I turned off the light and lay down, and almost immediately discovered a hangnail--turned on the lamp, FLASH! the bulb went. I turned on the lamp on the other side of the bed, replaced the bulb in the main lamp, took care of the hangnail, and lay down again, lights out.
It took me the usual half-hour for the flywheels to spin down; and I had to hit the can, so I got up again and--*sniff*--I could smell a hot/melting odor in the bathroom, something that smelled like electronics and plastic about to catch fire. I spent five fruitless minutes trying to find the source of the odor, then gave up and went back to bed.
So, another half-hour for the flywheels to spin down. I was laying in bed, and I was beginning to drift off, when the power failed for about 15 seconds. I've got three fans running; the sudden cessation of noise told me the power was off and there's always that pained oh, shit you get when that happens.
Though it came back on pretty quickly, the damage was done, and again--
Sighing, I rolled over and tried to go to sleep; the power failed again--briefly, again--and came back on even faster.
I went out to the kitchen to turn the light back on; the UPS that the modem and router are on (in Mom's room) was behaving...oddly, so I detoured there to check it out.
The UPS was clicking to itself, as if it couldn't decide what it was supposed to do. Turning it off and on did nothing. I swapped it for the UPS that Mom's computer was on, and this one clicked a bit and then started screeching periodically as if the power had failed.
Reseated the plug; it kept right on screeching. I checked the circuit breakers, thinking something might have tripped (Why would it do that??) and finally I went out to the garage for my multimeter. Turned on the stove light on my way, and FLASH it blew out. Argh!
By now it was past 4:30 and the eastern sky was getting light; I considered turning on the sink light, decided against it--it wasn't going to be dark that much longer--and continued my mission.
...went to the garage, got my meter; DMM reported that there was power at that outlet: 139.5 volts.
...even accounting for peak-to-peak reading rather than RMS, that's way too freakin' high. 140 volts is 20 volts higher than the nominal 120 volts you're supposed to get; it's 16.7% overvoltage. That might explain why I popped two light bulbs within the span of an hour. It might also explain the hot/melty smell: the UPS was giving up the ghost.
I left the modem UPS unplugged and triple-checked my results by testing the voltage at other outlets and having a look at what my computer's much more sophisticated UPS told me. I don't even have to have the computer on; there's a button on the front panel I can press to cycle through things like voltage and run time. It said 140 volts, plain as day.
Shut down the computer.
...with that, my little adventure came to an end; I lay back in bed and spent another half-hour waiting for the flywheels to spin down. I didn't get to sleep before 6 AM today.
* * *
I spent that half hour fruitlessly trying to remember whether or not a resonant circuit could exceed its input voltage. Say, for example, the conditions on a specific piece of the electrical grid just happened to hit the right combination such that it resonated precisely at 60 Hz. In some magical way, the inductance and capacitance were balanced correctly.
No matter how I thought about it, I couldn't see how that could be; most loads on the electrical grid are inductive and relatively few are capacitive. Think about the big power hogs in your house: they're all electric motors, things like AC compressors and laundry equipment and such.
Even if there was resonance, I didn't see how it could make 20 extra volts appear out of nowhere. Somehow--for some inane reason--the power company was delivering too much voltage.
The power companies play games with the juice in summertime; they have to. With all those air conditioners loading up the grid, they have to find ways to lower the stress on the infrastructure. There are a variety of reasons why, things like overheating equipment and reaching the output limit of the generators, but the reasons are largely unimportant for the purposes of this discussion.
One thing they do is to run the generators a little slower, so that the power is delivered at 59 Hz or maybe even 58. Not a lot slower, not enough to cause any real trouble for electric motors (which have to be designed for a specific frequency, and which will burn out if fed the wrong kind). And after the peak load period has passed, they'll run the generators that much faster to make up for it, just in case someone has a clock that takes its timing from the not-so-steady beat of the AC current powering it. So if they run 'em at 58 Hz for two hours, they'll run 'em at 62 Hz for two hours off-peak. But this isn't done unless things are dire, because as I said it's bad for electric motors to be run at the wrong frequency, even if it's only off by two or three percent. (At best, they won't run as efficiently, which actually makes the problem worse rather than better, because they use more power.) In fact, this isn't done all that much any more simply because the grid is so interconnected that everyone has to reduce frequency at the same time and by the same amount, lest chaos reign--and I mean the "eastern seaboard blacked out" level of chaos.
Another is to reduce the voltage. Instead of 120V, they'll supply 110. There's not a lot of leeway but there is some; and this particular trick doesn't require "make up" time--when the loading eases off you just go back to 120V. This is easy to manage, because if power station A is putting out 120V and power station B is putting out 110V, the customers won't know the difference...unless there turns out to be one crucial wire somewhere that has to bear the brunt of the difference in voltage. The reduction in voltage will usually be widespread enough that it causes no serious difficulties.
...but I can't think of a reason for them to run the voltage higher.
* * *
"Load is mostly inductive": the interesting thing about all this is something called power factor. It's the ratio of power actually delivered to a device versus the apparent power supplied.
Most of the time we figure volts times amps equals watts used. A 100 watt light bulb draws 0.83 amps; it's a simple resistive load so its power factor is 1. (Actually? It's a smidge less just because the filament is a tiny coil of tungsten wire--there's actually some inductance there--but we can ignore that.)
A compressor motor rated at two horsepower? Ehh...says here it draws about 12 amps...but depending on the motor's power factor it might draw less, or more, than that while it's running and pumping up your bike tire.
See: if it's part of a circuit which balances the inductance and capacitance correctly, the motor will run at its highest efficiency because the power being supplied to it (about 1,400 watts) will go into turning the compressor. (Since power factor is a measure of how much power is consumed versus how much is actually usefully applied, it's possible to have a power factor of 1 even with a real world example.)
In certain circumstances a device can have a power factor of 0: it consumes no net power. This is hard to make happen, but there are some limited circumstances where it's possible.
Power factor correction is only really used in industrial applications, where it matters--when you're running a big electric motor it makes sense to worry about power factor. For consumer-level equipment, it doesn't make that much of a difference. I mean, a 20" box fan may have a power factor of 0.7 (or less) but it doesn't matter because the thing uses about $0.30 worth of electricity in a week. The cost of adding capacitance to that motor to correct its power factor simply is not worth it. The fan would cost $50 (instead of $15) and it would save perhaps $5 worth of electricity over its lifetime.
* * *
Really, voltage is not as critical as it once was. It used to be that you'd burn stuff out all over the place if the voltage went too high; back when everything had big heavy transformers for voltage conversion it was easy to ruin something's power supply by feeding it too much juice.
But electronics got a lot cheaper, and these days just about everything has a switching power supply. This makes them lighter and cheaper, because silicon costs less than copper, and it weighs less to boot.
The floppy drive for the C-64, the C-1541, came in a case about 18" long and 6" high. It was a half-height floppy drive and could have been contained in a much smaller box but for the huge-ass transformer Commodore needed to put in there. The transformer weighs more than the entire rest of the drive combined...and it can be replaced with a circuit board weighing about three ounces.
Copper is expensive, and to make a transformer you have to wind lots of copper wire onto an iron core--and winding that wire is expensive, too, because it takes time, and then has to be tested afterwards. On the other hand, a robot can make about five switching power supply circuits in the time it takes a person using a machine to wind one transformer. Besides, transformers are fussy about voltage and frequency, and they won't work right if fed the wrong stuff.
Switching power supplies have the advantage of not giving a rip what's being fed to them: as long as it's AC of a fairly regular frequency, and as long as its characteristics are within a certain (broad) range, it doesn't matter. That's why you can look at the charger for (say) a camcorder and see that you can basically use it anywhere in the world: 120, 240, 50 Hz, 60 Hz, it's all good!
...and of course that way you don't have to make specific power supplies for each country, so it ends up making everything less expensive to produce.
* * *
Anyway--having gotten to sleep around 6, and having only slept for three hours--well, you can guess where I'm going now.