Bridge rectifier made with MOSFETs

[hkBattousai]

Today, I learned that an IC called LT4320 exists which is used for making bridge rectifier by using MOSFETs as ideal diodes.



I checked it on Digikey to see that its price varies between $3.6 and $18.9. Then I made a quick Google search to see if the same circuit can be done without using LT4320, and after inspecting several circuit models I end up with making the one below.



I understand that the peak value of the AC must be above some level to drive the MOSFET gates. And it should be below some level to keep the power loss on the resistors low. Would it work successfully between 12V-48V range (of course the resistor values will be recalculated after choosing a proper zener product)? Do you see any design mistake in this circuit? Would it work in a practical real life circuit?

[JohnnyBerg]

Would it work in a practical real life circuit?

No, it won't. The whole idea is that the MOSFET's are either of or complete on. So they do not dissipate any power.
In your circuits, the MOSFET's go on very slowly, because of the slow rising of the sinus. They will get very, very hot 

[hkBattousai]

Would it work in a practical real life circuit?

No, it won't. The whole idea is that the MOSFET's are either of or complete on. So they do not dissipate any power.
In your circuits, the MOSFET's go on very slowly, because of the slow rising of the sinus. They will get very, very hot 

I felt this could happen too. Thanks for clearing it out. I will try to improve this circuit by keeping it as cheap and simple as possible.

[langwadt]

Would it work in a practical real life circuit?

No, it won't. The whole idea is that the MOSFET's are either of or complete on. So they do not dissipate any power.
In your circuits, the MOSFET's go on very slowly, because of the slow rising of the sinus. They will get very, very hot 

Look at the direction of the mosfets, they have the body in the forward direction so they will work like a regular diodes until
they are turned on, power dissipation is not the issue

The problem is that once the fet turns on they won't turn off

[splin]

And perhaps even more fundamentally, the MOSFETs will turn on as soon as their gate threshold, say 2.5V, is reached. They must only be turned on as long as the AC input voltage is equal to, or exceeds the DC output voltage or they will short the output.

[NiHaoMike]

What would be interesting would be a 3 phase version for rectifying alternators. Apart from automotive applications, it can be quite useful for small alternative energy generators.

[krish2487]

There is nothing new in it.
Look for synchronous switching/rectification.
The entire idea of synchronous switching is to replace lossy diodes with near ideal FETs.


Though, it is more often seen in the likes of ZVS, stuff like that.
What LT has done is to package such switching with a ZCD to decide which FETS to turn on.
It can be done dicretely, though I suspect it will be a bit larger.


H11AA1 for the ZCD, steering logic to decide which fet to trigger and a optoisolated FET bootstrap driver.


EDIT: Sorry, posted the wrong opto by mistake. Fixed it now.

[digsys]

I used to make my own before the LT4320 came along, and it was a sheetload of parts ! I used opto drivers with xing logic from a low voltage tap.
THEN you have to take into account when to switch and IF the OP drops due to high load, keep the switch OFF or lose it in heat. The LT4320 does
a LOT of work, sensing current flow direction etc etc and like others have said, you HAVE to keep it saturated in ALL I/O conditions !!!
I've only ever paid $2-3 for the LT4320, admittedly in 100s, but there is no way you'll do as good / reliable job for less.

[janaf]

This is worth the effort for low voltages where diode voltage drop becomes proportionally a large loss. For higher voltages, it gets less important.

[Mad ID]

This is worth the effort for low voltages where diode voltage drop becomes proportionally a large loss. For higher voltages, it gets less important.

What about high voltage when you would otherwise have 2W+ dissipation, and with this thing you have 5x less...this saves board space and even enables some products which are not otherwise possible in terms of size etc.

[janaf]

This is worth the effort for low voltages where diode voltage drop becomes proportionally a large loss. For higher voltages, it gets less important.

What about high voltage when you would otherwise have 2W+ dissipation, and with this thing you have 5x less...this saves board space and even enables some products which are not otherwise possible in terms of size etc.

Could of course be such cases. Small mains transformers have 20-30% losses, big ones down to 4-5%. Depending on if the diode losses are small or big compared tho the transformer losses, maybe, maybe not its worth it. Hard to say generically. Do the maths....

[krish2487]

This is worth the effort for low voltages where diode voltage drop becomes proportionally a large loss. For higher voltages, it gets less important.

Or you have reasonably high voltage and at 10~20+ amps where the power dissipation in the diodes plays an important role in efficiency of the converter.  [size=78%] [/size]

[necessaryevil]

I have seen a schematic using opamps. The drawback: it uses 2 P-Mosfets and 2 N-mosfets instead of four N-mosfets. Those P-mosfets have worse specs and/or are more expensive. This doesn't mean it isn't usefull, depending on your application.

If you want to use four N-mosfets you will need some higher voltage to drive the gates of the high side mosfets, as krish2487 mentioned. This voltage should be 6-10 volts higher than the output voltage.

[NiHaoMike]

I have thought that a typical car alternator would benefit a lot from synchronous rectification. But it looks like the trend is to make a high voltage permanent magnet alternator, which means minimal losses in rectification, and then step that down with some sort of switching converter.

[Liv]

Would it work in a practical real life circuit?

Not for capacitive load. Circuit like this I'm using in DIY soldering station (see attachment). MOSFETs are absolutely cool. For capacitive load need slightly complicated circuit:

http://ic.pics.livejournal.com/leoniv/12037956/637167/637167_original.gif

Sorry for my english.

[eneuro]

Probably, you are not going to improve upon what they have already done for you.  Just use it...

Unfortunatelly, it is only 2 phase, so they did nothing for me for 3 phase, but adding 3 Hall current sensors in smart way I was able DIY it myself and it works for 2 phase too 

[hkBattousai]

I improved the design. Can you please make comments about this?



Please tell me if you find any mistakes on it.

[NiHaoMike]

You'll need some more advanced logic to prevent backflow from the output to the input. On the other hand, it could be the start of a surprisingly simple grid tie inverter! For experimentation, a low voltage version could be built that interfaces to the mains via a common AC wall wart.

[eneuro]

You'll need some more advanced logic to prevent backflow from the output to the input.

Two Hall current sensors does the job there, while we do not need high precision current measurements, while need to switch proper mosfets to bypass body diodes when current is above some threshold value eg. 1A. 
So, if current is below 1A whatever current direction is we switch all mosfets off than its body diodes conducts, but losess are small lets say about 1.5W per switch, but from Hall current sensors we know which diodes conducts, so if this current is above 1A, we know which mosfets should be bypassed 
At the end of half wave cycle (it will depend on load, so it can be inside half wave period, eg. when capacitor load will move this time to different voltage levels) current will drop again below 1A so we have some dead time when current is crossing set threshold (1A in this example).
Additionaly if mosfets are bypassed we monitor if current didn't changed direction (eg. regerative braking tries to push energy back througth our "not ideal but low power losess bridge" and if voltage will be higher than from bridge input, we need to switch off that bypassed mosfet too, to prevent flow in reverse direction).
Using one Hall sensor - since I know current direction- per half-bridge I know which side (low or high) should be bypassed, so control is very easy based on Hall sensor radiometric output.
 
Maybe it could be made cheaper without Hall current sensors, but I like to know current levels in my bridge anyway, so the same sensors can be used to monitor current.