> any simple passive overvoltage protection circuit that can go after the fuse and before the transformer?
i am guessing it could be a simple as a hi-watt low-ohm resistor in series with the 120 and 2 oppositely oriented zeners of appropriate voltage rating shunting to neutral...?
When over-volted, you'd get flat-top waves. That would keep simple DC supplies in check, but lamps would still burn over-bright.
Main problem: small overvoltage gives big dissipation, significant overvoltage gives HUGE dissipation.
OK, say that instead of getting 108V which should be 117V, I got the opposite problem: 126V. And while I should design for the full 20A of the circuit, I may only have 10A online normally. Hmmmmm. I size the resistor for a 9V drop, 9V/10A= 0.9 ohms 90 Watts.
And ASSume it is DC, to avoid sinish complication. And I set a 120V Zener across the line. At 126V dropped to 117V, the Zener sits cold.
But line voltage varies. Backhoes eat underground feeders for lunch. The patch-work is undersize so they tap-up to compensate. The sag varies with load. Say almost 10%, to 138V.
Now the 120V Zener conducts. The 0.9 ohm resistor feels 138V-120V= 18V. It passes current of 18V/0.9= 20A. Say my load still sucks 10A. The other 20A-10A= 10A must flow through the Zener. The Zener dissipates 120V*10A= 1,200 Watts. This is as much as my actual load eats. My electric bill doubles. Another 10% voltage, my Zener runs 3,600W and my bill quadruples (except the fuse should blow). If my load happened to all be off, the Zener is taking 4,800W, maybe for no good reason.
Shunt regulation of varying supply voltage is horribly lossy.
Actively switched resistors work, for overvoltage, with loss proportional to how much you are throwing away. For 10A at 18V over-voltage, 180W, which could be acceptable. However Murphy's Law says that if you design for 20% overvolt, the utility will give you 40% over. If you get into a bad neutral, your dropper may have to be as big as your load.
The classic way to handle large power is an autotransformer. For 1,200W load and 10% over-voltage, we only need a 120VA core. A multi-tapped winding can give 2% increments over any reasonable range. A voltage detector and motor can auto-adjust. Losses will generally be small. As John found, all mechanical contraptions can get stuck. All servo systems can oscillate: we have such a regulator which bumps up and down 2 Volts every 4 seconds all day and all night.
Another technique is a SOLA. This is a resonant transformer with a saturated leg. If the input frequency is right-on (it usually is), then saturation limits the output to 120V for quite a wide range of input (below full power, 95V to 140V). Early ones made very distorted waves; better ones came later. They work good; they are expensive and heavy and hot and loud. They also made them with low-volt windings: I had some OLD transistor computer power supplies where the "5.2VDC" was a lo-volt SOLA, two diodes, and a bunch of BIG caps. Line variations not a problem. Load could sag, but in those days the logic load was mighty constant (none of this modern dynamic power or SpeedStep junk) and not very fussy (5V, 6V, all the same to the logic).
I just picked up a box which both isolates 800VA and steps 12V: from 90V to 140V in gives 112V to 125 out. It uses power transistors to change taps. These don't stick, and can switch at zero-cross, but there are a dozen ICs controlling it and I'm sure they can do the wrong thing. I'm rather tempted to put a big mechanical switch on the taps: I don't have a constantly changing voltage, I have steady low voltage. It matters most on the test bench, where I can check my actual "line" voltage while reading audio power.
A newer way is to use an "audio" amplifier. One made to give many Amps and a few dozen volts of 60Hz. Put the output in series with the line. By adjusting the voltage and phase, it can buck or boost the output, potentially with milliVolt precision (a near-Zero impedance "wall outlet"). Done in linear fashion, it's awful inefficient. At 60Hz, switcher techniques are easy, and many "UPSes" use this method to compensate small line variations.
And that's your answer: use a good UPS. APS brand uses "Line Interactive" to indicate that it actively boost/bucks the line. Every UPS maker has this feature, but not on the $69 junk, and it can be hard to suss out from the glossy sales literature how far up the product line you have to go to get it.
As an extreme, add energy storage. Not just a cut-over for blackout, a full time online battery. Charge the battery from the wall. Invert battery potential back to AC. A large change in wall voltage makes a small change in battery voltage because it stores energy. Another trick is a motor, a flywheel, and an alternator. Motor can't change speed quickly because of the flywheel. (Anyway most AC motors are not much affected by voltage within reason: they turn at line frequency.) The alternator delivers clean steady AC, and its field regulator can fine-trim the voltage. But the cost of such schemes is generally way-too-much.
> trusty old Simpson 260, probably fair.
Great tool. Plenty accurate. Darn near indestructable. But as mentioned, going up and down poles is a tough life. It is a "trust but verify" situation.
