Alternator voltage regulator teardown

This project was what started the whole idea for Cramer Power Electronics. I've had several people talk to me about issues they'd seen with parts store replacement voltage regulators, ignition modules, and other parts. They wondered if I would be able to make something better. The first step was to have a look at what's out there now, so I bought a voltage regulator from the Friendly Local Auto Parts Store, tore it down, and put it under a microscope (literally) to see how it works and what quality standards the builders used. So if you're wondering how a solid state voltage regulator works or what you're getting when you buy budget priced automotive electronics, read on!

Teardown

From the outside, it looks like a stock 1970s era Mopar voltage regulator. The casing is stamped sheet metal with a molded plastic connector and a folded metal bracket.

The back is sealed with potting compound. There are several types of material that can be called potting compound; they are usually epoxies or urethane.

As this material cuts easily with a knife, I'm going to say it is most likely urethane. I switched to a box cutter starting with a knife. I could have used something more aggressive, but didn't know if there was anything inside the potting compound that might be damaged.

Enough cutting revealed a layer of sand under the potting compound. Once I had cut around the perimeter of the potting compound, I was able to pry it out to see what was under it.

The potting compound peeled off in one piece.

Here's what's underneath. The wire on the right was caught in the potting compound and broke off the circuit board.

Brushing away the sand revealed a single layer surface mount circuit board.

I used a microscope to examine the circuit board. Since it was a single sided board, it was easy to produce a schematic.

Schematics

This is a very simple circuit, using a large power transistor and a smaller transistor to switch the larger one. The switching transistor is controlled by a Zener diode, D2 on the circuit. When the voltage rises high enough, it flows through the zener diode and turns on the smaller transistor, which switches the large one off. When the voltage is low enough, current through D2 shuts off, the large transistor switches on, and grounds the field coil. There is a capcitive coupling between the output of the power transistor and the switching transistor that limits the maximum switching speed. The reason for this is that switching off the power transistor creates a large voltage spike, which the D1 diode dumps back into the supply voltage. If the switching is too rapid, this will send too much current through D1 and can burn out D1, or even the 12 volt feed wire.

The OE 1970s Chrysler voltage regulators incorporate a temperature sensor (thermistor) to match the charging voltage to the underhood temperature. The sensor wasn't on the battery itself, but in the voltage regulator, where they considered it a reasonable proxy for battery temperature. This regulator notably does not have any temperature sensing components.

Analysis

So, first, here's what they did well with this regulator.

The soldering work on the PCB itself is first-rate. I'm guessing it was run on an automated production line using solder paste and a reflow oven. It's not impossible to hand place components like this, but getting them placed that precisely isn't easy. And the solder joints are almost certainly not hand-soldered with an iron.

The circuit itself is a very simple design with not much to go wrong and not much that can kill it. The most likely things that could damage the circuit are water intrusion or an alternator field coil that draws more current than the power transistor can handle. Unfortunately, I could not find any markings on the power transistor, so I don't know what its maximum current is. Even though there aren't any discrete components for protecting it against voltage spikes, this circuit should be able to withstand spikes to 40 to 60 volts without any problem, possibly even higher.

And here are the areas of concern.

While the PCB soldering is done very well, the wire soldering is second-rate. While the joints look functional, these have large, sloppy solder blobs. The conformal coating job was also done poorly, leaving many areas of the board uncovered.

Then there's the matter of the sand filling. Sand filled electronics have valid applications where there's a concern about a box getting filled with explosive gas, such as natural gas pipeline stations, but that doesn't apply here. While this box didn't show any signs it would leak, if you made a mistake like leaving a big greasy fingerprint around the area where the urethane fills the box, it's game over. On the upside, it does make teardowns easier - perhaps their warranty department insisted on this. It's not an unacceptable way to build electronics, but it makes for a somewhat messy factory environment (the sand will get everywhere in the area where you pot the parts) and has a higher risk of failure.

As noted before, there's no temperature compensation. It runs at a fixed voltage.

And then there's the issue of tolerances. The resistors are cheap 5% tolerance pieces. Zener diode voltage tolerances are also around 5% unless you spring for something special. Theoretically, you could stack the tolerances so the voltage is off by 15%. That is unlikely to happen in practice as it's unlikely that the tolerances will all be off the same amount in the same direction. Most of these are probably going to be within 5% of their target voltage.