Connect with us

What fails a FET?

Discussion in 'Electronic Design' started by Andrew, Oct 25, 2006.

Scroll to continue with content
  1. Andrew

    Andrew Guest

    I have had a circuit working for some time now that has worked great.
    A particular portion uses a 2N7000 FET to invert a 5V signal, to 12V
    (inverted). The input is from a servo driver, outputting about 4.8V,
    directly to the gate. At the drain is a 10k resistor connected to the
    power supply, which is ~11V to ~14V depending on battery charge. At
    the resistor/drain node, the output is fed as clock to some CMOS chips.
    This clock feed has a 470pF filtering capacitor to ground also, if
    that is significant. The source is connected directly to ground.

    At any rate, this setup has worked great, but I have since done another
    revision of the board to clean up a few areas that should reduce noise.
    No changes were made to the schematic or even component placement,
    just routing.

    Within the first hour of operating this new board (the old boards, 4 of
    them, have at least 40 hours each, probably alot more on some of them,
    without any single problem), the transistor I described earlier failed.
    When viewing the (usually 4.8V peak) signal that is being inverted, at
    the point that it enters my board, it was about 100mV peak. The square
    nature of the signal was in tact, but the voltage was VERY low, so low
    that it would not trigger the FET.

    I measured the resistance between the Gate and Source and found that it
    was low - about 150 ohms +/- 10 ohms. I replaced it, and the circuit
    worked great again, for about another 20 minutes, then, the same thing
    happened!! Measured the resistance of the transistor (which, in a
    "good" circuit measures in the many many mega ohm range) and it was
    again, in the 150 ohm range. This explains the 100mV peak, anyway.

    So, I switched to a different board (same new rev, just a different
    physical board). During comparisons of the "bad" board and "good"
    board, somehow the second board's transistor blew too! Same problem.
    I replaced it, and decided to run it with a new servo driver board
    also, and now, for 2 hours it has been perfectly fine, but needless to
    say I am skeptical. I am trying to find a cause for the failure of
    this FET and I can't find one.

    So, that brings me to the point of this post. What are things that can
    cause a FET to fail? My guesses are over current the drain to source,
    or over voltage the gate to source. Those are my guesses anyway, but
    both are pretty impractical as the drain source current is limited by
    the 10k resistor (and at 12V, is way way way under the 800mA rating at
    Vgs = 5V), and the Vgs can be up to 20V and is at around 5V. Is there
    a way to tell what would cause the gate to source resistance to fail in
    the way I described?

    My guesses as to the cause of the failure are that a) there is a
    problem with the first board I tested, or b) there was some problem,
    unrelated to my board, from the input 0-4.8V servo driver board that
    blew the transistor. The power supply to the servo driver board is 6V,
    though, so I don't know how it would get above 20V, which is the max
    rating for Vgs.

    Anyway, how can you kill a FET, and which way of killing one (if any)
    will result in a very low gate to source resistance?

    Thanks in advance.
  2. Guest

    Maybe the gate voltage is being exceeded due to some noise somewhere?
    Maybe terminate the gate/source with a 10-50K resistor would help?

  3. Guest

    A shorted gate means the gate oxide has broken down, which implies
    excessive Vgs. That, in turn, could be from the circuit, or ESD

    James Arthur
  4. Joerg

    Joerg Guest

    Hello James,
    I'd second that. To prevent it Andrew could look up the max Vgs rating
    (the 2N7000 is usually 20V), then place a resistor between driving
    source and FET gate, preferably close to the gate. 10k or so, depending
    on how fast the signal needs to be. Then run one diode to the 14V supply
    and another to ground which limits the excursion to a range of -0.7V to
    14.7V. A BAV99 is nice for that if space is tight.
  5. Guest

    Yes, I've blown a few mosfets with excessive gate voltage. Adding a 10K
    resistor from gate to source usually fixes the ESD problems.

    BTW, the 16 white LED driver is working well at 86% efficiency. It's
    almost ready to mount inside the flashlight. I have it running here
    projecting a beam of light on the wall. Tomorrow will tell if it fits
    inside the housing.

  6. Guest

    Thanks for the update--let the blazing begin.

  7. lemonjuice

    lemonjuice Guest

    Antistatic soldering stations?
    ESD protection and overvoltage protection?
    Clamping using reverse biased zeners will lower performance but it
    will protect so will decoupling and the like.
    Many things ... can kill a FET ... they like BJTs can melt. A
    touch of a finger on the gate to internal ruptures can kill them...
    High temperatures can open up a lead bond in a circuit leading the
    circuit to fail then someone may try testing a FET... but that is
    something really to avoid your very last shot because you can SO
    easily kill it. As said above the thin oxide layer at the gate
    electrode ensures high input impedance but also makes it vulnerable to
    ESD. You need a low voltage ohm meter when testing it plus it has to
    be set to the highest resistance reading.

  8. neon


    Oct 21, 2006
    You should not be checking FET with meters period it could damage the FET Unlike transistors they get damaged from even cleaning the board with paper towels and they unlike transistors may work for a while and badabing it goes bad.
  9. Andrew

    Andrew Guest

    Thanks all for your input. I'm still a little new to effectively using
    transistors and the things that can happen with them in the "real"
    world. I didn't take any ESD care whatsoever when handling the
    transistors and when soldering. It is poor practice that experience
    has never forced me to change. Guess I've been lucky so far. I didn't
    realize transistors were so sensitive to ESD. This could certainly be
    what happened, and would explain what happened to the transistor on the
    board that failed before any minutes/hours were even put on it (besides
    testing - some of which was done with a DMM). I have plenty of room to
    try a 10k resistor from gate to source (which would be gate to ground,
    in my case) if the working board doesn't hold up. I'll be testing more
    this week so I'll try to update everyone.

    I'm curiou, though, is it possible to damage a transistor with ESD
    while handling/soldering, then install it, have it work perfect for a
    few hours, then fail suddenly because of the original ESD damage prior
    to/during install? Or would a failure like this be unlikely to be ESD
    and more likely to be a circuit problem?

    I really don't think I am over-voltaging the gate. With 4.8V max going
    in now, a 20V max Vgs, and no noise to speak of, I find it unlikely
    that it was over-voltaged, but I could be wrong.

    Thanks again for all your help!
  10. Andrew

    Andrew Guest

    One more thing I thought of, now that we are on the ESD topic (sort
    of). With the current board (as opposed to previous) I am now using a
    foam packing type insert to press the board against its plastic case,
    instead of letting it free float. The entire plastic box that houses
    the device can undergo some serious vibration. Is it possible that
    this foam pad is generating ESD or something? I don't know anything
    about ESD, I'm just thinking outloud. It is contacting the bottom of
    the board, where there are no compontents, but all through hole
    connections and vias (the transistor in question is through hole) are
    being contacted by the foam piece.

    Any chance this is related? Or very unlikely?
  11. MooseFET

    MooseFET Guest

    Andrew wrote:
    If two different material rub over each other, the electrons from one
    will tend to transfer to the other and build up a charge. You don't
    get a high voltage out of it until you move the charges apart. I would
    worry about anything that allows one plastic to move against another.
    It wouldn't be just the foam but also the plastic housing against
    mounting parts.

    You want to make sure that you have a much shorter spark path to
    something non-harmful and/or make sure that none of the parts are good

    Very often you end up solving the static problem when you solve the EMI
    problem. The conductive material you put in as shielding also works to
    take static charges away. You still need to make sure that the plasic
    foam between the metal and the PCB can't build up a charge.
  12. Andrew

    Andrew Guest

    I might have been a bit unclear. It is a plastic BUD type box with the
    circuit board inside. I don't know the composition of the foam piece,
    but it is foam and is stuffed between the lid of the plastic box and
    the bottom of the circuit board.

    I really doubt this is the problem, though, because I have "fried" one
    of these transistors without having the board in the box.

    I also experimented today with purposely trying to fry a transistor in
    the fashion that the ones are failing on my board, and believe it or
    not, I had some trouble. I put 37V (max from single power supply that
    we have) from gate to source and left it for 15 minutes or so. It was
    alright. I tried pulsing it and that didn't work either. I put the
    37V from source to gate instead and still it was alright both pulsed
    and continuous. I linked two supplies together and put up to 60V to
    both gate to source and source to gate, pulsed and continuous (for a
    few maybe 30 seconds) and it was still alright. I put 72V (max of both
    supplies in series) from gate to source and this did not take long at
    all to "fry" the transistor. The resistance from gate to source was
    very similar to the results of the transistors that failed on my board.

    With this type of voltage to fry these things, unless there is some
    other serious problem, I must have damaged the 3 transistors that
    failed in boards with ESD somehow. I seriously doubt (certainly have
    not seen it) that there is enough noise on the 4.8V signal to achieve
  13. Andrew

    Andrew Guest

    Wow. I'm stunned. I can reliably get one of these transistors to fail
    with ESD. There's nothing like seeing is believing! For anyone who is
    skeptical that ESD easily fails transistors, here is the experiment I
    just fried 5 transistors in a row easily doing.

    I grounded the source terminal to earth ground (used the one on the
    power supply). I stood on my padded chair, rubbed my shoes on the
    (nylon?) padded chair a couple times (don't need many at all) and
    touched the gate. One time out of 5 I felt/heard/thought I saw the
    shock, the other 4, I thought nothing happened. Tested resistance
    between gate and source, and BOOM, it was fried... very low resistance.

    I then considered that the soldering iron tip is earth ground and
    figured that it wouldn't surprise me for a second that that his how I
    damaged some transistors (that eventually prematurely failed), and
    fried other transistors on the spot. To confirm this theory, I held
    the gate pin with one hand, the soldering iron (on the rubber
    boot/handle) with the other, stood on the padded chair, then rubbed my
    feet a bit on the padded chair while intermittently touching the
    soldering iron tip to the source. Sure enough, after a couple tries
    the transistor was fried.

    Although I don't solder while standing on my padded chair, I can easily
    see how ESD is transferred to devices. I may have transferred a
    smaller charge while standing on the linoleum floor and soldering that
    simply damaged the FET. This damage, from what I've read, may often
    times still allow the chip to work alright - for a while - then fail,
    which is what some of my boards/transistors have done.

    ESD is much more serious than I thought. Regarding FETs and other
    sensitive devices at least. Needless to say I'll be taking MUCH more
    care in the future.
  14. Andrew

    Andrew Guest

    Sorry there are so many posts by me, but I'm trying to fill everyone in
    and keep some info in here for future searchers.

    I just wanted to note that right after I posted, I went to see if I
    could fry a transistor by holding the gate pin, rubbing my feet on the
    linoleum, and touching the soldering iron tip to the source pin. Sure
    enough, first try, I fried it. I'm really beginning to think this is
    case closed!

    Thanks for your help again everyone.
  15. PPP

    PPP Guest

    Thanks for your updates!

    I'm new to electronics design, and this info is very helpful for my
    future design endeavors.
  16. mark

    mark Guest

    I have one article re ESD damage where the authors claim that ESD damage is
    a thermal effect and therefore latent damage is almost non-existant. Don't
    believe it really, but have used the article to support the argument when
    my customer has asked me the same question.

    What are other people finding re latent damage?

    M Walter
  17. Andrew

    Andrew Guest

    This is the article that convinced me that there could be latent

    Seems plausable. An ESD spark can damage the chip in the same way that
    welding or arcing can. Check out the picture in that writeup, it makes
    sense to me. The chip works fine until it gets some time on it,
    excessive heat, or slightly more current than the damaged track will
    allow, which would have otherwise be acceptable on an undamaged chip.
  18. mark

    mark Guest

    Andrew wrote:

    Interesting article, thanks

    M Walter
  19. neon


    Oct 21, 2006
    you have not learn a thing plastic of any kind is HOT.
  20. Andrew

    Andrew Guest

    Further regarding latent damage... I just did another experiment where
    I tried to damage a transistor in the same fashion(s) and I had one
    that failed where the gate to source resistance measured like 9 mega
    ohms or so. Typically, when the transistors were failing, they were
    reading approx. 120 ohms, and a "good" or "new" one usually reads open
    on my fluke. This 9 mega ohm gate to source transistor appeared to
    work "alright" (no further damage done anyway) at 5 volts or so, and
    even at 10 volts, but once I put 37 volts (skipped to it, should have
    incremented, doh!) it blew instantly (so that gate to source resistance
    read 120 ohms, same as the other "blown" ones). As I mentioned before,
    even though they aren't rated for it continuous, the "new" "good"
    transistors will take the 37 volts no problem, for at least 15 seconds
    (most time I've done it for, but I've never blown a "good" one with 37
    volts before).

    Seems like proof to me that ESD can damage a chip such that it might
    *seem* ok, but will fail where an undamaged chip wouldn't.
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