Clean DC from wildly varying AC 1

[onekohm]

Hello all,

I have a pet project that is giving me a headache.
I am trying to power a microcontroller circuit from a magneto. The magneto is an AC source which varies in frequency from about 2 to 40 kHz. Waveform is very roughly sinusoidal, with peak voltage varying from about 8 to over 100 volts. Voltage peak increases with frequency.

The microcontroller and associated circuitry (mostly LEDs) will require max 2 A at 5 V.

I have been wrestling with this for a while. So far the best I have come up with is a 1:1 transformer to tap some power without affecting the rest of the system. Then a diode bridge followed by a zener diode and an emitter follower as an active clamp. After this caps and a linear regulator. I am worried the large variation in the AC's peak voltage will cause problems. I would also like to minimize wasted power as much as is possible.

Thanks for your help.

 

[itsmoked]

Bridge rectify the magneto output directly.(no transformer)
You must use diodes for the bridge that are fast enough for 40kHz,(not a big problem).

Charge a cap with an appropriate voltage and temperature rating that exceeds the magneto's output ability.

Select and use a buck switching controller to draw from the varying voltage stored on the cap to create whatever voltage you seek.

If you are actually using this result only to run LEDs then you should seek to make the buck regulator a current regulator instead of a voltage regulator.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[TurbineGen]

I agree with Keith, forget the transformer.

I'd use a switching voltage regulator for voltage control. Use capacitors and a choke on the output to clean up the current spikes to reduce rf interfernce. I would probably run it with a choke at the output followed by a few capacitors. You can also use a diode to clean this up even more.

I would stay away from the linear regulator just due to heating reasons. If you have 100 Volts peak, and are bringing that down to 5VDC at 2 Amps, that is going to be a tremendous amount of heat for the linear regulator to dissipate. A switching regulator uses short pulses and therefore will be significantly more efficient.



------------------------------------------------------------------------
If it is broken, fix it. If it isn't broken, I'll soon fix that.

 

[onekohm]

Thanks for your suggestions. I've been reading buck data sheets, and I will still need something to clamp the voltage peaks from the rectifier. So far the highest allowed input voltage I've found with the output I'm looking for is 60V. Is zener and emitter follower good for this, or do you have other suggestions there?

 

[Skogsgurra]

I am not so sure that the solution is as simple as Smoked puts it. First, if you do not use some sort of transformer - or SMPS variant - you will need to draw 2 A from the magneto at all voltages. That would mean something like 150 W at maximum voltage (assuming that RMS is around peak/sqrt(2)). Thinking further, It means that you will need to pull 2 A out of that magneto at lowest voltage, which is - if I read correctly - 8 V peak. Using a rectifier bridge will drop between 1.5 and 2 volts, so there isn't much left for your linear regulator. An LDO may work. But then again, you will have problems with the 100 V peaks. Using a zener is not a very good idea because it will consume immense amounts of current at maximum voltage and using a series resistor won't work because of the low voltage end.

An SMPS with an exceptionally broad input voltage range is probably the best solution. But still doubting if you can have your 2 amps at 5 V when running at low speed.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[OperaHouse]

I certainly agree. A friend of mine designed a line of universal 24-240V AC/DC industrial modules. He used LED current drivers fed into a zener, anything not used was wasted as heat. Fine for a device under 1W. I'm sure if he could have found a way to do it just as a voltage regulator, he would have done it. I think starting at 8V is a real issue in the design. You are out of mainstream commercial usage and believe this is beyond the capabilities of a neophyte. Do you really need 2A? Seems like a lot for LED's.

 

[onekohm]

Gunnar, you've hit my biggest problem right on the head.

I was hoping using a SMPS with 4-60 volt input would allow the zener, as I would only have to clip above 60 V. Is there a better method?

The low end is a problem. I haven't checked output of the exact magneto, but a very similar one I looked at a while ago had this output. I will be able to check out the exact output soon, hopefully it will be a bit better.

OperaHouse, the LEDs are not in series as I need to individually control them.

 

[BobM3]

Dual winding transformer? Use the microcontroller to sense the voltage into the SMPS. Have it switch windings to limit the range of voltage into the SMPS. Or you could switch the input capacitors to the SMPS to be either in parallel or series (again, using the microcontroller to detect the input voltage).

 

[itsmoked]

Gunnar.. Why would you always need to draw 2A from the magneto?

I'm not talking a linear! You need to draw 2A x 5V = 10W from the magneto.. That means at 80V only 125mA.

If you don't want to deal with the 100V then just add in a voltage controlled switch between the rectifiers and the cap that disconnects the cap whenever the cap voltage exceeds 50V. You're following the cap with a switching regulator it will use whatever the cap has on it and does not care that the cap voltage moves around.

If you are trying to run this thing down in the ditch where you shouldn't be, getting full power from it at 8 volts, you should check to see that the then required 2A is not going to be too much for the magneto's windings. If this is still OK then just select a buck/boost switching regulator. They can boost the voltage as well as buck the voltage. I'd still review the logic for that.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[TurbineGen]

The real problem I see is the 8Volt minimum voltage you need to work with. As stated earlier you drop 1.5 volts across a full wave recifier, but you could use a single diode and only drop 0.6-0.7. That allows you a little bit more voltage to play with, but not much.

Here's what I can come up with: two separate power sources.

Use a single diode (halfwave rectifier) to get your DC. Use a large capacitor as a filter to smooth the ripple. Now you need some type of selector switch, I'd use a DPDT two pole relay. A mechanical type might work fine if you can handle a short power dip in the order of a 200mS. You'll need a voltage sensor of some sort to switch the relay when output voltage exceeds 50V.

The relay will then drive two separate voltage regulators. One with be the 60V max input 5V switching regulator with a choke and added filter capacitor. The second will be a two stage regulator. The first stage would be some buck regulator that can get your voltage down below your 60V threshold, and from there you can use the 5V switching regulator as in the first output.

I think that should do it for you. I can't think of any other way to overcome the issue, but perhaps someone else has an idea.

Use a switching regulator to get your 5V.

------------------------------------------------------------------------
If it is broken, fix it. If it isn't broken, I'll soon fix that.

 

[Skogsgurra]

Yes, Smoked. That is correct when using an SMPS and running at decent voltage levels. But,there is a spec saying 2 A at 5 V at peak 8 V from the magneto. I have no idea if the magneto is an untypical one or not. But I cannot imagine taking 2 A from a typical magneto at 8/1.4 - 1.5 = 4.2 V. Especially if we are talking the ugly waveform you get when hooking a rectifier and capacitor to the magneto. Adjusting the spec to not include the lower voltage makes it a nobrainer.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[onekohm]

Thanks again for your help. I'll mull it over over the next few days. If the info I get from the actual magneto differs by much I'll get back to you with some new numbers.


Until then, Happy New Year everyone.

 

[MacGyverS2000]

Sounds like a hobby project for a bike-based LED lamp... magneto on the tire and a few high-wattage LEDs.

Dan - Owner

http://www.Hi-TecDesigns.com