• Please review our updated Terms and Rules here
  • Exhibitor application for VCF West 2022 is now open! If you are interested in exhibiting, please fill out the form here.

Change CPU in a PDP-8A from KK8A -> KK8E

anders_bzn

Experienced Member
Joined
Jan 27, 2016
Messages
232
Location
Sweden
I got my hands on a working KK8E CPU set. I tried to swap out the KK8A CPU to the KK8E but when I started the machine with the KK8E nothing worked. Quite soon I found a broken fuse on one the top G0018 regulator board. After replacing the fuse and swapped back to the KK8A CPU everything works as normal.

I have a 20 slot chassie (8A/420)

I have in the system MM8-AA, Dataram 16K core, RL8A, RX8E, KK8A, KM8AA, DKC8A and VC8E

Just removed KK8A and inserted the KK8E board set in the slots as described in http://bitsavers.trailing-edge.com/...2_PDP-8A_Miniprocessor_Users_Manual_Dec76.pdf on page 2-8 (pdf page number 49).

Is there something that I have missed?
 

m_thompson

Veteran Member
Joined
Jul 8, 2014
Messages
1,637
Location
Rhode Island, USA
The M8320 Bus Loads board?

I think that you need one with the KK8-E processor, but the bus loads are build into the KK8-A.
More info in EK-0PDP8-SP-001_PDP-8_Family_Configuration_Guide_Apr78.pdf
 

anders_bzn

Experienced Member
Joined
Jan 27, 2016
Messages
232
Location
Sweden
Thanks for the document. It gave me some clues. I noticed that M8320 should have an ECO made before it is placed in a 8A. DEC ECO M8320-00007. On a 8A there is +5V on AA1 pin, in a 8E it's a testpoint. Therefor should R55 be removed.

And when I looked on my board the R55 were still there and the DEC 4008 transistor array has released the magic smoke... Might explains why the fuse was broken. Where can I find a new DEC 4008?
 

AK6DN

Veteran Member
Joined
Aug 23, 2010
Messages
984
Location
Silicon Valley USA
Thanks for the document. It gave me some clues. I noticed that M8320 should have an ECO made before it is placed in a 8A. DEC ECO M8320-00007. On a 8A there is +5V on AA1 pin, in a 8E it's a testpoint. Therefor should R55 be removed.

And when I looked on my board the R55 were still there and the DEC 4008 transistor array has released the magic smoke... Might explains why the fuse was broken. Where can I find a new DEC 4008?

You must have the rev D variation (with the DIP transistor array and R55 stuffed) of the M8320. By the time DEC got to rev J variation the DIP transistor array was replaced with 3x 2N3725 discrete transistors, and R55 was removed from the schematic. DEC didn't change the board etch, but just poked the three discrete transistors into the right holes in the DIP footprint.

You can buy new discrete 2N3725 parts today ($4.23 each at Mouser: http://www.mouser.com/ProductDetail/Central-Semiconductor/2N3725/?qs=Ypxpq5eNvNW7L0ybaddGmg== ) not cheap but available.

Don

***UPDATE***

Just happened to find this on eBay: http://www.ebay.com/itm/130361357169?_trksid=p2057872.m2749.l2649&ssPageName=STRK:MEBIDX:IT as NOS

Based on the fact that DEC used 2N3725 to replace these parts I suspect this is (probably) identical to the DEC4008 part ... Motorola MPQ3725.
Per the datasheet for the following MPQ3725 part available from Mouser, the package (DIP14) and pinout of the four transistor devices is 100% compatible.

Also available non-eBay from Mouser here: http://www.mouser.com/ProductDetail/Central-Semiconductor/MPQ3725/?qs=Ypxpq5eNvNV3EZNYiXC7LQ== as brand new parts

***UPDATE***
 
Last edited:

anders_bzn

Experienced Member
Joined
Jan 27, 2016
Messages
232
Location
Sweden
You must have the rev D variation (with the DIP transistor array and R55 stuffed) of the M8320. By the time DEC got to rev J variation the DIP transistor array was replaced with 3x 2N3725 discrete transistors, and R55 was removed from the schematic. DEC didn't change the board etch, but just poked the three discrete transistors into the right holes in the DIP footprint.

You can buy new discrete 2N3725 parts today ($4.23 each at Mouser: http://www.mouser.com/ProductDetail/Central-Semiconductor/2N3725/?qs=Ypxpq5eNvNW7L0ybaddGmg== ) not cheap but available.

Don

***UPDATE***

Just happened to find this on eBay: http://www.ebay.com/itm/130361357169?_trksid=p2057872.m2749.l2649&ssPageName=STRK:MEBIDX:IT as NOS

Based on the fact that DEC used 2N3725 to replace these parts I suspect this is (probably) identical to the DEC4008 part ... Motorola MPQ3725.
Per the datasheet for the following MPQ3725 part available from Mouser, the package (DIP14) and pinout of the four transistor devices is 100% compatible.

Also available non-eBay from Mouser here: http://www.mouser.com/ProductDetail/Central-Semiconductor/MPQ3725/?qs=Ypxpq5eNvNV3EZNYiXC7LQ== as brand new parts

***UPDATE***

Thanks a lot!

2N3725 sounded familiar, I found a bag with about ~100 transistors. DEC must have used them in 8/L as well. I replaced the broken array with transistors and removed R55 yesterday but I didn't had a chance to test it.
 

anders_bzn

Experienced Member
Joined
Jan 27, 2016
Messages
232
Location
Sweden
Now I have tested some more. No luck. Is there something more I should know before I start to debug?

The cards is inserted in the following slots:

1 - M8320
18 - M8300
19 - M8310
20 - M8330

A picture:
pdp8a_kk8e.jpg

I can't even access the memory from programmers console. I got worried about the NAND gate on the M8320 so I changed it (it sits electrically tight connected to the broken transistor array).
 

m_thompson

Veteran Member
Joined
Jul 8, 2014
Messages
1,637
Location
Rhode Island, USA
There is some interesting rise-time accelerator logic on the bus loads board as well as voltage regulators. You should make sure that that part of the bus loads board is working.

Many of the signals from the programmers console are static. I believe that holding the Load Address switch will put the address on the Omnibus and leave it there until you release the switch. You could measure the voltage on the 15 address lines and see if they correspond to the switch settings. You could also look at the Address lights on the front panel to see if they correspond to the Omnibus signals.
 

AK6DN

Veteran Member
Joined
Aug 23, 2010
Messages
984
Location
Silicon Valley USA
There is some interesting rise-time accelerator logic on the bus loads board as well as voltage regulators. You should make sure that that part of the bus loads board is working.

I would certainly do that. Put the bus load board on an extender (if you have one, that is) and measure the voltages on some of the key nodes (the termination voltage generators on the board). If you don't have an extender put it in the last slot of the backplane and move all the other cards as far forward as possible to give you room for probing. For this card I have debugged it on the bench using just benchtop power supplies (5V, +/- 15V) and a bit of logic to drive some of the (like TP4_H, TP2_H, F_L, INT_STROBE_H) nodes that are 7400 gate inputs.

I would also try pulling the four board set and putting back your original hex board CPU and see if the system is working.

Many of the signals from the programmers console are static. I believe that holding the Load Address switch will put the address on the Omnibus and leave it there until you release the switch. You could measure the voltage on the 15 address lines and see if they correspond to the switch settings. You could also look at the Address lights on the front panel to see if they correspond to the Omnibus signals.

Unfortunately he has a PDP-8A system box and backplane, not a PDP-8e/f/m console, which would be much easier to debug as you indicate.

Don
 

anders_bzn

Experienced Member
Joined
Jan 27, 2016
Messages
232
Location
Sweden
I would certainly do that. Put the bus load board on an extender (if you have one, that is) and measure the voltages on some of the key nodes (the termination voltage generators on the board). If you don't have an extender put it in the last slot of the backplane and move all the other cards as far forward as possible to give you room for probing. For this card I have debugged it on the bench using just benchtop power supplies (5V, +/- 15V) and a bit of logic to drive some of the (like TP4_H, TP2_H, F_L, INT_STROBE_H) nodes that are 7400 gate inputs.

I would also try pulling the four board set and putting back your original hex board CPU and see if the system is working.



Unfortunately he has a PDP-8A system box and backplane, not a PDP-8e/f/m console, which would be much easier to debug as you indicate.

Don

I have been sitting here and studied the schematics of the bus load board an I think I understand how it's suppose to work. I do agree upon your suggestions. I have extension boards, no problem there. I have measured some parts with a multimeter, the puls transformer seams fine and the diodes for the voltage references as well (I'll measure the voltage one I have the card in the machine again).

I have tested the machine with the original KK8-A CPU, it works well. Booting OS/8 from disc.

EDIT: Grammar
 
Last edited:

anders_bzn

Experienced Member
Joined
Jan 27, 2016
Messages
232
Location
Sweden
During the test of the bus load board I found that "INT STROBE H" only had an amplitude of 500mV. I found that the driver 74S40 on the timing generator board was broken. I changed it (to a SN7440) and got the signal working again.

The bus load board seam ok but I'm not sure. In the PDP-8/A there is a text "Notice that an M8320 module (Bus Loads) is inserted in slot 1 of the 8A600 and 8A620 computers. This module must be modified before it can be used in the 8A600 and 8A620 computers. If the modification has been accomplished, R55 (Figure 2-4) will have been removed; if R55 is present. carry out the procedure outlined in DEC ECO M8320-00007."

This made me to remove R55, but I don't know if that's the easy way to identify that the ECO has been made or if it's the complete change.

I found a ECO change log: http://bitsavers.trailing-edge.com/pdf/dec/pdp8/pdp8e/PDP-8E_ECO_LOG_Dec74.pdf

But the ECO M8320-00007 isn't covered in that version. Does anybody have a link to a source or the ECO list on paper?
 

anders_bzn

Experienced Member
Joined
Jan 27, 2016
Messages
232
Location
Sweden
Thanks Jack, our boards look the same. The board I got seams to be Rev. E (mark on the handle). Can you please lock if you can see any markings on the transistors?

It should be 2N3725.
 

AK6DN

Veteran Member
Joined
Aug 23, 2010
Messages
984
Location
Silicon Valley USA
I have both revD and revJ M8320 boards for my PDP-8m. One thing I noticed is that with the revJ with the discrete transistors is you need to check that the metal transistor case does not short against a board in the adjacent slot (typically a core memory G227 X/Y driver board). The board pitch is only 0.5", so the transistors must be assembled as low as possible to the bus loads board to prevent the shorting. I found that adding a bit of kapton tape insulator on the top of the transistors made me sleep better at night... or add some insulating tape on the backside of the G227 core driver board in the correct spot.

Don
 

anders_bzn

Experienced Member
Joined
Jan 27, 2016
Messages
232
Location
Sweden
I like pictures, here is my board.

M8320-rev-e-ECO.jpg

I'll put some tape on the top of the transistors, that's a good advice. This card sits in the top position in the back plane. I checked the clearance and there was very little space in between the transistors and the chassis.
 
Last edited:

anders_bzn

Experienced Member
Joined
Jan 27, 2016
Messages
232
Location
Sweden
M8320-rev-e-broken.jpg

I found a picture on the broken transistor array. But I can't understand how it could brake by holding the test point on AA1 to 5V.
 

bobaboba

Experienced Member
Joined
Sep 7, 2013
Messages
389
Location
Scotland
Doesn't look like physical damage.
From the colour of the board it looks as though it has run very hot and destroyed the package.
 
Top