3725Altair 8800 Troubleshooting Help
- Apr 2, 2014Hello, all,
I have a nice original Altair 8800 system that includes the MITS CPU, later 18-slot motherboard, MITS floppy controller, and two 8" MITS floppy drives.
It also as a MITS 2SIO, a trio of MITS 16K Dynamic RAM boards, and a MITS 2K ROM board. It was built from a kit (K serial #), by someone who was clearly skilled, as the soldering work looks very nice.
The system ran beautifully up until yesterday, when a malfunction started. It started off intermittent, meaning that sometimes the system would work and other times, it wouldn't, and progressed to the point where it is malfunctioning continuously. When it is malfunctioning, the address lines count through the entire address space. RESET stops the operation as long as it is held on(as expected), but once released, it immediately resumes. Early on, as quickly as it started, the malfunction would stop, and the machine would run fine for a time, then, it'd go back to "counting" without the machine being touched at all. It is as if the processor is going into RUN state and the front panel is not able to bring it out of RUN state.
When it is in malfunction mode, activating the STOP switch has no effect. Holding STOP, then pressing and releasing RESET doesn't stop the behavior. I've checked the STOP/RUN switch, and it is working properly. When the malfunction is occurring, the RUN/STOP flip flop doesn't toggle when the RUN/STOP switch is activated (alternately RUN, then STOP). However, during times that the machine is behaving (which is less and less frequently), it does toggle properly.
I have removed all boards except one 16K memory board and the CPU, and the behavior persists. I have cleaned the edge connectors on the boards using De-Ox-It, and that didn't make any difference. I've reseated all of the IC's that are socketed, including the 8080A, and all fit firmly in their sockets. I've tried different memory boards, and even removed the memory completely, and it makes no difference.
I've checked the +8V to the front panel board, and it's running at +10, but the regulator on the front panel board is giving the ICs +5.00 V.
The main +8V to the motherboard is running at about +10V. The other motherboard voltages also check out within spec.
I checked both the front panel and motherboard voltages with a scope, and ripple is minimal, so the caps in the power supply seem to be OK.
I checked +5V at the regulator on the CPU board, and it too is good.
The malfunction does not appear to be shock related -- I can wiggle the front panel and CPU boards around, and even lift the machine slightly and jar it and it doesn't seem to create the malfunction, nor make it go away. It also does not appear to be temperature related - when it was intermittent, it would do it when cold, and warm. Now that the behavior has become permanent, it will sit there counting through memory from cold start for an hour, with no difference in the behavior.
Does anyone out there have any idea what might be going on with the machine? I'm wondering if anyone has had a similar experience, and may be able to provide some pointers as to where to look. I have all of the original MITS documentation for the machine, including schematics and board layouts. The front panel logic design with all the one-shots acting as de-bouncers and triggers for the EXAM/EXAM NEXT, DEPOSIT/DEPOSIT NEXT and SINGLE STEP functions makes me crazy. I'm wondering if one of the 74123's might have gotten flakey, then completely failed. If so, it could be messing with the PRDY signal, making the CPU think it's supposed to be running all the time. Are 74123's prone to failure?
Working on the front panel is almost impossible with the hard-wiring to the backplane on the right side, and the AC Power and 8V supply wiring on the left side, pretty much locking the panel in place. I don't want to mess with undoing the front-panel to backplane wiring, but have thought of putting a connector in on the 8V and AC Power wires so that at least the left-hand side of the front panel board will be free, and then panel can then be tipped up and over to get to the reverse side of the board. Would installing such a modification be a "bad thing"?
What do people think about using sockets when replacing chips? The original instructions have all of the ICs soldered to the board on the front panel. If something in the front panel logic has failed, would it be verboten to cut out the failed chip, clean out the holes, and put a high-quality socket in, so that in the event of issues in the future, IC's can be replaced more easily.
I am not really concerned about the collector value of the machine, as I never intend to sell it. I just want to have this machine running again, and will do what is necessary to facilitate its repair, but am interested in opinions about making modifications to make the machine more serviceable. Clearly, MITS didn't design the front panel to be "service friendly".
Any thoughts will be greatly appreciated.
- Next post in topic >>