[Received via e-mail: March 20, 1997]
First, a few words about aeroplanes so that the rest makes more sense:
After World War II, the four major powers -- USA, Russia, France and Britain -- were the only ones with nuclear weapons. And to deliver these weapons (a euphemism for "bomb the hell out of somebody") they all had significant delivery vehicles: bombers. Britain built three types and called them the V-bombers. They were the Vulcan, the Valiant and the Victor. The Vulcan was the most beautiful aeroplane, being an almost perfect delta-wing with a single large fin. The other two were progressively uglier. Anyhow as a young boy I was most interested in these planes and could pick one out at any distance.
Now, In 1974 I was told to go to Peru and fix a power supply on a system that had been blowing fuses every day for the past year. This was not some small power supply like we have now, but weighed in at sixteen hundred pounds. And a system could have two of them (which was part of the problem). The number of power supplies that a customer had was determined by the number of cabinets on the system. If the system had two I/O cabinets then the customer had two power supplies, even though the system may have been small and the load called for only one.
These power supplies worked1 by rectifying the mains voltage to produce 600 volts DC and then chopping this with a bridge of SCRs to produce an AC current that then went though a transformer to produce 6.5 volts at about 400 amps to supply the logic. It ran at about 2000 Hz (more on this, later). The amazing thing was that it worked.
The problem on this site was that the load was not enough on either of the two supplies and also they were out of adjustment. Anyhow, when I got there I quickly rebuilt the supply that was failing and put it on adjustment and was convinced that I could then leave.
"Oh no," they said. "This has been failing for the last 18 months and you are not going until it has worked for four days without any problems." I was young enough then not to consider that I might never leave Peru.
As I was now firmly stuck with nothing to do but wait for the next four days, I asked what I could do to help. They said that they needed some help at the Air Force computer site as they were installing another cabinet on the system. Although it was another model of computer, I was also knowledgeable on that style and so agreed to help.
The cabinet took some time to install as it had been sitting in a warehouse for some time and all the capacitors had to be reformed (another problem we don't have to worry about now). While I was there I started to talk to one of the Peruvian Air Force high-ups. (How did I know this? Well, he had lots of gold braid all over his uniform.) In our discussion I told him that I had flown into the main airport (which was both a commercial and a military airport) and had seen an ex-RAF Victor bomber.
"No, you didn't," he said.
Well I didn't get the point at first and kind of insisted that I had.
"No! You did not," he repeated, at which point I had to agree.
We did not discuss aeroplanes again, although we both knew that it would have been interesting.
I said that I would tell you more about the 2000 Hz frequency in the power supply. 2000 Hz is one of the most sensitive parts of human hearing, and also a good frequency to send over the phone. Well, while I was in Peru I realized that the adjustment could be done by ear. In fact connecting an oscilloscope up to the supply was dangerous because you had to make it electrically hot.
The next time that somebody called because they had the same problem, I told them to take the phone over to the power supply and follow my instructions. I then told them to take an adjustment tool and turn the screw until I said stop. They could turn the screw in either direction as long as they remembered which way they went. When I heard the supply make the wrong noise I told them to stop and then turn the screw the other way until I again said stop, but to remember how many turns they made. I then told them to divide this number by two and turn it back to that position.
Most of the people were concerned that I could do this over the phone, but I assured them that it would work and that they should call me if they still had the problem. The only calls I ever got were from people that had not believed but after a week had to call me to say that everything was now OK. Only a couple of them swore at that point, as they had been working on this problem for a long time and it shouldn't have been that easy, and then they laughed.
I tried to put this adjustment into a technical paper, but it kind of looked strange.
[Basically, the power supply worked in three stages.
The first stage converted "mains voltage" (which is alternating current, or AC, from a wall-socket) to direct current (DC). In this case it converted it to 600 volts DC -- roughly equivalent to 50 car batteries.
Once the power had been converted to DC in the first stage it could be converted back again to AC in the second stage (using a bridge of Silicon Controlled Rectifiers, or SCRs). Now, though, it could be made to oscillate two-thousand times each second (2000Hz), instead of at the frequency that originally came from the wall-socket. In the U.S. this is 60Hz, but the power supply had to work in many different countries where that might not be the case.
After the second stage had converted the high-voltage DC to AC at 2000Hz, the third stage, a transformer, converted the high-voltage AC to a lower voltage, but at a higher current. This was then used to supply power to the rest of the system. --tim]