| Back | Contents | Next |
Burglar alarm systems often use magnetically operated sensors to detect when a door or window is opened; in this case, the sensor was a simple pair of contacts that close when a magnet is near the sensor. If the magnet is near the sensor, the alarm control sees a normal voltage about midway between power and ground. If the magnet leaves the proximity of the sensor, as when a door is opened, the alarm control sees power supply voltage. Or if the wires to the sensor are shorted, the alarm control sees ground. Businesses can have quite sophisticated systems. Now, one place had an outdoor concrete structure with a magnetic sensor on the door. The zone was armed after hours, but sometimes someone had to get into that structure after hours, so there was a switch that was wired in parallel with the magnetic sensor contacts. If someone needed to get in, the security operator could just flip the double-pole switch and a red LED would come on, the LED being controlled by the other pair of double-pole switch contacts. The rest of the zone remained armed, while the zone would ignore that one outside structure door.
One day, the alarm computer reported a funny voltage on that zone. The voltage was not off enough to make an alarm, so the computer just gave a warning that maybe something was starting to come loose. The security operator, wondering, flipped the switch on and the voltage returned to normal. He flipped the switch back to off, and the voltage stayed normal, so he forgot about it. But you do not ignore problems with security systems. So, after about the third time, the owner got involved and figured the switch was bad. I arrived one fine day, new switch in hand, but after I studied the alarm system computer logs and looked at the wiring, it did not appear likely that the switch was the problem. After all, unless the switch was on (shorting the sensor circuit), it was not really even in the circuit, so its contact resistance would have no effect if the switch was off. The switch was sealed, so leakage across the poles seemed unlikely also.
So I left the switch alone and, key in hand, went out to that door thinking that either the underground cable or the magnetic sensor was bad. I opened the door and went back to the security panel, and behold, the voltage had not changed; it was still just off a bit. So now was it the underground cable or the magnetic sensor? I disconnected the magnetic sensor and now the controller showed that the door was opened. I put a jumper wire in place of the magnetic sensor and now the zone showed a normal voltage. I put the sensor back in the circuit with jumpers and the voltage was off a bit again. I again replaced the sensor with a jumper and checked the bypass switch with the door open. The red light went on and the zone voltage was good.
So I pulled the sensor out of its hole and it was wet as a fish and the terminals were all corroded. That explained a lot, but not why flipping the switch “fixed” the problem. Quite curious, I measured the switch resistance. The reading was about 1 kΩ, about half the 2 kΩ end-of-line resistor that formed one half of the resistor voltage divider that produced a normal zone voltage. My meter was telling me that the sensor should have been making the alarm go off, not just giving a slightly odd voltage. Wondering if my meter was bad, I measured the voltage across the sensor, expecting it to be zero. It was not; it was the same voltage as the computer log had indicated, about 0.3V. That sensor should not have been a source of voltage; that voltage was what fooled the resistance measurement function. On a hunch, I shorted the sensor leads together and measured the voltage again. It was zero. I repeated my original resistance measurement, and the sensor was dead shorted, even with no magnet. So not only were the contacts corroded shut, but the corrosion, water, and dissimilar metals somewhere in the switch had created, over time, an electrochemical cell.67 When I shorted the sensor leads, or the security operator had flipped the switch, the short drained the cell so that the now-dead cell did not create any strange voltages. In sum, what had seemed like a simple electrical problem was actually a chemistry problem.
Here was a simple thing nobody had thought of, which changed everybody’s entire understanding of something. The original assumption of a bad switch was dead wrong. The owner could not think of any other explanation. I never expected a chemistry problem. Only God knows all the mistakes and wrong assumptions that long-day creationists, Framework Hypothesis proponents, and proponents of Big Bang and evolution theories are making.
In this next one, I did not exactly come out of it with an enhanced reputation. There was a piece of equipment that consumed some fair amount of power that had to be available at all times, so there was no off switch. To work on it, you removed the fuse. The equipment had come over on the Mayflower and was showing its age. I had designed, built, and programmed a new controller that used a then-modern microcontroller chip. It worked just fine for a good while, but then one day it just sat there and sort of stuttered. I went up to the equipment, pulled the fuse and, not immediately finding any problem, put it back in to test it further. It immediately started working fine.
Now, most of us have experienced computer crashes or malfunctions. They do that. I used to say that, if cars were like computers, a minor adjustment to your carburetor could make your muffler fall off. But people don’t like equipment downtime, and trying to find the cause of a once-in-several-months glitch was about as appealing to me as getting life without parole, so I put a reset switch for the microcontroller on the front panel. It was a quick fix. After all, the standard advice before calling computer tech support is to first reboot. So the reset switch seemed to be a reasonable idea.
Sometime later, it started stuttering again, but the reset switch did not work. I pulled the fuse and found no obvious problem, put the fuse back in and it immediately started working fine, just as before. Now, this kind of thing does occasionally happen; microcontrollers can be quite complex little beasts and I had seen my share of odd behavior from time to time that turned out to be due to a bug in the microcontroller chip itself. That the microcontroller’s own internal watchdog “obviously” slept instead of barking seemed to confirm this. But the equipment was critical, so I decided that, if killing the power solved the problem, the best thing would be to change the microcontroller program to emit a periodic toggling signal and build a hardware watchdog that would cycle the power to the microcontroller if the toggling signal quit. The owner grudgingly bore the cost and I built and installed it. Things were fine for a good long while. Then all of a sudden the equipment quit completely. I went in and, sure enough, it was totally dead. I traced back through the chain of power supplies from the microcontroller all the way back to the fuse. There was voltage on the line side of the fuse, but not the load side. I pulled the fuse, and it looked perfectly good; the element was a big, easy-to-see, high-amperage thing. Wondering if I had made a mistake, I measured the fuse resistance and it was not all the way open, but had high resistance. I measured it again, and the resistance had changed. Pushing and pulling on it changed the resistance change some more.
So, all this time, the fuse was not blown but had an intermittent bad connection somewhere in it. The equipment stuttered because the fuse let in just enough current to charge up the microcontroller supply, then the microcontroller started up and began to operate the equipment. But the equipment load shorted the high resistance fuse connection out and the microcontroller lost its power also and the cycle repeated. So much trouble was caused by a defectively manufactured, but not blown, fuse. When I pulled it and put it back in, I suppose something was just loose enough to wipe the “contacts” and fix the open circuit. I have seen many intermittents, but that occasion was the first intermittent fuse I had seen. Looking for a complex problem, I had overlooked a simple possibility. Of course, another problem with troubleshooting intermittents is that you can look at it cross-eyed and “fix” it.
Small mistakes or small things overlooked can greatly mislead one! Unlike this case, where I got feedback on my work, proponents of creation by chance will never get feedback directly from the unobservable past. Alas, one day God will give them feedback much to their shame and eternal horror and torment in Hell.
67If you have a modern digital voltmeter you can make something like this yourself. Take a bright penny and a nickel. Put the nickel on the table, wet a piece of paper towel with salt water and put it on the nickel, leaving an edge exposed. Put the penny on the towel so it does not touch the nickel. Now measure the voltage across the penny and nickel. You should get about 0.3 volts.
| Back | Contents | Next |