check the basic gates as those 7440's and 7474's tend to be bad....
 
Second step is to check some of the foundational modules. Given that the system has a state which could be described as "bananas" I figured I would start by checking the 7440 based M617's and the 7447 (flip flop) based M216's. I'm sure the eternal "run" state when you click a switch (thus starting a state) is due to a bad flip flop, but there are probably a bunch of other problems.

To do this I bought a nice little TES200 chip tester and a 14 pin chip clip. It's a simple board, and with the chip clip and a dip header I'm able to test simple circuits on the cards.

And sure enough, things are a mess. Two 617's and two 216's are "good" meaning all chips pass tests. Three of the 617's and two other 216's are "bad". One of the 216's has *two* bad flip flop chips.

This was a spares box. Oh yes, it was.

The 617's are 4 input NAND gates, and I think most of them are used as big buffer or inverter drivers. Thus I can see why they fail: If they are powering a bunch of other circuit loads they can just fail over time in ways that smaller 2 input NAND gates don't. Flip flops just seem to flop, not sure why.

I may test the 115's, 113's, and 111's next. Looks like I also missed a 216 and another 617, will test those tomorrow. The other big board to check is the M160, not sure if I can test that with the tester.

And while this can't test more complex boards, the backbone of the pdp8/I and L are these relatively simple boards which are then wired up using the backplane. But it's a start, and it should lessen the need to try and trace problems throughout the unit.
 
So this morning I fixed all the bad 617's and 216's, put the cards in and fired up the computer. It still sets the "RUN" light when you do a load address, but it doesn't loop and the address loads into MA. So there's still something stuck, I think before I start logic probing I will test the 115, 111, 113's using the tester.
 
The M11's are also a basket case. Yep, this was a "parts" system no doubt. Well, swap chips we must....
 
I realize that this is not a pdp8/L.

It's a diorama of a pdp8/L. One made by field service to show just how crazy things could be.

The failure on the M11 in D2 might help explain why RUN is asserted even during a load address or Examine function. But it does look like some of the 111's have problems, so time to order a few 7400 NAND gates. I wonder if it's worth putting the boards on my pre-heater and just pulling all the chips and replacing them en masse. Probably not as it is normally one chip per board that is bad, which points to them stopping other 8/L's and being swapped into this system. But still, I can't wait to see what the G series logic is like.

For boards I can't fix I do have a complete set of pdp8/I and MM8/I boards from the long lost pdp8 that got hit by a forklift. But before I break those out I do want to try to fix as many of these as possible.
 
Wow. For the M111 and M113's every board but one (a M113) had bad 7400's. Literally, every single board had at least one, two, or all four chips bad. Sometimes the whole chip was bad, sometimes just a single gate. But all shot, I think either this batch of DEC's 7400's were crud or something very bad happened....

Oddly enough the M115's I pulled (3 input NAND) were all good. So I got that going for me......

Time to place another parts order and put this aside for a bit.
 
czunit said:
I realize that this is not a pdp8/L.

It's a diorama of a pdp8/L. One made by field service to show just how crazy things could be.
Click to expand...
I don't know what this is in reference to.

I have an 8/i in a closet that (last time I worked on it) had problems similar to your 8/L. Some year I should get around to running all the modules through the flip-chip tester. (I suspect the thing was over-voltaged or something, to have so much wrong with it.)

Vince
 
Interesting. I went into the attic and got the *BIG* box of M series modules. Oddly enough in a massive box of *stuff* I found no M111's and one M113. I do see some M133's and a lot of boards that are "NOR" logic, I'll take a picture of this.

Note: I just looked up the schematic on the M133. It looks to be identical to the 113 including pinouts, was it just a better version or is there some secret incompatibility?

Did the 8/I's use a different basic set of boards from the 8/L? Is there an inventory of which boards went into which type of computer out there?
 
jackrubin said:
Were you able to set up your TES200 with flying leads so you can test chips in circuit?
Click to expand...
Yes. Using heat shrink and good soldering I got it up and going. Some of the boards have.... oxidized pins on them so I have to clean them up but it works well.
 
vrs42 said:
I don't know what this is in reference to.

I have an 8/i in a closet that (last time I worked on it) had problems similar to your 8/L. Some year I should get around to running all the modules through the flip-chip tester. (I suspect the thing was over-voltaged or something, to have so much wrong with it.)

Vince
Click to expand...
It's in reference to the complete mess this 8/L is. The first one had perhaps half a dozen bad boards which was pretty easily fixable. This one has over 10 bad boards and that's just in the M111/113/216 logic. Either something very bad happened or it was a parts unit for a number of other 8/L's in the lab.
 
czunit said:
Interesting. I went into the attic and got the *BIG* box of M series modules. Oddly enough in a massive box of *stuff* I found no M111's and one M113. I do see some M133's and a lot of boards that are "NOR" logic, I'll take a picture of this.

Note: I just looked up the schematic on the M133. It looks to be identical to the 113 including pinouts, was it just a better version or is there some secret incompatibility?

Did the 8/I's use a different basic set of boards from the 8/L? Is there an inventory of which boards went into which type of computer out there?
Click to expand...
The printsets have Module Utilization pages which list the modules and show their placement on the backplane. I have an Excel spreadsheet showing the flip chips used in the 8/I and options but I never put one together for the 8/L.

www.dropbox.com

8-I_ModuleUsage.xlsx

Shared with Dropbox
www.dropbox.com www.dropbox.com
 
Well hm. I seem to be low on 113's and 111's don't exist. I think I'm going to have to send a letter to myself from 30 years ago because I seem to have a bunch of boards here. Maybe this wasn't from the 8/I, I wonder...

Here's a pair of pictures of the green and red boxes....


XGFuddP.jpg
srhWMq7.jpg
 
Well, this is a weird pile of stuff. Any idea what it came from?

QtyBoardDescription
7​
m1376-4-1 input NAND gate
2​
m101Bus data interface
2​
m1272-2-2-3 And/NOR gate
7​
m112nor gate
13​
m129and/or invert gate
24​
m13310-2 input nand gate
23​
m1353 input nand gate
24​
m1393-8 input nand gate
5​
m15524 pin dip mount
14​
m216flip flop
4​
m24516 pin IC card
5​
m238unknown
25​
m611high speed power inverter
7​
m1904 bit arathemetic module
2​
m6176 4 input nand
4​
m6276 4 input nand
5​
m623Not a clue lot o transistors
4​
m310timing delays
1​
m360variable delay
1​
m401variable clock
1​
m405crystal clock
1​
m939lot o diodes
2​
m00215 loads
1​
m101pos data interface
1​
m7718keyboard interface
1​
m77198/I interface
7​
m1788*6 data mixer
1​
m734io input bus ux
1​
m707tty xmiter
1​
m704no clue
1​
m107no clue
 
Yeah, I'm wondering about this: Putting on a 30 year old wayback helmet I remember salvaging the following from the old Psych lab:

pdp8/I: Running. Had a DW08, 2 cores, high speed papertape/punch.
pdp8/I: Rammed by a forklift tine destroying the wire wrap.
Big RK08: This was not the small RK8E, it was the half density monster with a light board and a lot of M series logic. Two drives working, one wrecked. Standalone, was probably attached to the 12.
A really big fpp12 that I rescued from a pdp12
A littler fpp-something that was damaged?
a MM8/I memory assembly with two cores
I might have had a spare RK8 controller plane that was damaged.

I never had a dectape or stuff like that. I believe I put the big FPP12 into the same rack as the working RK8, ran the thing for a few years, then sold it about 20 years ago to some local guy in MD. He was really happy. The pdp12 of course could not fit in my station wagon so it was left behind. I wonder if it's still there in that storeroom...

Anyway, what do those things look like? There seem to be parts that would fit an 8/I, but the odd part is a lot of those parts look like rebadged earlier parts (like the 115, 113, and 119 being the 133, 135, 139?) Some even have the old numbers on the boards in solder with new numbers on the M modules. Maybe this 8/I used later versions?

The presence of 4 310's and a 360 makes me think 8/I. But what are those M7718 and 7719's? And what are those arithmetic modules doing there? Or the monster number of high speed power inverters? And why no general registers, but a bunch of dual width 8*6 data mixers?
 
I believe I had a RK08 in my 8i. Two 1/2 density RK05 type platters. May have been referred to as an RK02? Yes, it had about 1/2 rack of flip chips as the controller. In my avatar you can probably see the two drives and the blinking panel. I have a video online of running a disk test and blinking the lights in the panel. Noisy as heck with the head moving back and forth. I never got more than 8k working with it. I had found a Fabritek 24k expansion unit. But it had some power issues. Like the -15v was totally fried. It had a TC08 with a TU56. It was quite the machine. Vince got the other 8i I had. This one went to Paul Allen's museum which is now closed and possibly defunct. Wish I could get it back. I can't say what all your flip chips were from. There are documents that show what goes in each slot of the 8i backplane. I used that to populate mine with an EAE which it did not have when I got it. Ah, I see Jack linked that document. I had not noticed that, it must not have been willing to show it to me when I was read only. Good luck with your machine. I could use another 8e if I can find one.
 
Yeah, what you have in the picture is *exactly* it. Except I had the drive and controller on the other side of my 8/I, same concept. And yes it was LOUD and worked and ran OS8 pretty well. I also had a Fabritek that kind of worked and a few other oddballs as well.

I don't miss it, that thing was just too big, and as a result I always wanted a little 8 like an 8/L. Now I have two of them, cute little things. I also have an 8/E but not sure what I'm going to do with it as it's kind of too simple a pdp8 with Omnibus and all.

AI (my KS10) went to LCM a number of years ago, but there is a note in the inside pocket that says if he ever decides to dump it or goes broke that I would come and get it with a U Haul. So I may wind up going up there if it winds up on a loading dock (like it did at MIT, Sandstorm, FTP, Digex, etc)
 
You must log in or register to reply here.
Back
Top