Dynamo Current and Voltage Regulator

Many owners have long ago replaced their dynamos by alternators, but in some cases that's not an attractive proposition: Antique cars loose originality if the dynamo is taken away. But my friend Germán had a more peculiar problem: He bought a sailing yacht which has an old Volvo Penta Diesel auxiliary engine, and this engine comes with a Dynastarter! This is a combination starter motor and dynamo. It's basically a 12 Volt compound motor that is used as such for starting, and as parallel-excited DC generator when the engine is running. It's coupled to the engine with two parallel V-belts. For starting it gets its current through a large relay, and as a generator it used the traditional setup with a voltage regulating solenoid, a current regulating solenoid, and a reverse current relay - a system that is very failure prone and has burned out the dynastarter several times.

So Germán wanted a fix. Either something that makes the dynastarter reliable, or else a complete change to a separate starter and modern alternator. Certainly the latter is the better solution, but it's hard to implement, because this engine doesn't lend itself to the easy installation of a starter gear. After I went with him on a nice sailing trip in winter 2004, and with the promise of another trip to the San Rafael Glacier, I couldn't do less than design and build a tailor-made electronic regulator for his dynastarter, in a sort of last attempt to avoid the large modification work on the engine!

Since this regulator will work just as well with a standard dynamo, and not finding any similar design anywhere in the web, I decided to put this project on my web site immediately after completing it. Certainly people out there are looking for such a regulator! So, here it is, for all you who are locked into using old electromechanical technology, but not afraid of helping it with some modern electronics!

Secondly, dynamos are not self-limiting, while alternators are. When an alternator runs faster, higher voltage is induced, but also the frequency of the AC is higher. Since there is a large inductance in the windings, the reactance increases with frequency, the two effects roughly compensate, such that a correctly designed alternator can never deliver a dangerous current! It limits its output current to a safe value, regardless of RPM. Dynamos instead just don't have this nice feature. As long as there is enough mechanical drive power, they will deliver whatever current a discharged battery can take, and will easily self-destroy if the current isn't limited externally.

Voltage regulation, on the other hand, can basically be implemented in the same way for dynamos and alternators. So, a dynamo regulator is basically an alternator regulator with additional current regulation and reverse current blocking.

So far, so good. Let's go to the details.

In the typical dynamo charging circuit, B+ and B- are the battery connections. D+ and D- go to the dynamo brushes, while DF is the field connection, with its other end returned to D+ inside the dynamo. Please note that not all dynamos return the field winding to the positive output! My regulator circuit, as shown here, works only with dynamos connected in this way. For further clarity, here's the internal wiring of a standard dynamo. A dynastarter is just like this but with an additional field winding added, connected from D+ to an additional terminal, which is used to apply battery current for starting.

In the regulator schematic I have provided B- and D- connection points, but of course the two are the same thing: System ground. On the real life implementation I just named all ground connections D-, following the standard used on the boat for which the regulator was built.
 

You can print out this schematic, so you don't have to scroll up and down the page while reading the explanation. Clicking on the schematic will bring up a 300dpi version for high quality printing (as long as your browser can handle the file size... Otherwise right-click, save to disk, and print from there).


Let's start analyzing the circuit in idle condition, when the engine is stopped. The battery voltage at B+ cannot pass to D+, because D1 blocks it. D+ is at ground level, because of the low resistance of the dynamo between D+ and D-. Q1 is biased off, which leaves the negative side of the entire OpAmp circuit floating. So the complete control circuit will float at +12V, and in this way will bias Q4 fully on. Q4 then holds DF down to ground, ready for dynamo startup, and for using the dynastarter as a starter.

There will be no power consumption other than the very low leakage of some components. Q1 is off, D1 and D5 are reverse biased, Q3 is off, D7 stays far below its conduction voltage, Q4 has an insulated gate, C1 and C8 supposedly don't conduct DC... Actually, the leakage in C8 is the main current drain, but this leakage gets down into the microampere range when the circuit is connected permanently to the battery. So, this circuit consumes essentially no idling current, while keeping DF grounded for easy dynamo startup.

When the dynamo starts rotating, it will self-excite through residual magnetism. The very low voltage initially generated (typically 0.2V) causes a small current to flow in the field winding, through Q4. This increases field strength, the generated voltage goes up, and soon we have a working dynamo. As soon as the voltage gets to about 6V, Q1 will switch on and pull the negative side of the OpAmp circuit to ground. The yellow LED will light, and the whole regulator circuit gets alive.

Let's see what happens when the dynamo is putting out about 12V. Q1 is on, the yellow LED is on, but the voltage is not high enough to start charging the battery, which might be near 12.3V. Pin 13 of the current comparator U1D will be at roughly 10.4V, while pin 12 will be about 0.15V lower. Thus, pin 14 will be essentially at zero, and the red LED will be off. Pin 2 of  the voltage comparator U1A will be at 4.25V, so that D4 is reverse biased. D5 too is reverse biased, since its anode is at 3.9V. Pin 3 of U1A is at a regulated 5V. So, pin 1 is at a high positive voltage, close to the battery voltage, and the green LED is off.

U2C is an oscillator that runs at roughly 100Hz. It produces an approximate triangle wave across its timing capacitor C5, which is used as a reference for the pulse width modulator U1B. Its pin 6 will constantly vary between one third and two thirds of the battery voltage. Because pin 5 is held up by U1A, far above 2/3 of the battery voltage, pin 7 will be high too, and through Q2, Q4 will be kept on. So the dynamo can run at full field current.

Let's assume that now the operator throttles up the engine. An unregulated dynamo would now far exceed its current rating, and burn out if the condition lasts long enough. But in this circuit, as soon as the voltage drop on R1 reaches 0.15V, current regulation sets in. Let's assume that the dynamo is charging 20A, with the battery voltage at 13V, and see how this condition is maintained:

The voltages at pins 12 and 13 will be almost exactly equal. What little difference appear between the battery voltage and the junction of R2 with R5 will be amplified by a factor of 1000 by U1D. So, a little variation of the current around the 20A set point causes a rather large variation of the voltage at pin 14. When this voltage is above roughly 5.5V, it will make D4 conduct, and raise the voltage at pin 2 of U1A. The red LED will be on.

When pin 2 reaches  the 5V present on pin 3, pin 1 will start going down, at a slew rate set by C4, the basic timing capacitor of the control loop. The larger the current excess, the faster the control loop will act. The voltage on pin 1 will end up somewhere between 1/3 and 2/3 of the battery voltage, so the green LED will be on, and U1B will produce a square wave at pin 7, its duty cycle being a function of the voltage at pin 1. R19 and R20 give U1B Schmitt trigger action, so that the switching will be as fast as possible, and very clean. The square wave will drive Q4 quickly on and off, thanks to the emitter follower formed by Q2 and Q3.

While Q4 is on, the full dynamo voltage is applied to the field winding, and its current increases in time, according to the winding's inductance. When Q4 switches off, the inductance of the field winding keeps field current flowing, with the circuit being closed through the freewheeling diode D8. The current will decrease in time, until Q4 switches on again. So, the voltage at DF is a square wave, while the current through the field winding is essentially DC with a small triangle wave component mounted on top.

Any variation in RPM will cause a tendency to change the current, which will make the circuit react, adjusting the duty cycle of the field winding drive to keep the output current constant.

As the battery is charged, its voltage will slowly rise, until reaching about 13.9V. At this level, the voltage at pin 2 of U1A reaches 5V, without needing any help from the current sensor circuit. The control loop will then control the field duty cycle to maintain a constant 13.9V, while the current will drop, making pin 14 go down and the red LED go off. The green LED will remain lit.

As long as the dynamo remains rotating fast enough, the circuit will keep regulating the output to 13.9V or to 20A, depending on load demands. It will never allow any of these two values to be exceeded. When the RPM of the dynamo are too low to maintain regulation, the circuit will just let it generate at full field current.

When the engine is shut down, D+ drops to ground, Q1 switches off, the entire circuit powers down, and remains off until the next use, with Q4 appropriately biased on.

Various notes

The fuse was included mainly as a safety measure against failure of D1. This diode gets pretty warm, and could conceivably fail if it comes loose from the heatsink. If it fails shorted, F1 would avoid the destruction of the dynamo through reverse current. Anyway the risk would be low, because a failure like this would most likely end up with D1 opening rather than shorting, but fuses are cheap, so it's not bad to include one...

If F1 is open, the dynamo voltage can no longer be regulated by sensing battery voltage, and also there would be no current to sense. In order to keep the dynamo voltage from soaring, D5 will inject some current to pin 2 of U1A, and so the circuit will now regulate the dynamo voltage, not battery voltage, to about 16.5V, keeping matters safe until someone replaces the fuse - hopefully before running the battery down!

D18 and D6 are there only to protect the MOSFET gates against abnormally high voltage, which could happen when the operator does stupid things such as starting the system without a battery connected. Otherwise they would not be needed.

C6 and R23 form a snubber which eats up any transients that could happen during the MOSFET switching. Most likely this snubber isn't needed, thanks to the high switching speed of D8 combined with the comparatively slow switching of the MOSFET, which is limited by the slew rate of the TL084.

C8 might not be needed when the battery wiring is short and direct, but for safety I suggest to include it in all cases. I added it during the tests of the prototype, after having the system break into self-oscillation at 300kHz due to battery cable inductance resonating with C1! This event was useful to see how fast Q4 can really switch! It produced quite a clean 300kHz square wave... Needless to say, switching at a nominal 100Hz, its switching losses are essentially nil. Maybe the emitter follower would not be necessary at all, but surely it adds a bit to efficiency of the regulator.

The use of Schottky diodes at D1 and D8 was decided because of their low voltage drop. This means less heat (mostly for D1), and heat is the worst enemy of electronics, as you know. Standard diodes could be substituted, but I don't recommend it.

U2 is not a Zener diode, but a precision voltage reference IC. To answer a very common question I often get: No, you should not replace it by a standard Zener diode. Zeners are not precise enough. This circuit avoids the use of potentiometers for voltage setting, because they are unreliable, particularly in a marine environment. It relies on precision components to get the correct voltage, and this won't work with Zeners. If you can't find an LM336Z-5.0, you can use some other 5V reference IC (also called "reference diode", even if it is much more than a diode). And no, you cannot use a 7805 or any other three-terminal regulator, because they would mess up the correct startup of the dynamo: While Q1 is halfways on, a three terminal regulator would produce a lower voltage on pin 3 than pin 2 of U1A, which would make the circuit switch off Q4 and interfere with proper dynamo startup.

Feel free to replace components. I used the IRF540, BS170, 30CT060, SB360, 2N4124 and 2N4403, etc, basically because I had them. Any parts with compatible ratings should work as well. Don't replace the TL084, unless you know very well what you are doing. It was chosen because it handles high input voltages well, and has a high slew rate. Many other OpAmps wouldn't work well here, and the TL084 is easy to find and inexpensive.

If you absolutely want to include a potentiometer for voltage setting, you can make part of R12 adjustable. For example, use a 10k trimpot in series with a 22k fixed resistor. Likewise, if you want an adjustable current limit, you can replace R2 by a 200 Ohm trimpot. But a safer way to trim the voltage and current setpoints is by soldering in some series or parallel resistors picked to give the desired levels. I did not go to this length, and just mounted fixed precision resistors.

R1 in my prototype is a parallel combination of two 0.015 Ohm, 5W resistors. If you can't find resistors low enough to make R1, you can use my old fence wire trick: Galvanized iron wire is cheap, easy to find, easily solderable, and has a resistance which is very convenient for making very low value resistors!  I have some galvanized iron wire here which has 1.25mm diameter, which would equate to roughly AWG 16. This wire measures 0.12 Ohm per meter at room temperature, which is very close (slightly higher) to the calculated value considering the standard volume resistivity of pure iron. The only trouble is that this wire is rather strongly temperature dependent. So one has to design the resistor in such a way that the wire won't heat up too much. For this resistor, two pieces of this wire in parallel, each  about 12cm long, would be fine.

Whatever you use, factory made resistors or iron wire, be sure to connect the sensing points right to the resistor. Even a small piece of PCB trace or wiring can add a considerable proportion to this small resistance, and your regulator would end up regulating to a much lower current than desired!

If you want to use this circuit for a dynamo of a different current rating, you can modify the value of R1 so that at the desired maximum current it will cause 0.15V of drop. D1 and F1 should be adjusted to suitable values. The rest of the circuit can most likely remain the same.

If you want to use it for a 6V or 24V dynamo, this would require a significantly different circuit. You can't properly drive this MOSFET with 6V minus the voltage drop in the driver, and if you drive it with 24V, you might burn it out.

And if your dynamo happens to have one end of the field winding connected to the negative side, then too you can't use this circuit directly. You would need to modify the output section, providing a P-channel switch with proper driving.

The section of green PCB was cut from the original solar regulator's PCB. It has the proper mounting holes and connection pads for the terminal strip, so I used it to install the power components. The control circuit was built on a piece of perfboard. D1 was mounted to the case using a very strong clamp, a thin mica insulator, thermal compound and two bolts. Since it generates a fair amount of heat and is used at 2/3 of its maximum rating continuously, it needs good cooling to remain reliable. Q4 instead is mounted with a nylon bolt, mica insulator and thermal compound. It works quite cool.

The fuse holder was already mounted, so I re-used it. There were some holes in the box for switches, pots and LEDs, three of which I re-used to mount my LEDs, even if their position wasn't very logical...

The whole circuit was sprayed with several coats of acrylic lacquer before mounting it in the box, and then several more coats after mounting. This creates a tough plastic-like coating that should prevent corrosion and conductive film forming for many years, even in the harsh maritime environment where this regulator will be used.

If you can understand Spanish, you might want to read the slightly humorous Instruction Manual which I gave the Captain of the yacht for which this regulator was built.

Updates