LTSpice Active Rectification Simulation

[ACharnley]

Hi,

Can't get this to work and about to bang my head against the wall.

The two FET's at the bottom are creating a short, yet it's exactly as per the image and I've had it working before.

(replace the fets with anything you like just to make it work or use the included lib).

Regards,

Andrew

 

[DGM]

Yes, it doesn't work because that half of the bridge will always be shorted. Each NMOS device will be shorted through the other (flyback?) diode. Look at the current through each section of that half-bridge.

This could probably be rearranged to work, but I'm not up to troubleshooting it atm.
Attached is a static rectifier I made for battery polarity idiotproofing in a product.
This was just a proof of concept with low-current generic parts.
It also demonstrates an attempt to limit Vgs to within a safe range.

See if that is of any help.

 

[DGM]

Okay, I see what's going on. This is a dual N+P device. You specified the same subckt name for both parts, and both appear to be behaving as PMOS devices.

IDK how you can refer to two identically named subckts from one lib file. There's probably a way.
I just edited the lib file such that the NMOS and PMOS subckts had unique names, and then used those unique names in the simulation.

 

[crutschow]

Looks okay to me with different MOSFETs.
The problem is as DGM noted.

I assume you know that circuit will not work with a capacitive filter on the output as the filter will discharge through the bottom MOSFETs.

 

[ACharnley]

Hi,

Ah! I thought LTspice would use the symbol "nmos" "pmos" and match to the correct subckt. That lib fle came from diodes inc.

crutschow -> I am, I have a fast PIC reading the voltage and another FET shutting off (digital comparator).

Now I only need to work out if the sckhottys are really necessary. I'm playing with 500mA max and the Vgs for these FETS is only ~2V for 500mA, so...

0-7v-0.5v*2*0.5 = 0.2W additional loss without them under 2V.

The system can't boot until 1V so it's a small window to care about!

Thanks all!

Andrew

 

[DGM]

Hi,
Ah! I thought LTspice would use the symbol "nmos" "pmos" and match to the correct subckt. That lib fle came from diodes inc.

I kind of figured that whoever made the library was expecting it to work that way. Maybe it does work in other software; that's kind of why I was unsure if it could be made to work.

 

[crutschow]

Now I only need to work out if the sckhottys are really necessary.

Why did you add them?

 

[ACharnley]

Why did you add them?

Efficiency below 2V (the FET's aren't on so the internal silicon's are being used). Although the circuit can only work above 1V so it's a small window.

What do you think, over-engineering?

 

[DGM]

Unless you're using the MCU for something else already, idk why you can't just use a comparator with a fifth mosfet to implement the filter isolator.

As far as the extra diodes are concerned, it all depends on how your mosfets handle the use of the body diode -- and if you can spare the cost/space for extra parts.
You can avoid the use of the body diodes near ZC in the bridge, and you can avoid the use of body diodes in the first half-cycle of the isolator if you use them.
That first half-cycle in the isolator may be the biggest stress to avoid, depending on your rectifier/filter capacity.

 

[ACharnley]

DGM, I assume you're talking about capacitor back-flow. Indeed I'm doing that already.

As for the diodes, they're buil;t into the FET's so the skhottkys are doing very little. I'll probably delete them.

 

[DGM]

DGM, I assume you're talking about capacitor back-flow. Indeed I'm doing that already.

As for the diodes, they're buil;t into the FET's so the skhottkys are doing very little. I'll probably delete them.

Yes, I know you're using a capacitive filter. I was merely commenting about the use of a microcontroller to do the job instead of an opamp.
Also, I'm aware that the mosfets implicitly include body diodes. My remarks were based on that assumption. There are marginal benefits to having the diodes.
It's a matter of weighing cost vs benefit.

 

[ACharnley]

Thought so. The MCU is fast and has the pins. I'm unsure how accurate it'll be compared to an OP but I reckon it'll be reasonable enough.

The diodes are no more than $0.02 each but they are reasonably large (1210) and smaller is better. The thing here is I don't know if they're actually useful. I actually have some other FET's before these doing load matching and they present the same issue. One of them is in series as an "on" by default which will only pass the voltage to the active rectifier > 2V, thus the sckhotty dilemma appears academic.

When opens up another question, if in a 6V AC SIN the 0-2V is not passable how much power is unavailable.