Voltage regulator

[sonic22]

Hello everyone...

Small introduction: I am 32, i am from Greece (so English is not my primary language), i have had some very basic electronics training when i was young and my main profession is an engineer-mechanic.

I have a motorcycle (a TMracing 300EN if anyone is interested) and for the past 3 months i have been trying to build a rectifier/regulator that can handle up to 110VAC.
Long story short, after much searching and trying deferment methods (transistor Vgs limits claimed the life of many components until i found out what was going wrong) i ended up with this as a "solution":



and the actual prototype








Problem is that without load the circuit works and i get a steady DC of about 13,3V. But when i put a load on the output i get something like 1,3V but no overheating or magic smoke.
So again to make a long story short, what am i doing wrong? Are the resistances too low? Do i need to add some diodes somewhere? I am too lost to try and trouble shoot this on my own with my very limited knowledge.


p.s.
I am trying to make this first post short as short as possible. I can provide additional info about the circuit if needed.
The circuit has no capacitors, fuses, heat sinks etc, since it was build as a proof of concept and it is turned on just for a few seconds at a time.
The "DIY power supply" is crud but for the hight Voltage-power i needed (i can get about 89VDC and more than 2Aamps) is what i could afford or be able to build. Yes it can be dangerous and i DO NOT recommend someone trying.




Thank you for your time.

[sonic22]

Looking at the pictures i noticed i had missed a line.



Now i am getting less than 10V without load... and almost nothing with one.
No overheating or sort circuit  though...

[mikerj]

You show NPN bipolar transistors in the schematic, but the part number is an N channel MOSFET and you mentioned Vgs, so are they MOSFETS or bipolars?

If they are MOSFETs then using them as a source follower in this manner (i.e. with no feedback) won't work very well.  Vgs increases with the Drain current demand, so your regulation will be extremely poor.  Vgs for a drain current of 250uA (i.e. virtualy nothing) is between 3 and 5 volts, but to get 10 Amps that's already up to ~6v and increases from that point.  There is also a strong temperature dependence.

Even if you modified the circuit to achieve reasonable regulation performance, you are quite likely to exceed the safe operating area for these devices.  Dropping around 30v on a FQPF18N50V2 means you can pass a maximum of just 1 amp continuously, and you'd be dissipating 30 Watts to do so.


Out of interest, what's wrong with the conventional SCR or MOSFET based shunt regulator that's fitted by the manufacturer?

[sonic22]

They are N channel Mosfets (FQPF18N50)

Using only a single zener-mosfet i was able to get almost 2Amps (that is all i need) but when the input voltage got above 50V the mosfet would fry. There was a lot of energy-heat coming from the mosfet (about 130C) but with a big enough heat sink it worked.

The things i am trying to power up (a 35W incandescent headlight and some LEDs)  are not very picky about perfect regulation. Anything upwards or 11V and bellow 16V is good enough, no matter the ripple or voltage drop.

The bike is used 99% of the time during the day and off road. But i do want it to be perfectly functional without the need of batteries, especially if they need to be charged off the bike. For some unexplained reason the bikes generator (simple singlefaze, coils with a permanent magnetic rotor) is putting out too many volts (at full throttle i get above 80VAC). Normal rec/regs just get destroyed and then the entire electrical system is destroyed. MOV protection is a must, but first i need to have something that can handle the voltage. They original rec/reg costs about 90Euro and it is a joke. (i have pictures when i took them apart if anyone wants to see).
SCR circuits are my last hope but from what i have seen they are much more complicated for those voltages-power. Also i do not have a reliable ACgenerator (at least not for those voltages) and so experimenting is much more difficult.

p.s. The reason the bikes generator is suddenly putting out so much voltage (normally it should have 48VAC max) is inexplicable (i have asked professionals everywhere) and the only solution would be to change the entire electrical system witch is VERY expensive.

Thanks for the reply.

[mikerj]

For some unexplained reason the bikes generator (simple single faze coils with a permanent magnetic rotor) is putting out too many volts (at full throttle i get above 80VAC) and normal rec/reg just get destroyed and then the entire electrical system is destroyed.

80V+ is absolutely normal for the the kind of permanent magnet alternators used on motorcyle engines when they are unloaded.  The voltage will rise in proportion to engine speed (voltage proportional to rate of change of magnetic flux).

MOV protection is a must, but first i need to have something that can handle the voltage. They original rec/reg costs about 90Euro and it is a joke. (i have pictures when i took them apart if anyone wants to see).
SCR circuits are my last hope but from what i have seen they are much more complicated for those voltages-power. Also i do not have a reliable ACgenerator (at list not for those voltages) and so experimenting is much more difficult.

The SCR based regulators are very simple indeed, they simply shunt (short out) the alternator when the voltage waveform rises above a reference voltage.  This means the power dissipated in the regulator is low, as an SCR is either fully on or off.  You could implement the same method with your MOSFETs.  When the SCR is turned on, the energy is dissipated within the stator windings but will be much lower than you may expect i.e. the current won't be I=80/Rstator because the leakage inductance of these alternators is quite high by design, so they behave more like a current source than a voltage source.  Have a look at this schematic for an example of a single phase SCR shunt regulator.

You can buy a reasonably inexpensive universal single phase shunt regulator that will do the job, or you could even use the three phase regulator from a larger motorcycle and simply leave one of the phases disconnected.
Most larger bikes use a 3 phase system, but you could still use one of the regulators by connecting up your alternator output to two phases.

[sonic22]

universal single phase shunt regulator[/url] that will do the job, or you could even use the three phase regulator from a larger motorcycle and simply leave one of the phases disconnected.

Here is my collection of " motorcycle rectifier-regulators"



3 on top are 3 phase and the 2 at the bottom are the original ones.
All of them fried immediately, all of them are zener-Thyristor (SCR) based.
That is why i am trying something different.

[sonic22]

In any case i am very interested to find out why my device does not work.
Is it something i missed or did not do right or is there a more complex reason?

[flynwill]

What is the actual input voltage under in your test conditions?  As others have mentioned these MOSFETS need a gate voltage of 6-10V to turn on fully  (looking at the data sheet Vgs=6V will get you 1 amp).   Since you have 5 of these in series you need at least 30V  in to get any output at all.   If you do have enough input voltage, with a 1 amp load the output voltage will be about 9V.  If you're trying to charge your 12V lead-acid battery you need about 14V.  I think if you measure each of the gate voltages you will get a clearer picture of what is happening.

If you had built the regulator from bi-polar transistors as it is drawn it would work better, but still have relatively poor regulation.  Either way you'd be better off with a single device and a bigger heatsink.

As others have mentioned this not the best way to accomplish this.

[mikerj]

In any case i am very interested to find out why my device does not work.
Is it something i missed or did not do right or is there a more complex reason?

Are you certain that the winding on the stator is floating, i.e. are there two separate wires coming from the stator, and neither connect to ground (the frame)?

Some bike (mainly small mopeds and off-road bikes) have simple electrical systems designed to be powered entirely from the generator with no battery, and in this case one side of the stator is grounded and everything runs on AC.  If this is the case you can not use a full wave reg/rect, and this may explain why you have been blowing them up.

If it is supposed to have a floating stator, disconnect it from the wiring loom and use a multimeter to check for continuity to ground.  It's possible for the insulation to break down and cause a short to ground.

[sonic22]

@flywill
I have tried input voltages all the way up to 89VDC. The output voltage is always bellow 10VDC but i can see the multimeter fluctuating. I will connect a usb-osciloscope i have later to get a better understanding of what is going on.
I am not trying to charge a battery. The power generated is used directly.
I will take a few more orderly measurement later and report.

@mikerj
The stator is electrically isolated from the frame/ground. I used a multimeter to check continuity and there was none.


I really thought i had done some silly-noob mistake in the circuit that would be obvious at a glance to everyone around here. It appears things are more complicated... no?

[mikerj]

Does anything else connector to the stator, or does it it connect only to the input of the reg/rect?

[sonic22]

So i took some measurements today...
No load Vg for 60VDCinput
59,4V  16,7V   15,6V   14,4V   13,7V
with load
59,4V    4,3V     2,6V     1,1V     0,04V

Also did a diode check on the mosfets source-drain. They all conducted one way.

@mikerj nothing is conected to the stator. Only thing is that the stator has 2 single phase windings in it and i only use one of them. They are electrically isolated from each other and/or the frame. They also give out the same voltage.

[flynwill]

So one of your first two stages isn't working as expected.  Did you measure the voltage at the connections between them?  My guess is incorrect wiring or you mixed up which zener was which.

What is your "load"?

[mikerj]

Looking at the underside of the board, it's not obvious how the top of the first stage resistor and the MOSFET drain connect to the positive output of the bridge rectifier.  If the MOSFET gate was floating on the first stage it may explain things.

[dom0]

Forget MOSFET source followers with no global feedback. It will never work satisfactory.

With BJTs you can do this, when you use darlingtons you can just use one Zener and a resistor ladder.

Note: you get -2 mV/K tempco per stage, double that for darlingtons. Probably doesn't matter in this application.

[sonic22]

You are right!

I had failed to connect the output of the bridge rec to the first mosfet Drain.
I did and now for a 89VDC input i get an output of 12,3VDC.
It is a bit lower than perfect but OK.

Voltage at the gates (@89Vdc) are now
87,6V   87,1V   68.7V   38,7V   14,8V.


Next big problem is that when i put a load on (like a 10Watt lamp) the power supply goes wanly (both current-voltage meters go out and there is a dziiit sound but nothing like when i short circuit it) and i measure about 8VDC output on the load.
The power supply can handle 10Amp @ 12V and the converter is capable of 600Watt.
Strange...
The mosfets now produce lots of heat, as expected.

Thank you everyone!

p.s.@domO are you sure that the Darlingtons can handle the voltage-current i need? What configuration do you have in mind?

[Ian.M]

A better approach would be a shunt regulator.   First you need to determine the maximum possible current you need to sink.   The easiest way of doing that would be to connect the PM alternator output to a bridge rectifier and the bridge output to a 12V car battery which should not be fully charged.  Run the engine at 90% of redline, measure the battery charging current, and ASSUME the peak current at redline is no more than double the measured current.

You then choose a power resistor to pass that current at a nominal 12V, and build a shunt regulator with the resistor in series with the power transistor's collector lead.  Half of the worst case dissipation will be in the resistor, the other half in the transistor, which needs to be heatsinked accordingly.   As the transistor will have a fairly low HFE at its operating current, you will need a Darlington configuration to provide enough base current.   Drive the base of the Darlington with a 12V Zener from the positive rail with 100R in series to protect against transiently exceeding the max permissible base current and add 1k pulldowns to both transistor bases to avoid leakage current problems when it gets hot.   That should stabilise the positive rail at between 13V and 13.5V.  You should also fit a large electrolytic capacitor across the bridge rectifier output, sized to not sag more than 1V at full load current between half cycles from the bridge rectifier.  It will need to be rated for a ripple current equal to the current measured above + 20%.

[dom0]

p.s.@domO are you sure that the Darlingtons can handle the voltage-current i need? What configuration do you have in mind?

Safe operating area. MOSFETs are typically advertised with their maximum conduction ability, which has barely anything to do with the performance in linear mode, which this is about.

See attachment. All parts are example only, I would especially not recommend a 2N3055 here, but some higher beta device. Also, integrated darlingtons are a good alternative. D2, D3 compensate for temperature, depending on the zener used (whether it's a zero-drift one or not) you might want another Si diode there to get a low tempco. But that probably doesn't matter anyway.

You'll likely see a better load regulation and can achieve lower tempco when substituting Q6/Q3 with a Sziklai pair. Another alternative would be to regulate to only, say, 16 Volts and then use a standard three terminal reg behind that.

[sonic22]

Have a look at this schematic for an example of a single phase SCR shunt regulator.

I am having trouble figuring this out...

First of the PNP transistor has a emitter-base maximum voltage of 5V. Is that not a problem?
Second if i understand correctly the bridge rectifier will send a very high DC voltage to the battery-electrical system until the SCRs basically short circuit the AC generator voltage witch would happen almost constantly because of the very high input voltage.

[edavid]

Here is my collection of " motorcycle rectifier-regulators"

Is it possible that they are fakes?

http://mastercircuits.blogspot.com/2014/09/another-motorcycle-voltage-regulator.html

[mikerj]

Have a look at this schematic for an example of a single phase SCR shunt regulator.

I am having trouble figuring this out...

First of the PNP transistor has a emitter-base maximum voltage of 5V. Is that not a problem?
Second if i understand correctly the bridge rectifier will send a very high DC voltage to the battery-electrical system until the SCRs basically short circuit the AC generator voltage witch would happen almost constantly because of the very high input voltage.

5v is the maximum reverse bias voltage.  The base-emitter junction is forwards biased in this circuit i.e. the base should never be at a higher voltage than the emitter, unless you connect the battery the wrong way around...

The stator voltage will be loaded down by the battery and the rest of the electrical system, so even without the regulator the rectified output voltage will be much lower than the open circuit value.  When the regulator operates, it only does so for the part of the output waveform that pushes the system voltage too high, since the SCR will switch off on the next zero crossing.  Also remember what I said about these permanent magnet alternators; the high leakage inductance means they are inherently current limited and designed to be regulated in this way.  The vast majority of motorcycles use a shunt regulator like this (esp. Japanese bikes), though usually 3 phase for the larger bikes.