How to Make a Reliable Motorcycle Voltage Regulator

This article is made due to many queries after the post about DeadBug Prototyping, so I've decided to make a whole instructable to explain how the motorcycle regulator relay works and how to make it.

Disclaimer:

1. Some schemes where taken from the basic author of the circuit, Skrut, he developed the basics of this schematic. I made my vision of it and chose my details. And wrote this article.
2. Some of text parts are alike to DIY forums due to that it was discussed by me and/or other people in the process of making this project.
3. The first part of the article is theoretical explanation for people who have little knowledge in electronics, so it is replete with not quite correct comparison and simplifications. Please do not poke me in the face with electrical engineering textbook to teach me Kirchhoff's law. Anyway, in other case you can skip it or move to Step 3.
4. English is not my native, so I ask you to treat any mistakes indulgently.

If you are OK with listed above, then you are welcome for further reading.

It all started with the my friend, who have asked me to resolve the problem with Voltage Regulator on Honda X4 urgently. But! This schematic will perfectly work on any motorcycle with three-phase generator, for instance CBR* and others. To know if it fits you have to look at the generic regulator, if it has one or two red, one or two black and three "other-colored" wires (pink, brown etc.), then it's likely our case. In Honda X4 phase wires and plus-minus wires are separated into two power connectors one has three terminals, the other - four.

So, let's begin. First, it turned out that motorcycle VR (Voltage Regulator) - it's not a car VR, who could have guessed ;)

There are two differences between them and they are very serious.

1. Auto VR is a stabilizer. Motorcycle VR is a rectifier + stabilizer.
2. Auto VR adjusts the voltage on the field winding of the generator. Motorcycle VR regulates the output voltage of the generator.

There are motorcycles with automotive-type generators, but there are only few of them. Here I'd like to digress on the subject of "what is current, voltage, and a voltage regulator." Electric current is, as we know from school physics course, a "directional movement of electrons". Let's not dig into details now, it is just important to understand the main thing - an electric current has many parameters, but we need two most important of them - current and voltage. The current is measured in amperes, and the voltage is measured in volts. To understand what it is, imagine that your wire is the channel, and the current - water flowing over it. So the current is water flow rate, and the voltage - water level in the channel. For further understanding of the text that's enough.

Now about stabilizer that we'll need in this project.

We won't bother about rectifiers - a diode is a diode (captain obvious was here). The objective of any voltage regulator - to receive a voltage and lower it to the certain desired level and hold it at that level. The operating principles of the stabilizers are divided into impulse, linear and shunt. The shunt stabilizer makes excess voltage "bypass the consumer". The simplest shunt regulator is assembled from two parts - the resistor and zener diode.

Zener diode, it is such a funny thing that when the voltage is less than needed, he (zenner diode or ZD) pretends absent (ie, allegedly ragged wire), and when the voltage is more than you need, he pretends to be a wire (ie starts to freely conduct current). Imagine valve with spring, that's the same principle. It works like this. When the voltage is less than you need, a zener diode does not conduct current, all the current goes to the consumer. Water flow is low, the valve is closed. When voltage somehow is increased and becomes more than necessary, zener diode begins to conduct, and all superfluous voltage "falls" bypassing the consumer through the zener diode to the ground. Water flow is high, the valve is opened and pouring out the excess water. Thus, our power, our "water level" all time stays approximately the same value. Everything seems be fine, but there is no zener diode for high currents. This valve can only be a small diameter. Therefore, to make a stabilizer for high-current with the use of zener diode only is impossible. How to deal with it I will explain later.

Linear stabilizer operates on the principle when the voltage is too high, it creates additional difficulties to pass through. The best comparison - the toilet water tank. If he level in the tank is small - the valve is open, the water is poured into it, if the level rises - the float pulls up, the valve is closing, the hole almost closed when water reached the correct level, the valve is closed. Flush, level dropped, the water ran out, and it all goes all over again. Same thing here, only it's very fast.

On the first picture on the charts we see blue, red and green three phases of our generator passed through the diode bridge. The phases are shifted by 1/3 half cycle relative to each other and summing the rectified three-phase regulation, we obtain the desired value as generated above in magenta.

It is worth to mention that we consider the processes occurring in only one full turn of the shaft of the generator, ie, in reality, there is a fact that the output has acidic battery as a HUGE capacity we get stable 14,5V.

This is actually the norm of life for any system of stabilization of three-phase AC voltage.

Now, crime news.

As I said earlier, the most likely cause of failure of the relay - the breakdown of any diode in the rectifier bridge. All because of one little diode, two of the three generator windings are converted into unnecessary stuff. Second picture shows a graph in the form of "boobs" - this is the only survived phase. The other two (shown in brown dotted line) - are dead for us from now.

If you looked carefully at the charts, you probably noticed that the voltage at the survived phase rises up to nearly 25V. And in a working system it is no more than 10V. This is not a mistake.

How did it happen? Very simple:

Regulator scheme immediately suspects something is not right and decides at all costs, to keep required 14.5V output, bu the way, there is a battery at the output, which perfectly smoothes surges and controls the required 14.5V. But measuring the output voltage you are unlikely to find the correct voltage.

It is worth noting that we speak of the shunt relay controller. This regulation is very different from car schemes and of course car regulators do not fit our problems. Namely, car voltage regulator regulates the current on the stimulating winding with the intent produce stable voltage on the stator. And motorcycle shunt relay controller works so: when the voltage across the winding of the stator winding is exceeded it shunts the generator winding, not giving the voltage to rise above provisions, and this is due to the fact that as the causative agent in the motorcycle a permanent magnet is used and it requires no excitation.

Do not forget that some motorcycles use "car" regulatory system. Usually, it is easy to determine if you know where generator is located. If he has a separate building - it is likely, a generator with an excitation winding at anchor. If it is located inside the engine - a high probability is that it's the armature permanent magnet, which means somewhere our friend shunt regulator is hidden.

On the picture, perhaps, the most complete regulator circuit is shown that can be found in the manual on Japanese motorcycles. Here we can see that the voltage from the three-phase alternator (AC generator), driven by a permanent magnet, is rectified through three-phase diode bridge (diode bridge highlighted in green) and it all gets straight to the battery (battery). A regulation is engaged in "secret" Integrated circuit that controls powerfull thyristors (highlighted in red) that short-cut the generator to itself through the diode bridge if the voltage exceeds limits.

In general, everything is simple, but there is a serious problem - a rectifying diode bridge is very heavily loaded and richly heated, because they are forced to overcome all the current produced by the generator through themselves. The most common failure of VR is the failure of one of the diodes, ie breakdown of any diode. In this scenario, the scheme does not cease to give tension, but no longer able to cope with its tasks.

Here is a list of the necessary details:

1. Several Zener Diodes at 14B 0,5W
2. Resistors of 300 ohms 5Watt - 4 pcs
3. Capacitor 1000pF - 1 pc
4. IC Darlington Array ULN2003A DIP16 - 1pc
5. Triac BTA26-600 - 3 pcs
6. Diode bridges 35A like KBPC3510 - 3 pcs
7. Single-core and multi-core copper wires 2-2.5 square mm
8. Epoxy adhesive. 500 ml
9. Male connectors
10. Thermal paste (highly important)
11. M3 screws - min 3pc
12. Huge or big radiator, the higher the ribs - the better it will work

Note: small parts are better to take with a stock, they cost a penny, but they are lost or broken very easily.

Note2: I couldn't find 14V Zener Diode, so I used 13V Diode like BZX55C13, it gave me 13.5-14V overall output

Note3: in bridge name "KBPC3510" 35 means 35 ampere (you can take more but it's way too overprotected 35 is more than enough) and 10 means 1000 volts, so if there is 12 it would mean 1200 volts also overprotection, peak voltages woun't exceed 300-400 V, so 1000 is the golden point. Same for the triac marking.

On the above pictures with schemes power lines are highlighted thick, it is no coincidence. When assembling, you should use a serious copper wire of 2-2.5 square mm. Thin lines show the low-voltage lines, where you can take advantage of flexible thin wires. ULN2003 is better to buy a DIP package, because it is easier to operate with it's' pins.

I gave both schemes, for 3-phase and three single-phase bridges, but I strongly recommend to make VR with three separated bridges, because it makes this VR almost ultimate.

First of all carefully prepare the layout depending on the radiator size and height. I tried 15 different layouts before I decided to cut out 5 ribs out of 13 i had basicly. The more ribs you have and the higher they are - the better clooling yiu will receive. But do not locate elements too close - in that case they will heat each other. The most heating elements are BTA26-600 and bridges.

When you are supeshure in your layout you can cut out the ribs for parts.

I gave my radiator to the miler, who cut the ribs, but the radiator needed much work after that because the cutout was very coarse.

I took the rasp and finished the surface to perfectly fine. This step is HIGHLY important, because the better is the surface the better bridges and BTAs will adjoin to the radiator and it will guarantee good heat transfer.

After that check the layout one more time and mark the holes for mounting.

Drill holes, cut the screw-thread.

Apply thermal paste, this is also VERY important. The CPU paste will do fine.

I give the photos step-by-step, so comments are unnecessary.

Nevertheless:

You need to get an appropriate wire somewhere. Refer to the photos to see the wire thickness and bridges layout.

Note how I bent BTA's and bridge's contacts.

While soldering make make good warm-up of soldering spots and use flux, this must give you good shiny mirror-like soldering and will guarantee long work of the device. If the soldering spot is matted, then you either didn't heat it enough or didn't gave enough of flux. But try not to overheat. If you polished the heat-sink and applied the thermal paste, than the heat will transfer to heat-sink, spread and everything will be ok.

LOOK VERY CAREFUL how i connected the "stabilitron" Zenner Diode, note the mark on it, it must look towards BTA bus.

Check 5 times how you located the bridges.

Check not less than two times how you located the ULN2003A IC.

Note, that I placed the resistor between 1st and 8th pin of ULN2003A under the IC, it gave me extra space for BTAs.

5, I mean five times check what wire are you soldering and where.

On these photos the wires are:

• RED - the Plus.
• Brown - the Minus.
• Yellow-Green - three brother-phases.

Those wires must be very tough and multicore, all the load processed by VR is put on them. That is why I soldered two plus wires on opposite ends of the Plus Bus and two minus wires on sides of Minus Bus. My wires are 4 sq.mm., this is approximately 3 mm diameter for multicore wire.

Check if your wiring is correct again. Better print all the schemes and pages from datasheets to be sure.

I know that I already make you sick, sorry for that, but check the wiring again before this step. I've checked my build even if I moved it from one table to another. You can never be 110% sure.

First of all, check the integrity of the diode bridges, this poor thing suffer most.

To do this:

• check it with the multimeter set to Ohms or beeper.
• check leads in both directions from phase to plus contact. One way it has to beep in the opposite - no.
• do the same with the direction of phase to minus. The result must be the same.
• Do these operations with all three phases (all three bridges that is)

If any phase-to-pole check goes both ways - diode bridge is faulty and broken or you have a short cut between copper rails. Requires replacement of a bridge or bending the rails.

Personally I skipped further parts of this step, but you can check everything.

To check ULN2003A you need to:

• connect the output "+" and "-" to the 12V battery, ie energize the circuit.
• Between the 12th and the 8th leg of ULN2003 should be a shortcut or low resistance.
• Now, if we'll shortcut our Zener Diod (solder a jumper) 16th to 8th legs have NOT to be shortcut.

If it does not work: somewhere screwed up or bought the faulty chip. (That's why I recommend to by a couple of these)

Check the operation of BTA triac:

• energize the circuit from 12V battery
• check shortcuts between each phase and the "+" must be infinite or MOhm resistance, no shortcuts.
• Shortcut Zener Diode (solder a jumper) - between the phases and the "+" must appear a beep (they are shortcut).

Does not work: looking for faulty BTA semistors - change with a new one.

Important part!! (as always)

In experiments with the running motor better do NOT connect on-board electric network as a load because in some weird cases of force majeure there is a possibility to burn out all the electronics on the bike. In certain cases, when there is an fatal error made during mounting and soldering of the relay, the VR can issue a 300V!

Therefore I recommend to use as a load 55-60W 12 V bulb connected directly to the battery like on the picture. Disconnected output circuit won't work, so you need any load like "bulb + car battery" anyway.

Always, under all circumstances, keep terminals connected to the battery on a running motorcycle! If not, than this is a sure way to kill its brain! (if you have already done so, and the brain is still alive, you just got lucky)

Now, check that your output wires do not shortcut, do not lay on the bike frame, do not lay on your hands etc.

Start the bike an use multimeter to check the voltage on the output of VR.

Generally, you need to find a zener diode so that the whole scheme gave the needed voltage. For a start try to understand what is the correct voltage. If an ordinary battery is acid, its rate of charge lays within 13.8 - 14.5 V. If the battery is gel, the rate is 14.5-15.5 V. Different sources say different. In general, it is of course the choice of each biker and everything depends on the specific conditions. I used a 13V zenner and had 13.5-14V output, if you'll use 14V diode, you'll probably get 14.3-15V. Suit yourself depending on your power consumption (Xenon lights or heat bulb, onboard music, etc.)

It is important to take into account that you need to check voltage on the position of about 1/3 of the engine speed, ie, if you have a motor spinning up to 10 000 RPM, then calibrate at 3000 RPM. At different RPM voltage must not vary significantly +/- 0.5V, this will be quite normal.

If you checked all you could check and everything works fine, than you are a winner, the whole thing can be filled with epoxy glue (compound).

If you want the scheme to work long and reliable, it is a mandatory part because of the bumps and dirt - it will not live long (a day or two). For fans speed up the process: adding more thickener to glue, epoxy resin may lose their dielectric properties. So, in the preparation of the adhesive, follow the instructions. I'm bad at portioning so I bought epoxy with already measured amounts by caring dosing machines at the factory.

Now you will need a plastic vessel, a syringe and gloves from nearby apothecary. All needed parts are on the fist photo. Transparent liquid is epoxy, near it stands hardener.

Make a very good bath in your heat sink. I made my out of paper sticky, it seeped from all holes. Better make it out of baking wax paper and turn it with duct tape. The problem is caused bu the epoxy at the time it process a chemical reaction. It heats up to 60 degrees Celsius and nearly begins to boil letting out all the gases and bubbles. It also becomes very fluent unlike it is at the start of reaction.

Pure epoxy and hardener together, mix it well, follow instructions on the box or wherever they are.

Watch for the bubbles gather together into big ones. when they do so, pop them with a needle. At the start (30 minutes or so) you can help small bubbles to gather into big ones, afterwards look on the consistency of the epoxy not to make craters on the surface.

Wait 24h or whatever is written on your epoxy box.

Install VR on your bike, better to do it like on last photos - outside. This will give you an opportunity to check temperature first couple of days, and will give good cooling airflow through the heat sink. After that VR can move in other place. I made a meter long wires, so after I figured out that it becomes barely warm it moved under the seat.