Connect with us

varistor as snubber

Discussion in 'Electronic Design' started by artem_bond, Apr 27, 2004.

Scroll to continue with content
  1. artem_bond

    artem_bond Guest

    Hello all,

    I'm working on SMPS and I can't understand why people didn't used
    varistors in snubbers. I try to used varistor parallel to transformer
    primary coil and It's working well. I have no idea how I can check
    surge voltage on FET but I think varistor can decrease this value by
    volt-amps diagram and big spurious capacitance. But I can't find any
    mentions about using varistors in SMPS in snubber cell.
  2. Varistors are a bad idea for repetitive spikes, they have a "wear out"
    mechanism. Use silicon, not semiconductors that are made by grinding
    stuff up and baking it.

    Best regards,
    Spehro Pefhany
  3. Well, in the old TV (vacuum tube time) you would see HV varistors (VDR)
    across the vertical output a lot.
    Of all the hundreds / thausands I repaired I have -perhasp- seen one
    defective one.
    I have used VDR extensively to protect inductive switched loads, across
    mains as peak limiter, etc.., never a problem (used Siemens).
    Do you have data on failure rates?
    As to the OP
    In a switch mode, these things will dissipate some power.
    Better design you do not need them.
    But why should it not work...
    Have not used these in a switchmode.
  4. John Fields

    John Fields Guest

    MOVs are made by sintering conductive material in a non-conductive
    binder in such a way that when a the voltage across the MOV rises to a
    point which is high enough to arc between the conductive grains, the
    MOV presents a low impedance to the voltage, (much like when a gas
    tube ionizes) which then causes current to flow through the MOV until
    the voltage falls to a point where the internal arcing is quenched.
    However, because of the high local temperatures reached within the MOV
    because of the energy released within it, some of the conductive
    material will melt and will fuse together, forming conductive channels
    through the MOV. This is a phenomenon which is inherent in the device
    and the damage caused with each hit is cumulative in that with every
    hit the MOV takes, it becomes quiescently more and more conductive,
    until the point is reached where, with normal operating voltage across
    it its power dissipation capabilities are exceeded and it fails. It
    _always_ fails shorted, usually with spectacular results, and has been
    known to have started fires when it melts or explodes before finally
    clearing open.
  5. Yes, they recently revised UL standards to require a current limiting
    device in series with varistors, because of fires caused by thermal
    runaway in the nasty little things, just as you say. I've seen plenty
    of failures in ZnO varistors- Japanese, American and Taiwanese types.
    Their clamping characteristics are also pretty crummy. They certainly
    have their uses, but..

    Best regards,
    Spehro Pefhany
  6. Terry Given

    Terry Given Guest

    It is probably the varistor capacitance that is doing most of the work -
    look at the flyback waveform, if the peak is rounded the MOV isnt clamping;
    if its flat (or at least straight) the MOV is clamping. A simple test would
    be to replace the MOV with its equivalent (ie measured) capacitance, and see
    how your circuit behaves.

    A capacitive snubber presents a nasty turn-on load for the main power switch
    very true. IIRC Its more like a cumulative pulse energy rating - many little
    pulses, few short pulses etc. Each time a MOV clamps a signal, it is
    damaged; enough damage and it fails, initially short-circuit - so MOVs
    should ALWAYS have a protective fuse (somewhere in the circuit). Given
    enough energy, it will fail open-circuit (ie component gone)

    actual clamping voltage is fairly poorly controlled too, and the slope
    resistance tends to be quite a bit higher than a good tranzorb

  7. All very well, but in a vertical output stage the thing gets hits 50 times
    per second, that say 4 hours per day and for 10 years, and still working!
    The theoretical way *I* understood these things worked was certainly not
    I was under the impression that these things were semiconductor based.
    So I just looked up some papers.
    There is really ONLY significant degradation when used outside the specified
    peak current range.
    Have a read here perhaps:

    When used within range (specs) no 'arcing' or such will happen.
    It is good practice to measure the current (if repetitive) to see what
    the thing is actually (periodically) exposed to.
    The kA devices, when used as a 'snubber' in electronics will only get A or mA
    pulses so are way way below the rated kA surge.
    In such a case, no degradation occurs.
    In case of -main protection- from one time surges such as lightning, you will
    (very likely) have to replace... and a lot of other stuff too.
    In such a case teh series fuse is needed to break the circuit.
    But this was about snubbers (OP) and here I think these devices have proved
    reliable, at least in V output in TV.
  8. Also want to add, regarding to the OP, that in case of using
    VDR in switch mode, we need not only look at the heat dissipated,
    (that sets the lifetime in hours MTBF), but also at the capacitance.
    Thermal run-away is the most dominant failure mode
    (according to, read conclusion
    and see graph in figure 3 )
    for frequencies of 10 kHz or more I think it is very
    possible the capacitance of these devices will short / have some effect on
    the operation of the switch mode :)
    I have no real values here at hand, maybe someone knows the values.
    It does not seem to be in the data sheets.
    There are pspice models available for at least some varistors.

    Maybe that is why I never felt like using these in a 100kHz switcher...
    But perhaps this capacitance could be used to your advantage too.
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day