• Please review our updated Terms and Rules here

EAE for PDP-8/e Repair

m_thompson

Veteran Member
Joined
Jul 8, 2014
Messages
1,985
Location
Rhode Island, USA
I just got the EAE boards for my PDP-8/e. OS/8 won't run with them installed, so there is some debugging and repair to do. Eventually I plan to install an RK8E in this system, so I will need a check the ECO levels of all of the boards processor and disk controller boards.

Maintenance Volume 2 says to run Instruction Test 1 & 2, install just the M8340 EAE Decoder and Step Counter board, connect it to the M8330 Timing Generator board with the over-the-top connector, and rerun Instruction Test 1 & 2. Both diags work OK without the M8340 installed, but Instruction Test 2 fails when an interrupt is not generated.

Looking at the prints, the M8340 has SN7474 ICs directly connected to the Omnibus. I have seen LOTS of 7474 failures in PDP-8 systems, I will probably look there first.
 
After spending some time looking at posts from others with EAE issues I am now wondering if these boards are worth fixing.

The M8340 is etch revision D, S/N 144. The current assembly revision is F.
The M8341 is etch revision B, assembly revision C. The current assembly revision is E.

Maybe I can determine what changes were made and upgrade the revision level of the boards that I have.
 
After spending some time looking at posts from others with EAE issues I am now wondering if these boards are worth fixing.

The M8340 is etch revision D, S/N 144. The current assembly revision is F.
The M8341 is etch revision B, assembly revision C. The current assembly revision is E.

Maybe I can determine what changes were made and upgrade the revision level of the boards that I have.

I am still battling with the couple of EAE sets in my possession. I've got a good bit of debugging to do.

In case you haven't seen it elsewhere, here's some debugging I've done so far on one set: http://imgur.com/a/QVRLs

Best of luck, and let me know if I can be of assistance.

Kyle
 
I don't think that debugging the no-interrupt with the M8340 installed will be too difficult to do. It is probably hanging a signal like INT IN PROG.

I read through the ECO log and found LOTS of ECOs for the 8e boards and lots of requirements for certain revisions of boards. My EAE boards are early revisions, so there may be interesting issues that were fixed in later versions.
 
What revision level are the EAE boards that you have.

I'll check more in depth tonight, but I've been playing with, at the moment, etch Rev. D of both M8340 and M8341, as I recall. According to the DEC ECO Log, this combination shouldn't work. I'm trying to get a schematic of the earlier revisions to compare to, but I know they killed off a few gates that were causing too much delay for one board (imagine that, 7474 inputs tied directly to the Omnibus!).

Kyle
 
imagine that, 7474 inputs tied directly to the Omnibus!.
I was really surprised t see that.

I only have the schematics that are on BitSavers. We should take pictures of both sides of the different PCB and assembly revisions of the boards to reverse engineer the changes. Maybe we could recreate the earlier schematics, and document the changes in the ECOs. I really would like to get these boards working.

One of the ECOs was for 80 slot systems. That won't matter much in my 20 slot system.
 
I connected the 'scope to INT IN PROG without the M8340 installed, ran a short program to enable interrupts and sit in a JMP loop. I then generated an interrupt and INT IN PROG went high for about 500ns, and the interrupts were disabled. With the M8340 installed there is no activity on INT IN PROG. I thought that this would be easy to debug and just cut pin 7 on E16, the SP384A IC that is the receiver for INT IN PROG. Well, that didn't change anything, so I am now looking through the M8330 prints to see what else in the interrupt logic could be affected by installing the M8340.
 
I was really surprised t see that.

I only have the schematics that are on BitSavers. We should take pictures of both sides of the different PCB and assembly revisions of the boards to reverse engineer the changes. Maybe we could recreate the earlier schematics, and document the changes in the ECOs. I really would like to get these boards working.

One of the ECOs was for 80 slot systems. That won't matter much in my 20 slot system.

I'm working on that already. I've got most of the schematic for the latest revision laid out in KiCad, and will duplicate that and make the appropriate changes for the earlier revisions.

Once I get that schematic finished, I'm planning on laying out a modern EAE board (just a single board). Someone else suggested I go ahead and add a couple of DACs, a 12-bit input port, and an autoloader to make the ultimate Spacewar! board for an 8/E. However, I think I'll try to do everything individually at first. It's my goal to have an autoloader board finished by the end of summer, but we'll see how my time goes.

Kyle
 
I didn't see schematics or layouts for this already, or at least they're not on his site as best I can tell. Do you have them?

It's on my list of things to do, but I don't think I've done it yet. A few boards (RX8E comes to mind) are
misfiled on the website, but I don't the the EAE boards are among them.

Vince
 

I'm afraid that creating the projects and archiving the photos is as far as I've
got to date. I'd be happy to archive photos of different revisions, as well as
ECO history or whatever else you've got.

Vince
 
Without the M8340 and M8341 installed the 8/e will pass instruction test #2, so interrupts are working. With the M8340 installed and connected to the M8330 instruction test #2 fails with a missing interrupt. Without the over-the-top connection between the M8330 and the M8340 it will pass instruction test #2.

That meant that the problem had to be in connector J that connects to the M8330. I marked all of the connector signals on the M8340 board schematic in a different color depending on if they went to the Omnibus, the M8330, or the M8341. I looked at the signals on the J connector and compared them to the revision F prints. (I don't have any earlier prints) I found that contact JV2 was a ground on my M8340 and was used to pass the RESTART signal from the M8341 to the M8330 on the prints. I cut the ground connection for the JV2 contact, and now instruction test #2 will run. I will solder a jumper between contact JV2 and HH2 on the M8340.

I installed the M8341 without the over-the-top connectors, and found that address line #1 is always active. That problem should not be difficult to find.
 
I just sent out an order for a small PCB that will have a DEC-style edge connector socket on one side and a 40-pin header on the other. For over-the-top OMNIBUS boards, the socket will fit on the E/F/J/H top connector with the 40-pin header facing up. You can then use an arbitrary length 40-conductor IDC ribbon cable to connect to a second adapter on another board. This will let you use an extender board with over-the-top board sets while still allowing normal operation for testing. Here's a view of the layout -
attachment.php

Sorry for such a mumbly description - I don't have a 3D drawing of the DEC socket to use in constructing a rendering of the board, but basically, the socket points down and the header points up. Boards should be back in about 10 days and then I'll be able to send a few out to you and Kyle for checkout.

Jack
 

Attachments

  • DEC_Extender_Adapter.jpg
    DEC_Extender_Adapter.jpg
    16.7 KB · Views: 1
Have you many of these on order Jack? I made up some extenders with ribbon cable a short while ago but it's a lot of soldering to make up more and I have an RK8E to debug.
 
The initial order is for six boards, enough to make up three sets of cables. This is only my second attempt at PCB creation, so I need to be pretty sure I haven't messed things up too badly before sharing the layout. Once things are checked out, I will put in a larger order. I have enough of the DEC-style connectors to make up up 50 sets of cables so there should be plenty to go around.

Note that the connector I used (Electronic Connector Service ECS-2418-11-30) is an exact match for the DEC single-high connector at the socket end but the solder pins are in parallel rather than staggered rows as on the original DEC part.

http://ecsconn.com/ProductDetails/2...-125-Contact-Spacing-Dec-Style-Press-Fit.html

Jack
 
I used the Douglas connectors sawn in half, all I could find at the time, but these look much better. Does the pcb allow normal omnibus card spacing or is it a little wider (not that it matters much)?

Please keep your progress visible on this board - separate thread maybe?
Thanks
Bob
 
Good idea re separate thread - I'll do that right now.

As for spacing, the adapter should be about the width of the existing over-the-top blocks. No problem on the extended end but things could get a bit tight if there is a card-handle directly in front of one of the boards remaining at normal height. Hopefully the boards could be separated one slot on the backplane without causing trouble.

Jack
 
Back
Top