Over voltage Protection Circuit without using Fuse

[Pankaj Khatri]

Can anyone tell me how this circuit is working?

I have taken this circuit from.. https://www.maximintegrated.com/cn/app-notes/index.mvp/id/760

 

[MikeMl]

What is wrong with the explanation of How it Works in the .pdf?

 

[Grossel]

If you're somehow familiary with BJT's, then I suggest you try to draw it in opposite sign - that is swap all polarity of components and supply voltage.

That makes it much easier to understand, since people just are more used to NPN. Or somewhat in same range of problems if one are used to measure in milimetres and suddenly have to deal with inches.

 

[ACharnley]

I'm trying to design something similar only for AC. It's proving more troublesome than I imagined, I keep running into using P FET's as voltage followers. Anyone already cracked this?

 

[ACharnley]

Here's what I have so far.

 

[Grossel]

For AC, you'll probably look for a solution that involves a current transformer, or if three phase you'll need three CT's.

 

[ACharnley]

Sounds lossy, in the perfect world it would be FET based.

 

[dknguyen]

Sounds lossy, in the perfect world it would be FET based.

Technically, in a perfect world it would be able to handle infinite overvoltage and overcurrent.

 

[ronsimpson]

I'm trying to design something similar only for AC.

Looking at the circuit, you have a limiter for the positive part of the AC signal, also a limiter for the negative part of the AC.
If you put one limiter after the bridge you only need one.

 

[ACharnley]

It doesn't work anyhow, I went back to the drawing board and... this works perfectly!

I've spec'd it to work up to 100V and cut off is 30V (play with the 18.8k resistors). At 100V there's about 6mA pull at the LM393 which is the rated max. Finding an ideal P-FET is more troublesome but FDMC86139P looks good. I couldn't get it to sim with their spice model though.

If anyone can assist with the over-voltage, you can see that I've added protection for the FET's but I'd appreciate an extra eyes on. Sim says it's ok.

More of a worry, the transistor collector-base will see 100v which may overstress it despite the 22k resistor. I added D12 as a fail safe but doubt it does anything useful.

An alternative would be to control each FET separately using two comparators. A few less diodes but more complexity.

 

[ACharnley]

And also R7 shouldn't be required but it won't simulate without it.

 

[dknguyen]

And also R7 shouldn't be required but it won't simulate without it.

LM339 is open-collector so R7 is required (or at least, good practice in a real circuit). For the simulation it may be more important since it needs a number to work with and a floating copper wire can't be handled numerically.

 

[ACharnley]

Yup, but it's draining D12. I'd not do it in a circuit here unless using a FET.

 

[eTech]

And also R7 shouldn't be required but it won't simulate without it.

It should be required because if it’s missing the output will switch from open circuit to V-.

 

[dknguyen]

Yup, but it's draining D12. I'd not do it in a circuit here unless using a FET.

You're missing the point. D12 has nothing to do with it. Tell me, what voltage do you think the simulator should assign to the output of LM339 is not outputting a LO (i.e. when the output goes high impedance).

Do you just expect things to simulate when there are indeterminate values in the system?

 

[ACharnley]

Leakage from the comparator or transistor pin?

After working with LTSpice for a few months I now use resistors and capacitors (that would be omitted in real-life) to make simulations run.