Weekend Recap, or "The Magic Smoke Got Out"
On Saturday we started working on the second core. Timing the rising edges of the Field 0 and Field 1 pulses was as far as we got before a probe slipped in the backplane and shorted a pin, letting out some smoke.
We powered it off and then back on again to see things weren't working correctly.
Inspecting further, we found the probe was connected to -30V. I pulled out cards in the area while Warren tested them on his tester. I pulled out an M216 in slot E9 and saw it had some crispy looking pins. Warren tested it and confirmed that it was dead.
We replaced all three 7474 chips and put the card back in.
We tried sending Warrens Tune program (for tuning the cores) but we got no sound from the speaker. We tried to run it and it wouldn't run. Strange.
Warren determined that the SMA (7500, skip on minus AC) and SPA (7510, skip on plus AC) were not skipping. So he put together a little program to use while we were tracking down the source of this.
0200 / 7604 LAS (load AC with switch register)
0201 / 7510 SPA (skip on plus AC)
0202 / 7001 (increment AC)
0203 / 5200 (jump to 200)
We tested the M121 in H33 and the M113 in J31. Both were fine. We checked the clock in signal on J29 pin B1 and this was fine.
Skip FF & H bit on pg. 74 of maintenance manual vol 3 was not the problem.
Next we checked bit 0 of the AC.
We swapped H15/J15 (Bits 0 and 1) with H20/J20 (Bits 10 and 11). No change. Swapped back.
Maybe bit 5 of the MB register was the problem. We swapped H17/J17 (Bits 4 and 5) with H20/J20. No change. Swapped back.
We called it a night after this.
On Sunday we tested bit 0 of the AC on H33 J1. This was good, it had a nice output showing highs and lows.
Next we looked at the output of bit 5 of the MB. We connected the scope to J17R1 and R2. (After connecting them to the wrong pins a couple times). There seemed to a problem here. The top graph in both pictures is the output from the Skip Flip Flop. R1 and R2 are the high and low outputs of the flip flop for MB05. You can see R1 changes based on the pulses from the Skip Flip Flop.
R1 was fine, but R2 was wrong. You can see that there is a little flicker of something where it tries to go to high, but something keeps pulling it low.
So, MB05 (0)L (or MB05(1)H, same thing with a different name) is the problem signal.
The next task was to find where exactly that signal went.
In Operate Group 1 it goes to the CLL instruction (7100, clear link). In Group 2 it goes to SMA (skip on minus AC, which we knew didn't work). In the EAE instructions it goes to MQA (7501, inclusive OR, MQ with AC). It also goes to a number of IOT instructions.
We checked the memory inhibit which used MB05(0) and it was fine. We checked the MB05(1)L signal going to the console/display and that was fine too.
We checked the buffer for the I/O cable M15 K2. The M623 module was fine.
M119 in K10 (used for the CPR on pg. 45) was fine.
M113 in H10 (Link logic, pg. 53) was fine.
M617 in M24 (ext. memory buffer) was fine.
M119 in A31 (IOT 615x tape controller maintenance register) was fine.
M217 in E07 (used for KW12) has bad logic on pins K1, A1, F2, and E2. We will replace this later as it's for the real time clock option and not the main processor or memory.
A615 E/F36 and E/F37 (DSC D-A for character display)
M117 in C23 had a cracked solder joint. Reflowed the solder. Didn't solve our problem.
In the end it turns out I accidentally plugged a G882 into F3 when it should have been in F5. This was pulling the output to ground which is why R2 was always low. Oops.
This fixed some of our problems. The skip instructions work now, but we can't run the programs from MTTTY still. We tried to load one and it sent and seemed to be a good load, but it started to print a bunch of NULLs out to the screen and it wouldn't actually run the program we loaded.
So something is still not quite right.
We powered it off and then back on again to see things weren't working correctly.
Inspecting further, we found the probe was connected to -30V. I pulled out cards in the area while Warren tested them on his tester. I pulled out an M216 in slot E9 and saw it had some crispy looking pins. Warren tested it and confirmed that it was dead.
We replaced all three 7474 chips and put the card back in.
We tried sending Warrens Tune program (for tuning the cores) but we got no sound from the speaker. We tried to run it and it wouldn't run. Strange.
Warren determined that the SMA (7500, skip on minus AC) and SPA (7510, skip on plus AC) were not skipping. So he put together a little program to use while we were tracking down the source of this.
0200 / 7604 LAS (load AC with switch register)
0201 / 7510 SPA (skip on plus AC)
0202 / 7001 (increment AC)
0203 / 5200 (jump to 200)
We tested the M121 in H33 and the M113 in J31. Both were fine. We checked the clock in signal on J29 pin B1 and this was fine.
Skip FF & H bit on pg. 74 of maintenance manual vol 3 was not the problem.
Next we checked bit 0 of the AC.
We swapped H15/J15 (Bits 0 and 1) with H20/J20 (Bits 10 and 11). No change. Swapped back.
Maybe bit 5 of the MB register was the problem. We swapped H17/J17 (Bits 4 and 5) with H20/J20. No change. Swapped back.
We called it a night after this.
On Sunday we tested bit 0 of the AC on H33 J1. This was good, it had a nice output showing highs and lows.
Next we looked at the output of bit 5 of the MB. We connected the scope to J17R1 and R2. (After connecting them to the wrong pins a couple times). There seemed to a problem here. The top graph in both pictures is the output from the Skip Flip Flop. R1 and R2 are the high and low outputs of the flip flop for MB05. You can see R1 changes based on the pulses from the Skip Flip Flop.
 |
| R1 |
 |
| R2 |
So, MB05 (0)L (or MB05(1)H, same thing with a different name) is the problem signal.
The next task was to find where exactly that signal went.
In Operate Group 1 it goes to the CLL instruction (7100, clear link). In Group 2 it goes to SMA (skip on minus AC, which we knew didn't work). In the EAE instructions it goes to MQA (7501, inclusive OR, MQ with AC). It also goes to a number of IOT instructions.
We checked the memory inhibit which used MB05(0) and it was fine. We checked the MB05(1)L signal going to the console/display and that was fine too.
We checked the buffer for the I/O cable M15 K2. The M623 module was fine.
M119 in K10 (used for the CPR on pg. 45) was fine.
M113 in H10 (Link logic, pg. 53) was fine.
M617 in M24 (ext. memory buffer) was fine.
M119 in A31 (IOT 615x tape controller maintenance register) was fine.
M217 in E07 (used for KW12) has bad logic on pins K1, A1, F2, and E2. We will replace this later as it's for the real time clock option and not the main processor or memory.
A615 E/F36 and E/F37 (DSC D-A for character display)
M117 in C23 had a cracked solder joint. Reflowed the solder. Didn't solve our problem.
In the end it turns out I accidentally plugged a G882 into F3 when it should have been in F5. This was pulling the output to ground which is why R2 was always low. Oops.
This fixed some of our problems. The skip instructions work now, but we can't run the programs from MTTTY still. We tried to load one and it sent and seemed to be a good load, but it started to print a bunch of NULLs out to the screen and it wouldn't actually run the program we loaded.
So something is still not quite right.
I did that on a pdp-8i long ago. A probe slipped and I had several G series cards to replace.
ReplyDelete