MOSFET bridge rectifier for DIY shaking torch? || Decent plan for project?

Thread Starter

-live wire-

Joined Dec 22, 2017
959
There is this "shaking torch" project that I am revisiting. It is where you shake this thing and a powerful magnet generates a few volts AC in a coil (I have found a max of ~ 2.5V AC RMS). Then it charges a supercap or battery, which powers a light or something. This way, it is not dependent upon batteries. It would be good if you were stranded in the woods or somewhere else, or can never seem to find batteries. I kind of gave up before because I did not have all the required parts, and there were BIG power losses.

Now, I want to design a more efficient and practical design. I should be getting some buck and boost converters, N- and P-channel MOSFETs, and other parts soon. The first part of the shaking torch circuit must rectify the AC. This is where I lost a lot of power before, with a silicon FBR. Up to 1.2V (with 2 diodes), so 1/2 of the power! But I have heard about using MOSFETs for rectification (though mainly for reverse polarity protection), with minimal power losses. So how would I go about building a FBR with MOSFETs? Is this possible and practical?

Also, here is my plan for the whole project:
~2.5 VRMS AC from coil & magnet --> MOSFET full bridge rectifier (to get~2.5V DC) --> supercap (5.5V rated, 7.5F, 90 mOhms) --> on/off switch --> resistor (.5 to 10 Ohms) or something else here? --> boost converter (to constant 3.3 V) --> LEDs (80 - 120 mA at 3.3V)

So is it wise to place a resistor in series with the boost converter? If so, what value? I am leaning towards using a low value one, but is that ideal for the boost converter working for low cap voltages and charging the cap faster? I need it to work from 2.5V down to .35-.5V. I am also concerned about the converter exponentially needing more current, and getting fried by the cap. So what is a good solution to these problems, with minimal other parts? Thanks.
 

cmartinez

Joined Jan 17, 2007
8,220
There is this "shaking torch" project that I am revisiting. It is where you shake this thing and a powerful magnet generates a few volts AC in a coil (I have found a max of ~ 2.5V AC RMS). Then it charges a supercap or battery, which powers a light or something. This way, it is not dependent upon batteries. It would be good if you were stranded in the woods or somewhere else, or can never seem to find batteries. I kind of gave up before because I did not have all the required parts, and there were BIG power losses.

Now, I want to design a more efficient and practical design. I should be getting some buck and boost converters, N- and P-channel MOSFETs, and other parts soon. The first part of the shaking torch circuit must rectify the AC. This is where I lost a lot of power before, with a silicon FBR. Up to 1.2V (with 2 diodes), so 1/2 of the power! But I have heard about using MOSFETs for rectification (though mainly for reverse polarity protection), with minimal power losses. So how would I go about building a FBR with MOSFETs? Is this possible and practical?

Also, here is my plan for the whole project:
~2.5 VRMS AC from coil & magnet --> MOSFET full bridge rectifier (to get~2.5V DC) --> supercap (5.5V rated, 7.5F, 90 mOhms) --> on/off switch --> resistor (.5 to 10 Ohms) or something else here? --> boost converter (to constant 3.3 V) --> LEDs (80 - 120 mA at 3.3V)

So is it wise to place a resistor in series with the boost converter? If so, what value? I am leaning towards using a low value one, but is that ideal for the boost converter working for low cap voltages and charging the cap faster? I need it to work from 2.5V down to .35-.5V. I am also concerned about the converter exponentially needing more current, and getting fried by the cap. So what is a good solution to these problems, with minimal other parts? Thanks.
This thread that I'm currently working on visits a few of the concepts that you've just described. Anyway, @ebp was kind enough to point me to an energy harvesting line of chips manufactured by Linear Tech. It's fascinating reading.
 

Alec_t

Joined Sep 17, 2013
14,280
You might find it difficult to source MOSFETs with a low enough Vgs(thr) voltage to turn them on fully at the low voltages the generator makes, but in principle the MOSFETS would be more efficient than regular diodes.
 

MisterBill2

Joined Jan 23, 2018
18,176
You might find it difficult to source MOSFETs with a low enough Vgs(thr) voltage to turn them on fully at the low voltages the generator makes, but in principle the MOSFETS would be more efficient than regular diodes.

The simple fix is to put a lot more turns on the coil and add some external iron to improve the efficiency. Also, possibly a stronger magnet is in order.
One other thing would be to use a full wave voltage doubler circuit instead of a full wave bridge circuit. Of course then you will need two of the big supercaps, but you won't need any voltage booster circuit. Also, you lose the two extra diode forward voltage drops. So this is indeed a better type of circuit, except for the cost of the extra supercap.
 

Thread Starter

-live wire-

Joined Dec 22, 2017
959
I think that you would have better luck with a geared up PM motor & crank.
Changing the entire project is kind of impractical, and it would also be a waste because I have already built the coil/magnet part.

The simple fix is to put a lot more turns on the coil and add some external iron to improve the efficiency. Also, possibly a stronger magnet is in order. One other thing would be to use a full wave voltage doubler circuit instead of a full wave bridge circuit. Of course then you will need two of the big supercaps, but you won't need any voltage booster circuit. Also, you lose the two extra diode forward voltage drops. So this is indeed a better type of circuit, except for the cost of the extra supercap.
I could try adding some iron on the outside. Adding more turns is impractical. I am using a really powerful neodymium magnet, and there is not really space a larger one. I do have some extra supercaps, so that is not a problem. However, there is still the issue of needing to use diodes. So how exactly do you use a MOSFET as a drop-in replacement for a diode? And, in addition to that, there will be no regulation on the output.

But how do you deal with the increased current demands of the boost converter? Would an inductor work? Lol.
 

MisterBill2

Joined Jan 23, 2018
18,176
Changing the entire project is kind of impractical, and it would also be a waste because I have already built the coil/magnet part.



I could try adding some iron on the outside. Adding more turns is impractical. I am using a really powerful neodymium magnet, and there is not really space a larger one. I do have some extra supercaps, so that is not a problem. However, there is still the issue of needing to use diodes. So how exactly do you use a MOSFET as a drop-in replacement for a diode? And, in addition to that, there will be no regulation on the output.

But how do you deal with the increased current demands of the boost converter? Would an inductor work? Lol.
If you used the full wave voltage doubler circuit you would not need a boost converter because the circuit gives you twice the peak voltage . In addition, it has only one diode drop instead of the two diode drops in a bridge circuit. Look up the full wave voltage doubler circuit and see what I mean. The only big change would be adding the second supercap.So you will have two caps in series, one charged on the positive peak and the other charged on the negative peak. About as simple as can be.
 

Bernard

Joined Aug 7, 2008
5,784
One pass gives about 50 msec. of light from a LED. Maybe 10 shakes / sec. ( one up, one down / shake ). would provide some light. My prediction for success- 2 %.
 

Thread Starter

-live wire-

Joined Dec 22, 2017
959
Anyways, I am working on turning this old computer charger (20V 4A) into a crude high-power function generator/variable bench power supply. It would be very useful for testing this circuit (and many others). I would not having to be shaking it constantly, and be able to look at other things. So, the computer charger first goes through an Arduino-controlled H-bridge, is filtered with some caps (for stable DC) and then there is a current shunt on the output so you can get constant current and set current limits. That would go through an op-amp and be fed back into the Arduino. I am just starting to make it, and still need to get some parts.

I am having some difficulties with the H-bridge. I cannot seem to get it to work so that you control the charger with the Arduino (PWM, polarity). How exactly should I wire up the H-bridge? I understand the basic working principles of an H-bridge, but I cannot seem to get it to work with a completely separate signal and power supply. For simplicity, I am just trying to get half of it to work, with only one P- and N- MOSFET. What happens is there is the full 20V on the output, regardless of the signal. Here is my setup:

I connected the source of the p-Fet, and gate of the n-fet to 5V. The drains are obviously my outputs. The gate of the p-Fet goes to arduino Gnd. The source of the n-fet to Arduino gnd. The charger gnd to n-fet source, +20V to p-fet source.

And I have gotten the individual MOSFETs to work, so is there a problem with isolation or something? I suspect that the charger is just feeding into the gates or something like that. If so, how do I avoid that?WIN_20180313_15_18_20_Pro.jpg
 

MisterBill2

Joined Jan 23, 2018
18,176
Anyways, I am working on turning this old computer charger (20V 4A) into a crude high-power function generator/variable bench power supply. It would be very useful for testing this circuit (and many others). I would not having to be shaking it constantly, and be able to look at other things. So, the computer charger first goes through an Arduino-controlled H-bridge, is filtered with some caps (for stable DC) and then there is a current shunt on the output so you can get constant current and set current limits. That would go through an op-amp and be fed back into the Arduino. I am just starting to make it, and still need to get some parts.

I am having some difficulties with the H-bridge. I cannot seem to get it to work so that you control the charger with the Arduino (PWM, polarity). How exactly should I wire up the H-bridge? I understand the basic working principles of an H-bridge, but I cannot seem to get it to work with a completely separate signal and power supply. For simplicity, I am just trying to get half of it to work, with only one P- and N- MOSFET. What happens is there is the full 20V on the output, regardless of the signal. Here is my setup:

I connected the source of the p-Fet, and gate of the n-fet to 5V. The drains are obviously my outputs. The gate of the p-Fet goes to arduino Gnd. The source of the n-fet to Arduino gnd. The charger gnd to n-fet source, +20V to p-fet source.

And I have gotten the individual MOSFETs to work, so is there a problem with isolation or something? I suspect that the charger is just feeding into the gates or something like that. If so, how do I avoid that?View attachment 148233
How did this discussion move from a shake-charged LED torch to an H-bridge experimenter's power supply? And for regulating a charger, or anything else, why bother with an aduino toy processor when an analog approach will be far more effective?
 

Thread Starter

-live wire-

Joined Dec 22, 2017
959
The Arduino would be far better at, say, generating arbitrary waveforms at low frequencies. It is not just for a variable voltage with current limiting/ cc mode. Not to go off on too much of a tangent (I know...), but how should I wire up my H-bridge so it can control the adapter with the Arduino? I am having problems with needing the common + and gnd.
 

MisterBill2

Joined Jan 23, 2018
18,176
The Arduino would be far better at, say, generating arbitrary waveforms at low frequencies. It is not just for a variable voltage with current limiting/ cc mode. Not to go off on too much of a tangent (I know...), but how should I wire up my H-bridge so it can control the adapter with the Arduino? I am having problems with needing the common + and gnd.
Sorry but I can't help there. I avoid even reading about those little Arduino things. Also, they are not suitable for use in a voltage regulator circuit.
 

Elektryx

Joined Mar 14, 2018
3
Following this thread, I had a similar question about using FETs for rectification.

I think the whole Arduino/H-bridge that -live wire- is talking about is a "shaking torch" simulator jig, am I correct?
If so, the attached H-bridge diagram is what you are trying to hook up. Be careful with grounding here, if there is an isolation transformer on your power supply or your Arduino, you should be able to hook the Arduino ground to the power supply ground OK, they both need to have isolation from the main line if you want to measure waveforms across the inductor with a scope.

I'm going to have to take a look at this voltage doubler circuit (I'm familiar with it), it's an idea I hadn't thought of. Hmmm... if I replace the diodes in the doubler with FETs?
 

Attachments

Thread Starter

-live wire-

Joined Dec 22, 2017
959
Yes, that's what I am trying to do. I think the adapter is a SMPS, based on size and weight. However, I could be wrong. That diagram was a little bit helpful, but it was kind of confusing.
 

Thread Starter

-live wire-

Joined Dec 22, 2017
959
Speaking of terrible isolation and unsafe designs, this is my H-bridge so far. WIN_20180314_16_51_18_Pro.jpg
Don't worry. I will add some hot glue to insulate it in the end. This is just temporary. But if it was a consumer product, I would need a really good lawyer. Anyways, I am going to try to turn the p-fets on with some more n-fets instead of the arduino. Hopefully it will work.
 
Top