Modeling an MOV?

It probably isn't single-valued, so you might need a subcircuit with a few flip flops in it. ;)

Cheers

Phil Hobbs

Try different search terms? I'll bet that physicists don't think "resistance vs. voltage". I'll bet they think "current vs. voltage". So maybe search on "current in metal oxide varistors"?

What does Wikipedia say?

I'll bet that when you finally get down to the nitty-gritty, you'll discover that there's a bazillion different effects, and for any given MOV you just have to do a best fit to measured data. But I'm a skeptic.

It snaps back? I've modeled BVceo snap back without flip-flops, so, if I can get a definitive equation (or curve), I can likely model it. ...Jim Thompson

Here's one:-

Basically just a voltage dependent resistance plus some parasitics- they don't really have hysteresis.

Actually, Wikipedia has some curves, which pretty much back up my hypothesis that it depends on the material. Think "lots of back to back diodes in series-parallel", and you get an idea of the behavior.

Jim Thompson a écrit :

Resistance vs voltage???

V= A I^B + R I from what one can eyeball from the curves in

and you'll be close enough. Take a few points, then LS fit.

If you really need the differential resistance, then derivation is left to the student.

Have just done it for the V150LA20A (fig 14) and parameters are

a=338.334, b=0.0320318, r=0.0462244

A plot looks just OK.

Fred Bartoli a écrit :

If one look at fig 9 you see that there are several power regimes, so for the 320LA10, one might try fitting with

a i^b + c i^d + r i

Thanks! Should have known... Brian Hirasuna was one of PSpice's best support engineers, along with Charlie Edmondson (and Mohi :-)... they actually know simulators _and_ circuits! ...Jim Thompson

=A0 =A0 =A0 ...Jim Thompson

Speaking of MOVs. Can I just hookup three of them between Hot, Neutral and Ground wires? I brought a single outlet surge protector for my daughter, but she refuse to carry another item with her laptop. I might have to build a smaller one directly into the power bar or plug. Otherwise, her laptop will be the surge protector.

For the limited number of MOVs that I have modelled, it appears that the simpler model above works quite nicely. A bit of resistance helps to fit the upturn at the end of the curve quite nicely.

on

=A0 =A0 =A0 ...Jim Thompson

..

Hello Linnix,

MOVs can fail, so you have to assess safety. You may download application info from some manufacturers (like Epcos).

Some reading may save you from putting yourself or your family in danger.

Wim PA3DJS

tion

=A0 =A0 =A0 =A0 ...Jim Thompson

n...

Hopefully, it will just fail and remove itself from the circuit. She just won't carry another box/plug. I am trying to tap (and insulate) three MOVs on the power cord itself, to make it as small as possible. I will go read more about it. What other components do I need?

uoted text -

he's trying to tell you that MOVs across the line can catch fire so take care of how/where you enclose them.

Mark

text -

They generally fail to a dead short- unless/until there is enough current available to blow them open, which involves a certain amount of charring etc.

e quoted text -

OK, thanks. I can add themofuses to them. I might put them in a power switch, so she can switch it off as well.

equation

IIRC the V-I curves look a lot like those of neon lamps. There are some similarities in the physical processes.

They wear out too? Or is it they they're damaged by taking a heavy hit?

How to model that?

Grant.

If that doesn't do it, Epcos datasheet for MOV "Series/Type: S05, S07, S10, S14, S20" claim to have PSpice models in a bullet point.

I've not gone chasing those models, as I'm repairing a 200A MOSFET bridge from a friend's fried welder -- six blown (as in some are cratered) 80A MOSFETs. And five of six 22R gate resistors went open, which makes me wonder how far back the damage goes, the dual gate driver tests (in circuit probing) same as other channel (TC4427).

Grant.

I just tried the Brian Hirasuna "model"... PSpice barfs... too much crap ;-)

My suspicions are that my favorite function, TANH, may be the best fit, since all derivatives exist. ...Jim Thompson