Brake and battery lights on

Never disconnect the battery on a modern vehicle if the engine is running, it's a really fast way to destroy the regulator circuits which are commonly in the power-train computer.

From the description I would say you have some bad cables, either from the battery to the alternator or the battery cables themselves. They will corrode inside the insulation, or right at the terminals. I would start by removing and cleaning the cables at both ends, then have the alternator tested on the vehicle as well as having the battery tested. One or both could be bad.

Do not disconnect any cable with the engine running!

Actually, you do not know anything about your charging system. It could be fine or it could be out, but your tests will not tell you. Every year I did a demonstration on the generator test bench for my students to show what can happen when you do that. It was real easy for the voltage to reach over thirty five volts. That will destroy any computer on the vehicle, the generator's internal diodes and built-in voltage regulator, and any light bulbs that are turned on.

The thinking is that if you disconnect either cable and the engine stays running, the generator must be working but a lot of them will stop working due to the voltage regulator responding to the dips in the "ripple" voltage being produced. That will make a perfectly good generator appear to be bad so that test is not valid.

If a mechanic is caught pulling this stunt he will typically get one verbal warning. For the second offense he will be fired. It is that big a deal.

Some generators respond to the high points in the ripple. That momentary higher voltage goes right back to the field winding and creates a stronger magnetic field. That stronger electromagnet creates a higher output voltage which again creates a stronger electromagnet. It is a vicious circle and voltage can keep on rising until something gives out. The main thing that smooths out that ripple so it does not affect the voltage regulator or the generator is the battery.

Three things are needed to generate the output current. They are a magnet, (electromagnet, in this case), a coil of wire, and most importantly, movement between them. That is why the belt needs to make it spin. One thing that can save you from doing damage by removing a battery cable is not raising engine speed. Generators are relatively inefficient at low engine speeds and their output voltage is less likely to rise to dangerous levels, as long as you don't raise engine speed.

One other thing to keep in mind is batteries give off explosive hydrogen gas. Regardless if your generator is working or not there is going to be a big spark when you remove a battery cable with the engine running. Either the generator's current will be recharging the battery, and that can be up to 20 amps, or the battery is going to be supplying the car's electrical systems, and that can easily be over 30 amps. That kind of current is going to create a big spark when a connection is broken or reconnected. Small arc welders run as low as 40 - 60 amps and look at the sparks they create. The reason we do not hear about more battery explosions is because people are careful to not disconnect the cables when there is current flowing through them. It is also why there are huge warning labels on all battery chargers to be sure they are turned off before connecting or disconnecting them from the battery.

Another common generator problem is one defective diode out of the six. You will lose exactly two thirds of the generator's capacity but system voltage will remain normal or it could even be just a little high from the voltage regulator responding to the greatly increased dips in the ripple voltage.

It is always a good idea to wear safety glasses when working around car batteries, but if you still insist on removing a cable while the engine is running, a face shield makes more sense, and have plenty of water on hand to wash any acid off the vehicle's paint.

Ford used to have a really nice generator design that allowed testing right on the back of the unit. Only Chrysler alternators are easier to diagnose. Unfortunately the engineers do not really care about ease of service on GMs and many other brands.

The way you tell if the charging system is working is to measure the battery voltage while the engine is running. It must be between 13.75 and 14.75 volts. There still could be a bad diode though. You need a professional load tester to test for that. Ripple will be very high and the most output current you will get will be one third of the generator's design value. That is not enough to meet the demands of the electrical system under all conditions so the battery will have to make up the difference, until it runs down. You still could have a generator problem. Have your mechanic perform a load test on the system. That will tell us where to go next.

You described two different conditions, so I would start by checking for a fuse that blew in between those two conditions. In this case, if you find a blown fuse, I would not be concerned there's something shorted. If you could look at normal generator output, it is a three-phase output with each phase going up to around 18 volts, then down to 0 volts. The battery smooths that out, then we see it as something between 13.75 to 14.75 volts. Without the battery in the circuit, you will have the 18-volt peaks on those output wave forms, and that higher voltage, (electrical pressure), means higher current flow, and that can cause a fuse to blow when there is no real defect in that circuit.

You are right that you should see the "battery" light turn on when you turn on the ignition switch. On most brands of cars other than Chrysler products, the "battery" light circuit is the turn-on circuit that tells the voltage regulator to turn on and start running the generator. Do not assume yet your generator is defective. If it is not getting the turn-on signal, it will appear as though it is not working.

On a lot of cars, the "brake" warning light also gets twelve volts applied through the ignition switch, so it may be on the same circuit as the "battery" light. There needs to be a way to provide a bulb-check, but the bulb has to turn off after that self-test. It is common to tie that into the generator too since it is going to start working when the engine starts. The working charging system turns the "battery" warning light off, and it can be used to turn the "brake" warning light off too.

I had a major house fire a few years ago. I do not have internet access, so I have to drive into town to use someone's wireless system. Sometimes I do that every other night, but I will make it a point to be here tomorrow night to see how you are doing. Once any blown fuses are handled, use an inexpensive digital voltmeter to measure the battery's voltage while the engine is running. If you find it's around 12.6 volts or less, the charging system is not working. If you find between 13.75 and 14.75 volts, that only means it is okay to perform the rest of the tests, but that requires a professional load tester. Your mechanic will test the full-load output current, and "ripple" voltage.

The best way I can describe ripple voltage is to imagine a three-piston water pump that is filling the municipal water tower. Up in the tower all you see is the water level is rising very slowly. Down at the pump, you get a strong pulse of water flow, (and pressure), a gap, another pulse, a gap, and so on. If you could graph the pressure coming out of the pump, you'd see pulses of pressure. The pressure would never go to 0 pounds because when one piston was done pushing out water, another one would just be reaching its maximum pressure. One piston is always supplying some pressure. Ripple would be seen as the difference between when one piston was at maximum pressure, and when they all were collectively at the lowest pressure.

The same thing happens in the AC generator. (To use correct terminology, we use the term "generator". "Alternator" was developed for cars in 1960 by Chrysler, and they copyrighted the term, but everyone knows what we mean when we call it an alternator). Up to around 18 volts is developed in one phase, or winding, of the output circuit. The resulting current flow is directed through a pair of diodes, (one-way valves for electrical current flow), then out to the battery and the rest of the electrical system. As the voltage begins to go back down to 0 volts, the second phase is coming up towards 18 volts, but it is not there yet. And the third phase is right behind it. You may never see the output voltage drop lower than, oh, say 11 volts, because there is always one phase coming up on its maximum output voltage.

The professional load tester is able to see that 11 volts and the 18 volts, then it calculates the ripple voltage as the difference between them. There are a few load testers that provide a printout and they actually tell the ripple voltage, but so you do not think you're being cheated, most testers only have a bar graph that shows relative ripple voltage as "low", "high", or something in between, then it is up to the mechanic to interpret the results. Only two things will cause ripple voltage to be high. One is to remove a battery cable. The battery smooths out the effects of pulsing voltage just like the water tower smooths out the pulses of water flow. The second cause is a failed diode, and that is not uncommon.

To continue my wondrous story about the three phases, the spinning electromagnet has north and south alternating poles, (you will not be tested on this later), and that means the current in the output windings changes direction continuously and repeatedly. If it was not for the diodes, the current would go in and out of the battery hundreds of times per second. That would not charge the battery, and it would severely overheat very quickly. After one phase reached its 18 volts, then went down to 0 volts, the polarity of the magnetic field changes, and that phase develops negative 18 volts. It is the diodes that redirect current flow so it is always going out toward the battery. Two diodes conduct current during the positive half of the cycle, and a different two conduct during the negative half of the cycle. There are only a minimum of six diodes, so they are shared by the phases at different times.

The whole reason for sharing this great story is it is fairly common to have one failed diode. Typically they short, then the high current causes them to burn open, similar to a blown fuse. To add to the excitement, the diodes are always in two groups of three diodes. If any one diode in each group were to short at the same time, you'd have a direct short to ground, and a real lot of smoke. Today most cars have a very large fuse bolted into the under-hood fuse box for that circuit. Older cars used fusible links that were just short sections of a smaller diameter wire that was the weak link in the chain. Its insulation was designed to not melt or burn.

When one diode of the six fails, since it was used by all three phases, it affects the output current that can be generated from each one. The most you will be able to get during the brief full-load output current test is exactly one third of the generator's design value. It is physically impossible for a generator to develop more than it was designed to develop. With one failed diode in the common 90-amp generator, all you will be able to get is 30 amps, and that is not enough to run the entire electrical system under all conditions. The battery will have to make up the difference as it slowly runs down over days or weeks. Or, here is the reason for sharing all this, many voltage regulators see that missing phase, and turn on the "battery" warning light in response. Output voltage might even be a fuzz high due to the regulator pushing the generator to work harder when it sees that drop in voltage. It sees the drop; you see correct or slightly high battery voltage. When the voltage regulator is built into the generator, it sees the voltage there, similar to you seeing the water pressure at the pump. But you are measuring voltage at the battery which equates to measuring the pressure in the water tower.

The point of all this is to explain why the engine could stall even though the charging system is working fine, and why the generator could have a bad diode and the engine still runs with the battery disconnected. That is why you cannot draw any conclusions from disconnecting the battery. Many decades ago that was a common trick with the old DC generators, but those did not use diodes, and they were real inefficient at low speeds. In fact, at idle, they typically did not develop any output current. It was common to have to use a portable charger to recharge the battery once or twice a month. They also all used separate mechanical voltage regulators that could totally stop output current from being produced if voltage went too high. Add to that there were no sensitive electronics on those cars. When alternators showed up in the 1960's, mechanics kept doing that trick, not understanding why the test was not valid.

To sum this up, check the fuses, then measure the battery voltage with the engine running. If the voltage is okay at 13.75 to 14.75 volts, but the warning lights are still on, have the professional load test performed to check for a bad diode.