Hi brother CJ. I know they'd rather I didn't do this, but I feel compelled to add another chapter because I suspect there's something else going on here. The "Battery" light can turn on for a few different reasons other than a no-charge condition. If this were a true no-charge problem, the battery would run dead after less than an hour of driving, and much faster on GMs with their daytime running lights that turn themselves on. The fact the battery holds up for a few days is typical of an under-charge due to a failed diode.
With one failed diode of the six, you will lose exactly two thirds of the generator's output current capacity. Psharkdizzle didn't bother to list the engine size or transmission type so I can't look up what is normally in there, but for the common 90 amp generator, 30 amps is not nearly enough to meet the demands of the entire electrical system under all conditions. The electric fuel pump, head lights, ignition system, and radio alone can draw over 25 amps. That 30 amps you might be able to get is only at a higher engine speed. At idle you'll be lucky to get 10 amps. Use the brake lights and power windows a few times, or use the heater fan, and there's no way a generator with a bad diode can keep up. The battery has to make up the difference until it slowly runs down over days or weeks.
The misleading part of this is the charging voltage as measured at the battery with the engine running, as you correctly recommended, will usually appear to be just fine, within that acceptable range of 13.75 to 14.75 volts, (a little more or less depending on which text book you look at). What that voltage test is good for is to see if you have something or nothing, but the second part of the test requires a professional load tester. If battery voltage stays at 12.6 volts or less with the engine running, there is no current coming from the generator at all. If the voltage is in the acceptable range, we only know there is some output from the generator. That's when the load tester is needed. That will test for full-load output current and "ripple" voltage. If one of the diodes has failed, ripple voltage will be high, and, as I mentioned previously, the maximum current will be exactly one third of what it's supposed to be.
Psharkdizzle, there's no way to suger-coat this. In my opinion, GM had the world's second best generator design through the 1986 model year, but for 1987 they redesigned it, and it's one of the worst designs ever. Due to their design, they develop huge, harmful voltage spikes that can damage those internal diodes and voltage regulator, and interfere with computer sensor signals. This is where the test you had done on your battery is invalid for this symptom. The battery is the key component in damping and absorbing those voltage spikes, but as they age and the lead flakes off the plates, they lose their ability to do that. They will still crank the engine just fine so they'll work in an '86 or older car. It is real common to go through four to six replacement generators in the life of a GM car, but to reduce that number of repeat failures, replace the battery at the same time, unless it's less than about two years old, even if it's cranking the engine normally. If you do not replace it, expect the replacement generator to fail again anywhere from five minutes to five months. This is such a common problem, I copy and paste a standard reply rather than retyping it every day. People post your same symptom here over and over.
Also, to clear up any potential misunderstanding, the "Battery" light really has nothing to do with the battery, at least not directly. That light should say "Charging System". It means the generator is not recharging the battery as you're driving. As I mentioned, with a dead charging system, a good, fully-charged battery will have a hard time running the electrical system for a full hour. Since yours lasts much longer than that before you have to charge it again, you can extend your driving time by turning off everything electrical you don't need. Have that professional load test performed, then let us know what was found for full-load current.
Monday, October 5th, 2015 AT 9:14 PM