w_tom said:
Wow. So many grossly inaccurate responses in one thread.
The classic: disconnecting a battery while car is running. Ted comes close to explaining that this is wrong - but provides incorrect reasoning.
"...and massive surge caused by pulling the battery out..."
Is what I said and was exactly what you said here.
That disconnect recommendation
could create load dump - a voltage spike of between 50 and 200 volts on 12 volt electronics - depending on which auto industry standard is consulted.
I think you assumed the rest of my post was an attempt to explain why a "load dump" AKA surge, occurs. It wasn't, I was attempting to explain why surges damage semiconductors a whole lot faster than "normal" usage. You should also note that Bill Putney already listed a massive surge yesterday as a problem, even used the same terminology you did. (load dump)
Many people have this incorrect idea that semiconductors are the same as fuses, this is wrong. A fuse is designed to burn through almost instantainously. Semiconductors by contrast can stand short overvoltages - but those will shorten their lives.
Simple with proper tools and basic knowledge. But even Ted provided defective information. Semiconductors do not wear out. Even the idea that power cycling light bulbs causes premature failure - a myth not supported by any technical facts. Light bulb life expectancy is predominately due to voltage applied and hours of operation - or mechanical shock when powered.
This is absolutely, completely, totally wrong. Entropy occurs in everything. First of all, semiconductors DO wear out. It may take hundreds of years if the semiconductor is operated in a tremendously derated fashion as most are, but they will wear out. There is a physical reason for this, it is because as electrons move through the semiconductor, they generate tiny amounts of heat. Over a long period of time the heat will cause random moves of molecules across the semiconductor joint. Once enough molecules move, the joint is ruined and that is that.
This is why if you remove a heatsink from a high power modern computer CPU and run it flat out until it stops functioning, it will be permanently ruined, it will never be stable again. The joints there are so small that once enough heat builds up, it takes little time for the material to be permanently altered. People talk about "burning out" semiconductors (like diodes in alternators) but this is in fact sloppy slang, burning is an oxidization process, and yes some semiconductors can be ruined this way, but most "burned out" semiconductors don't burn, they suffer massive material changes due to excess heat caused by overvoltage, or simply because of too much electricity moving through the material.
TO repeat, in most electronics, the process takes many years because the semiconductor joint is so gigantic that there are trillions of molecules of material there, and molecular migration happens to such a small number of molecules that usually the device built around the semiconductor will have long since disintegrated before the semiconductor is seriously degraded.
But in many cases, such as overvoltage, overheating, and otherwise operating a device out of it's specs, the process is drastically speeded up.
And this process also crops up as a problem if high precision is involved. This is why radio transmitters must be periodically retuned, and why high precision test instruments must be recertified, it is because their use over time will slightly alter their readings, as the semiconductors in them wear. High precision electronic test instruments all have recalibration controls (well at least good quality ones do) that allow them to be recalibrated to a standard.
And as far as the lightbulb thing goes, yes it is true that the hours of operation of a light bulb filament are more important to it's lifespan than the number of on-off transitions. You are right in that this example of mine wasn't really accurate. Put it down to trying to shorten a more complicated explanation, which as a nitpicker you obviously are demanding, so I'll spell it out here.
Light bulb filaments do wear out, for the same reasons semiconductors wear out - molecular migration of the tungsten metal (or whatever metal is used in the filament) from the filament to elsewhere. Inside a light bulb of course, that "elsewhere" is the inside glass wall of the bulb itself. In a light bulb the process is of course much faster because so much more heat is produced that the molecules are all moving around a lot faster.
What I was trying to get at, though, is that they typically blow when turned on due to the surge of current that happens then. The reason there is a surge then is that when the filament is cold, it's resistance is lower, thus more current with the same applied voltage.
Obviously if you had a lightbulb on 24 hours a day, 365 days a year, without any surges or overvoltages going through it, it's filament would last longer than one that is turned off and back on a few times a day, and tests will bear this out. But eventually enough material will migrate that the filament will lose structural integrity, sag, and break under it's own weight.
experience. That disconnecting battery is a classic example of an irresponsible post. Poster probably never heard of load dump - and the potential destruction of all automotive electronics.
It's also a classic example of someone figuring out a shortcut that is destructive to the device in question, but not in a way that is immediately obvious to the customer.
It is much the same thing as when people doing body work on a car don't bang out the dent as far as it can be banged out, and instead mold and build up huge chunks of Bondo. It's the same thing as the mechanic that loses a screw because of piss poor fastner control, and figured "they won't see that it's missing and there's lots other screws around it" which is one of my pet peeves.
Ted
