Best way to invert this voltage?

[ACharnley]

Playing and learning, an area I haven't any experience with is voltage inversion. What would be the best way to flip the green part shown here and add it to the DC effectively giving a full wave rectifier?

Ideally the method should support up to 100V and be the best efficiency possible.

Cheers, Andrew

 

[ronsimpson]

What is wrong with 4 diodes?

 

[ACharnley]

Efficiency. Why do you think I've been spending so much time with active rectification (having issues due to that inductor).

 

[ronsimpson]

Schottky diodes?

 

[rjenkinsgb]

Have a look at this - an "active rectifier" controller suitable for up to 200V input.
https://www.infineon.com/dgdl/ir1167aspbf.pdf?fileId=5546d462533600a4015355c45d9c1655

There is an example full bridge circuit using these on the second page of this document:
https://www.power-mag.com/pdf/feature_pdf/1242900929_input-bridges.pdf

 

[ronsimpson]

I did this on my project. Good diodes. Only two so there is less loss.
Note the voltage is 2X and the current is 1/2 of normal.

 

[ronsimpson]

active rectifier

I have used these. It only rectifies the positive half. Not full wave.
This really works at 10 amps where diode loss is a real problem. At 1A I would not because the IC and other parts have loss.

 

[ACharnley]

Nice find - looking at that chip now.

Trouble I'm having with active rectification is as soon as a FET is turned on (very close to zero crossing) it causes a voltage spike typically in the opposite signal (i.e both lines go positive) and this causes LTSpice's digital controls to go mad. Adding capacitance to offset the inductance helps but needs switching in and out depending upon the frequency (which I'm doing) and it's not perfect. Thus, I began looking at using one AC as ground and something like an isolated DC inverter.

 

[ronsimpson]

What is the current level?

 

[ACharnley]

I did this on my project. Good diodes. Only two so there is less loss.
Note the voltage is 2X and the current is 1/2 of normal.
View attachment 114623

Ah the classic voltage doubler . I'm having enough trouble finding 100V fets let alone 200V fets though! A 200V buck would be even harder to find.

At the moment I'm sinking the 0-100V into a super-capacitor with PWM, which causes the inductive supply to drop (I hold it at about 6V and switch off once the capacitors are charged).

 

[ACharnley]

What is the current level?

Dynamo's are 500mA rated but peak higher with the right capacitance. Generally 1-1.5A is possible.

 

[ACharnley]

£1.4 each, might as well buy LT's active controller for £6 (and I would if it worked at a lower voltage).

 

[rjenkinsgb]

as soon as a FET is turned on (very close to zero crossing) it causes a voltage spike

The active rec controller ICs have a minimum "on" time with the signal polarity ignored, specifically to prevent false operation due to ringing effects.

 

[ronsimpson]

In another direction:
I know you are unhappy about 1.5 volts loss out of 50 to 100 volts. BUT
Are you doing "maximum power point" on your dyno? If not you are loosing more power there.

 

[ACharnley]

The active rec controller ICs have a minimum "on" time with the signal polarity ignored, specifically to prevent false operation due to ringing effects.

So shoot through at a very low voltage? Worth investigating.

 

[ACharnley]

In another direction:
I know you are unhappy about 1.5 volts loss out of 50 to 100 volts. BUT
Are you doing "maximum power point" on your dyno? If not you are loosing more power there.

Don't forget 50-100V is almost no current draw. It's normally around the 6V mark where the voltage loss becomes critical.

 

[ACharnley]

Quick question Ron, the voltage doubler is effectively shorting the capacitors for half a cycle, was there any outcome of the related discussion that was had a few weeks ago on whether this advisable long term? I guess it only stands for electrolytics not tantalums?

 

[ronsimpson]

the voltage doubler is effectively shorting the capacitors for half a cycle,

No.
When the left side of the signal goes negative it hangs up on the bottom diode and will not go negative. So the right side pushes up and charges the bottom cap. Next then the left side goes positive it pushes up and charges the top cap. In one cycle you will have the peak voltage on each cap and they add to make P-P voltage.

 

[ronsimpson]

Don't forget 50-100V is almost no current draw. It's normally around the 6V mark where the voltage loss becomes critical.

I think you have a problem where your unloaded voltage could get to 100V but your loaded voltage is 6V. At max speed; I would put a 1 watt Zener in the circuit to limit the voltage. (wild guess) I think that a 25V Zener (maybe lower) will clamp the no load voltage. This will help your "high voltage MOSFET" problem.

Years ago I had a problem that my "generator" had a no load voltage of 1200 to 1500 volts so they wanted me to use 2000 volt transistors. For the prototype I shorted out the generator when there was no load. I used 650 volt transistors and could have used lower.

Do you have a graph of current & voltage at high speed.
------edited------
No data on. I have looked for hours.

Found this: Speed of cycle, output voltage held at 6 volts. (Zener diode as load)
speed watts volts current
RPMs
5___0.5___6.00___0.126
15___1.89___6.00___0.369
30___4.09___6.00___0.560
This is for 18 ohm load. (fixed load)
5___0.45___2.45___0.115
15___1.89___5.70___0.325
30___3.21___7.23___0.435
I am looking for a graph of what load produces the most power at different conditions. (graph 5, 10, 20, 40 , 80 ohms at each speed)
Also looking for data where the current is very low and the voltage is high.
I also want the constant voltage done at 3V, 6V, 9V 12V.

I think there is an amassing lack of data and knowledge on this subject.

 

[ACharnley]

Hi,

I'm a bit a head of you in the research, first the zeners - 3W aren't reliable. They get too hot at 20V. Shorting the generator (collapsing the field) using a triac doesn't work as the inductive nature of the thing causes it to miss it's zero crossing, same for ST's fancy TTL controlled version. Doing the same with a transistor DC side sounded ideal but the zero crossing point becomes predictive. I was switching off at ~0.01V and it was "boost"ing to over 250V. I contemplated clamping those spikes but considered it was becoming messy. The transistor was still dropping 0.7V and getting hot.

So I'm handling the full 100V, which requires N-FET's, so high side driving. I actually revisted your voltage doubler and integrated active rectification for the diodes. It's working perfectly in LTSpice and is my solution going forward.

A.