MOV or TVS to damp transient when offline flyback is switched ON?

[treez]

Hello, regarding the attached offline , 24W flyback schematic, would a MOV or TVS be better placed where indicated?
The purpose of the MOV (or TVS) is to staunch the overvoltage which happens due to the resonance of the leakage inductance of the common mode choke and the input capacitors.
(LTspice simulation also attached)

 

[KlausST]

Hi,

i'd prefer TVS, they are faster.
Additionally it's said that MOVs are getting lower ohmic every high energy pulse.
Last week i read about this, but i've never seen this effect with the MOVs i used...

Klaus

 

[dick_freebird]

I have read that MOVs degrade with each breakdown, but
that's a pretty old recollection. Maybe they are better than
back when they were young and sexy and I was paying
attention.

If you intend repetitive breakdown use then you want that
rating-basis, and you may have to dig for it.

 

[schmitt trigger]

There is an excellent Littelfuse app note, which compares the merits of the different clamping devices. They make all of the different types, so you cannot accuse them of having a vested interest of recommending one vs another.

For fastest response, which is your case, TVS diodes would be your best option.

 

[treez]

Hello
The MOV question is back...where should the MOV be placed...?

Page 12 of the following shows a 240W offline PSU schematic. The MOV is placed after the input LC filter.
Surely you agree that this is totally wrong?, and the MOV should appear after the input fuse?
After all, the input LC filter will ring up to a high voltage whenever the mains is applied at the instant of mains peak, and the MOV will end up quenching this ringing voltage…which will reduce its lifetime unnecessarily.
A MOV is simply to quench line overvoltage transients.

If it is required to damp down input LC filter overvoltage transients caused by ringing at application of mains at the instant of mains peak, then the filter inductors should be placed after the diode bridge and should feature diodes across them (cathode on input connector side)
Alternatively a TVS should be used following the mains input filer, though this is not recommended.

Please see page 12..
https://www.onsemi.com/pub_link/Collateral/DN05067-D.PDF

 

[chuckey]

These clamping devices work against the line impedance, more line impedance the better. if you connected one straight across the power grid, where the line impedance is extremely low, pulses would see it burn out rapidly as the let through current would be too high?
Frank

 

[treez]

yes I agree, but.....line transients above 380V are rare....wheras power-up transients happen every time the power supply is plugged in, which could be several times a day...this is why the MOV must go at the front end..surely?

 

[FvM]

I don't believe that you are discussing a real problem. Most offline SMPS input filters are designed similar to the page 12 schematic and apparently run well.

The filter should be designed to pass respective EN61000-4-x tests and I'm not sure if the suggested "improvement" would serve this purpose.

 

[treez]

the attached LTspice simulation shows the problem that the MOV in this position faces...you can see in the location where the MOV is, it has to quench a voltage that would ring up to 490V.
That is going to reduce the MOVs lifetime unless it is made beefy and expensive.

 

[FvM]

I don't believe that the switch-on transient involves much stress for the MOV. It's energy is only a few percent of the MOV rating. The more interesting question is how fast the energy will be absorbed by the MOV and what's the peak voltage with connected MOV. You would add a MOV model to the simulation.

 

[treez]

Thanks, as you allude, the MOV is quenching the input filter ring. That job should be done by a TVS....placed where they put the MOV, ie after the LC input filter.
The MOV should go just after the fuse, and is not for quenching input filter ringing, but for quenching short 1KV transients which occur from time to time on the mains.

 

[treez]

Woops my apologies, I have put the phasing of the common mode choke the wrong way round......here is the corrected simulation (LTspice) version, actually , the problem is still just as bad so the above questions remain the same