Indigo 1: Dead Serial Port?

[hbent]

I have an Indigo 1 that's been running just fine for many years. The only weird thing about it is that no matter what I set in the PROM, it won't auto boot. This isn't really a problem, I just hook up a serial cable when starting up the machine, type "boot" at the prompt, and we're off on our merry way. Now, however, all of a sudden the machine is stuck at an orange LED (and doesn't chime, or make noise of any sort) if I have a serial cable connected to port 1 in any way - connected on the other end, not connected on the other end, doesn't matter. If I unplug the serial cable it chimes, starts right up, no problem. Unfortunately that isn't going to work for me as even though I have a keyboard and monitor connected, it never gets far enough to initialize them. Any suggestions on what to look for? There's nothing obviously wrong with the port and both of my 8 pin serial cables seem fine. I haven't yet pulled the CPU board but I suppose that's the next logical step.

 

[Elf]

Wow, that is pretty strange, just having the serial cable connected changes the behavior of the machine?

I wonder if something on one of the control lines of the serial cable has shorted (with time and use), or perhaps some physical defect has developed on the board and the serial cable is providing just enough torque to aggravate it?

I would say it is worth investigating electrically, but I'm not sure what tools you would have available for that. Perhaps at the very least check that none of the lines on the cable(s) are shorted to any of the other lines, try a new cable, and try providing various types of pressure to the cable in the port to see if it makes any difference. Basic stuff, but something basic must be going wrong if an unconnected cable just plugged into the port changes its behavior.

 

[weblacky]

I’m unsure about SGIs specifically but it’s often common, modern, practice to check for cable insertion by monitoring the grounding on the port. So I’d agree with Elf, likely there is a short. Now perhaps not a dead short, but even a high-resistance connection of rust or something can trigger such a feature.

I'd suggest (since you have two serial ports) that you attempt to probe two configurations:

1. Ground shield to pin (every pin).
2. All neighboring pins for connections between each other.

Now, let's be clear here. This port is already soldered to the PCB and already has IC connections. So pins will be connected in some fashion right now. I suggest you use compare serial 1 port with serial port 2. Do probing 1 & 2 on serial port 1 and record the values, then do it again on serial port 2 and compare values.