Currently Overloaded...

You don't normally expect zonking amounts of current to be flying around inside a computer (unless you've packed it solid with extra disk drives), so tagging a couple of skinny wires to one end of the circuit board is probably an eminently sensible approach. Those five DC volts will eventually find their way along the copper tracks and wander into the odd chip when required, and it's fairly unlikely you'll get flash-over between the connector pins or a nasty smell as several amps of current rumble uncontrolled through the resistors and capacitors.

In fact, the working bits inside a chip are now so small and close together that we worry about quantum effects (and if all the dead cats will interfere with calculations). I mean, what if an odd electron decided to go walkabout one day and debit your account at the bank instead of crediting it? If the nice man at Intel happened to sneeze when they were making the chip for your machine, it could mean that all of your spreadsheets are calculating the wrong answer.

See, this is the thing. We're so focused on digital stuff, miniaturized electronics, and tiny voltages squirming their way through the ever-increasing complexity of modern machines that we tend to forget that real electricity doesn't behave anything like this. It needs big chunky wires, and often involves several real amperes rather than those wimpy milliamp things - as I discovered when I went to see if I could help a friend sort out a problem with his model railway (railroad) layout last week.

When I was younger, and a practitioner in the art of miniaturized transport modeling, the only way you could get the locos to move was to shove lots of DC current through the track, accompanied (in the later period as electronics began to blossom into the modeling world) with a high frequency pulsed current that zapped the bits of dust and cat hair that might impede the flow of amps into the tiny electric motors. It you put two locos on the same track, they went round together and the speed controller box got hotter and hotter until the reset button popped out. Unless, of course, you'd wired them different so they went opposite ways. Then the head-on collision usually occurred before the reset button had time to react.

I remember reading, just at the end of my modeling days, about the new electronics that were coming to the hobby. Digital Command Control (DCC) was the upcoming thing. You just shove 15 volts AC into all of the track all of the time, and the individual control modules in each loco, turnout motor, and accessory allow you to individually control each one. Up to 99 separate channels, and superbly fine speed control as well because the chip can fire pulsed bursts of current into the electric motor, rather than just feeding it a constant voltage. And, from what I've seen in working layouts recently, it really does make everything easier and better. Another first for technology. It even uses wireless now so that the "driver" can wander around getting the best viewpoint without being constrained by (or tripping over) trailing cables.

However, the trouble with my friend's layout was that the fine control only worked for about 30 seconds before the control box decided all was not well and turned off the power. We initially suspected a short circuit, but my multi-meter could find no sign of one. We tested several different locomotives with no effect. Finally, we started measuring track voltages and current consumption in multiple places around the layout.

And, yes, even with DCC and the magic of electronics, model railway locos still do absorb quite a lot of power. The voltage drop as each one started was quite noticeable and it soon became clear that the controller was cutting out not when there were too many amps coming out of it, but when it detected that the voltage in the track had dropped below some predefined limit. Yet it took 30 seconds or so for this to occur.

Finally, after investigating the wiring layout, it became clear why. My friend had made the same assumption that computer system builders do - that you just need to tag on some skinny wires at one end, and the volts will meander through to where they're needed. In fact, this system used skinny wires to carry the track power all the way from the remote power unit to one end of a long and remote siding so that the whole layout (containing a very large four-track main line loop) dragged these volts through several very thin connectors, and lots of fishplates between track sections that (although soldered together) would move around with changes in humidity and movement of the baseboards and the floor underneath.

One of these joints was probably not as good as it could have been, and was increasingly resisting the not inconsiderable amount of current flowing through it. My suggestion was to go back to the principles of the electrician. Run couple of bus bars made of thick single-strand copper wire all the way from the power unit, continuing underneath the whole layout, and tag the track and everything else into it at regular intervals.

I suspect that it's not a solution you could apply to a recalcitrant server, though another friend did have lots of problems with a disk array that kept failing when he was driving it off an old 200 Watt power supply, so maybe there's correlation there. Amazingly, this friend uses speaker cable for his extra-high hi-fi stuff that could comfortably carry the output of a small power station. Mind you, at ten pounds (in money) per metre, it's probably a bit expensive for wiring up a model railway.