One week at 100Mhz: Scary power failures

The one sad part about working with this machine is that I have discovered one or two mechanical flaws that are worrisome. And this time it’s not just the lack of USB ports.

I mentioned there is one detached key cap — something that could probably be fixed very quickly with a drop of cyanoacrilate — and that there is an occasional line of blue pixels running down the screen. But those are both overlookable … if that’s a word … since neither one really prevents me from using it.

On the other hand, this box has a sketchy power switch. If you remember old machines like this, some of them had two “power buttons” — one a simple power activator, like a push button on an elevator, and another one which was actually a switch that physically cut power to the system. Some old desktops put that switch at the back, and it was usually a rocker switch of some kind.

This has a similar arrangement, with a plastic slider as the “rocker” switch on the back left, and a sort of “restart” kind of button on the front. If I power down the machine and the AC is still attached, I can turn on the system by pressing that activation button. Or I can completely sever the power by switching the plastic switch at the back.

The fault lies somewhere in that rear power switch, which is probably worn with age. Occasionally it doesn’t slide fully to the “on” position, and the current will snap in and out at a frightening rate. It’s like an electric pulse popping through the entire system every half-second. The screen throbs white, the drives cluck and click as their power is applied and then cut immediately. It’s exceptionally scary. 😯

Generally I try to avoid messing with that switch, leaving it in a fully “on” position and using a power strip “surge protector” to cut power when I leave for the day, or when I want to be sure the AC adapter isn’t pulling a current. It seems to be an okay solution.

The same switch might be to blame for another strange phenomenon. Right now the weather is turning from cold to cool, and the humidity is up as the spring rains set in. In the morning, when the machine has been powered down for the night, it’s starting at a very low temperature (I do not have “central heat,” as is common in some other countries) and as it heats up, some of the components flex in their braces.

So usually after about 20 minutes of use, I’ll hear the distinct “pop” of a component shifting as it expands, and the power cycles on the machine. It usually happens after I place my hands on the plamrests, changing the pressure on the frame and allowing the components to slide slightly. And I have a feeling the guilty party is that same switch again, shifting and losing its connection when the temperature reaches a certain point and its board relaxes.

As it gets warmer it shouldn’t be as much of a problem. But it does mean in the here and now, I occasionally walk away from the machine and come back to find it rebooting and doing a file system check. 👿

Neither of these things came into play in the past because the system was never really on for very long. I would boot a kernel, try a configuration, test a piece of software, then shut down. And rarely did I rely on it for so much typing and “hands-on” use.

But neither of these things bodes well for the machine beyond my personal use. It will take a geek of considerable patience to put up with a sketchy power switch, and it will take a geek of even more considerable expertise to consider repairing that. I don’t count myself among the electrically-minded technophiles; if a component stops working, that’s where I draw the line for usability.

Oh, and I am also a kludge with a soldering iron. 🙄

But these are the things you discover when you take on a charity case and put it back to work again. Every machine has its quirks, I just wish this one’s weren’t quite so dramatic. 😯