I made a 3-phase power rectifier circuit with "ideal" diodes. Is there anything I should change?

I made this 3 phase power rectifier with ideal diodes, instead of actual diodes. I basically just copied an ideal diode diagram and pasted it 6 times. It's for use on a wind turbine project, to rectify the power from the turbine windings. The load is the circle atop the diagram.

The turbine voltage can be 0..30 V in extreme conditions. Its frequency can get up to about 3 kHz in strong wind.

My main goal of using an ideal rectifier over a traditional one is to reduce the thermal losses.

It runs okay in my simulator, but I would really appreciate a review of it.

Also, I noticed most of the ideal diode diagrams I've found use P-channel MOSFETs along with PNP transistors. I'm normally used to seeing N-channel MOSFETs being utilized for switching instead of P-channel. Are they P-channel because it's harder to drive an N-channel MOSFET since the gate voltage would need to be higher than the supply voltage? If that's the case, I'd be fine with adding a boost circuit to run the N-channel components.

• ((DC) "Bus voltage" would need to stay below max gate voltage.) Apr 17 at 9:54
• (You can export a falstad.com design as an URL for others to tinker with.) Apr 17 at 9:55
• How about just a step up transformer and regular diodes? Apr 17 at 11:18
• @winny, how would that reduce thermal losses, which is the OP's stated goal? Apr 17 at 12:21
• @winny, that explanation is more help to the OP now Apr 17 at 12:43

The "ideal diode" circuit is intended for switching between two DC sources with minimal losses. It has a limited voltage compliance range, in the sense that both ends of the "diode" need to be far enough above ground voltage in order to keep the PNP comparator circuit working properly and the MOSFETs fully turned on. Specifically, you can't ground one end of the "diode" and expect it to rectify an AC signal with low loss. When you do that, you're actually just using the MOSFET's body diode as a rectifier.

Another limitation is that the voltage can't exceed either the $$\V_{gs}\$$ rating of the MOSFETs or the reverse $$\V_{be}\$$ rating of the BJTs.

Bottom line, this approach is a non-starter for what you're trying to do.

If you really want to use six N-channel MOSFETs, you'll need three proper half-bridge gate drivers, each with its own boost circuit, and you'll need to generate six different control signals as illustrated in the graph below.

Basically, for each phase, you need to switch on its high-side MOSFET when it is the highest available voltage, and switch on its low-side MOSFET when it is the lowest available voltage.

simulate this circuit – Schematic created using CircuitLab

If you really want to get sophisticated, take a look at regenerative braking circuits for electric vehicles. It's essentially the same concept.

• thanks for the help. In addition to regenerative braking circuits, would the circuit diagram from a household inverter generator work as well? From my understanding, most Honda, Harbor Freight, and Generac inverter generators rectify the generator output before regulating it, then output it as single phase AC again for household use. Would the circuits from those work as well? I'm assuming they're using some form of ideal diode instead of traditional bridge rectifiers. Apr 17 at 23:44
• Also, since 3 out of the 6 ideal diodes are on the low side, should those be the ones with N-channel MOSFETS? Apr 17 at 23:47
• I believe your understanding is wrong. I've never seen a household generator that was anything other than a simple alternator connected directly to the output sockets. Regulation is usually achieved by varying the field current. Apr 18 at 4:25
• I've got both an old generator that is set up like you said, and also a newer inverter generator that has a lot more going on. I don't know exactly how these newer ones work, but I figured there has to be some form of AC rectification if they use an inverter. Apr 18 at 5:10

I'm not sure if you are aware but your circuit doesn't make use of the MOSFETs properly i.e. you are not using the MOSFETs as ideal diodes. Instead, you are using their body diodes which is no different that using actual silicon diodes.

So, what you see at the output is still the rectified 3P sinewave but rectified by the MOSFETs' body diodes.

NOTE: I will not repeat the max gate-source voltage problem here as it's already pointed out by others.

You can use p-channel for the high-side mosfets, this will allow you to keep the gate voltage between the rails and not above the V+ rail as an N-channel would require. Then, you can use op-amps to turn on the high-side p-fets when the phase goes a few millivolts above the V+ rail, and turn on the low-side n-fets when the phase goes a few millivolts below V- (or GND). Below is a circuit I have implemented and tested to be functionnal. It only goes up to 5V though. I'm pretty sure you could get a similar circuit to work up to 30V with some modifications.

Edit: For some reasons some net tags appear blank. They are "VCC". Sorry, this is the best I can do on mobile right now.

• best I can [do] right now Date: 2020-03-15 In case you did not create the schematic yourself, please disclose where you got it from and who originated it. Apr 20 at 5:07
• Schematic is my own work. I can share a .pdf with anyone interested. Apr 21 at 18:27

Have a look at the Linear DC2465 demo. It's basically what you want, although you might have to use higher voltwge rated mosfets for your purposes.

https://www.linear.com/demo/DC2465

Another thing no one seems to be mentioning: its very important that your bridge performs the commutation correctly. i.e. There should only ever be 2 bridges conducting,the one at the highest voltage, and the one at the lowest voltage. If, for example your switch-high criteria is just V>=midpoint, you will effectively always be shorting the middle leg.