Connect with us

Relay Suppression Diode Failure

Discussion in 'Electronic Design' started by Mark P, Jun 18, 2007.

Scroll to continue with content
  1. Mark P

    Mark P Guest

    I trying to determine the cause of a relay suppression diode failure.
    The diode is built into the relay can (T-05 type) and is a standard
    switching type diode - according to the manufacturer. The relay coil
    specs: L=600mH and R=850 ohms. The relay was driven with a 2 second
    ON pulse followed by 2 seconds OFF. This duty cycle was continued for
    three minutes. The relay was driven with a transistor switch on the
    low side. The voltage on the coil was 13V. The relay failed and after
    it was opened up you could see that the diode was cooked.

    When the transistor switch turns OFF the diode suppresses the voltage
    transient. Is there sufficient energy dissipated in the diode over 3
    minutes to cause it to fail? Is there a way to calculate or estimate
    the diode junction temp rise? The manufacturer has no thermal data on
    the part.
     
  2. Jim Thompson

    Jim Thompson Guest

    Only 70uJ, unless your measured numbers are wrong. Perhaps relay
    installed backwards so that diode is forward-biased when relay is
    powered?

    ...Jim Thompson
     
  3. Thomas Magma

    Thomas Magma Guest

    Usually when a component 'cooks', it is due to current and not voltage. When
    the relay switches off and the coil collapses, it produces high voltage but
    does not have high current potential due to the lack of capacitance.

    I would suggest double checking to see if the relay coil is wired in the
    right polarity.

    Thomas
     
  4. Just to keep thermal things in perspective:

    The average power dumped into the resistance of the coil is
    close to
    1/2 * (V^2/R)= 99 mW. The peak energy dumped into the diode
    is very approximately 11 mW (15 mA times about .7 volts
    drop), but the average power dumped into the diode over the
    4 second cycle is only about 1.6 uW (by simulation), because
    most of the stored inductive energy is dissipated in the
    coil resistance during discharge. So the temperature rise
    in the diode, is far more likely to be caused by its close
    proximity to a coil dissipating about 6000 times as much
    power as the diode is. All this assumes that the diode was
    not defective, to start with.

    Even if you activated the coil every 0.004 second, for 0.002
    second on time, the ratio only drops to about 46 to 1.

    I think you might measure the copper coil resistance during
    the test to use it as a temperature monitor.
     
  5. nospam

    nospam Guest

    The energy stored in the relay inductance is i^2L/2.

    13v and 850 ohms gives 15.3mA.

    15.3mA in 600mH gives 70uJ of stored energy.

    If the diode had to absorb all the stored energy (which it doesn't) the
    power dissipated would be 70uJ every 4 seconds or 15.5uW.

    The estimated junction temperature rise from dissipating 15.5uW is zero.

    More likely you had transient voltage spikes on the 13v supply (possibly
    caused by whatever you were switching with the relay) which caused the
    diode to break down and then cook because it and the transistor were
    shorting the 13v supply. That or you had the coil wired the wrong way
    round and were trying to short out the supply with the diode from the
    beginning.




    --
     
  6. (snip)
    Okay, please tell me what what duty cycle and frequency
    produces the most power dissipation in this diode. And what
    power is the diode dissipating under those conditions?
     
  7. Jamie

    Jamie Guest

    I stated this once before but some one kind of thought I was a fruit cake.
    The fly back energy released from the coil produces current in the
    diode, if you keep pulsing the relay, the diode does not have any time
    to cool down., it will at some point avalanche due to heat etc.

    Most likely if you look at the spec's you may find more specific data
    on the relay about duty cycles., if that data isn't there, then the
    OEM didn't do their home work.
     
  8. Jamie

    Jamie Guest

    Hmm, we use relays at work (DC) that employ AC type TVS diodes to clamp
    the coil, above the rated voltage on the relay. they don't seem to have
    a polarity issue.

    Now, I have seen a few times in machines we have a circuit that now
    and then goes hay wire and causes one of these relays to chatter until
    the machine powers down or the diode shorts and then blows the control
    fuse which then nicely powers the machine down for you! :)
     
  9. Eeyore

    Eeyore Guest

    I've known small relays to 'take out' 1N914/4148 small signal diodes. I always
    use a 1N400x type now.

    I don't think it's average energy btw, but peak current.

    Presumably if you could persuade the driving transistor to turn off more slowly
    it might be ok.

    Graham
     
  10. Fred Bloggs

    Fred Bloggs Guest

    Almost certainly the case, and the "proper" polarity was determined by
    an old meter movement ohmmeter where the red lead (+) is negative and
    the black lead (-) is positive:)
     
  11. Or perhaps the PCB layout dude/dudette was more familiar with modern
    component layouts that generally show the part from the *top*.
    Everything else is generally symmetrical on TO-5 relays. ;-)


    Best regards,
    Spehro Pefhany
     
  12. Jasen

    Jasen Guest

    yeah but 15 pulses per minute isn't going to be much of a strain on
    that.

    Bye.
    Jasen
     
  13. Depends on the ambient temperature and how many microKelvins you are
    away from destroying the junction... it could just be the nanostraw
    that breaks the camel's back.


    Best regards,
    Spehro Pefhany
     
  14. Mark P

    Mark P Guest

    Thanks for all the inputs. I also calculated the energy dissipated in
    the diode and determined it was very low. The relay had to be in
    correctly, because it worked for 3+ minutes before failing. After the
    failure the relay, you coild still operate the coil, but debris from
    the exploding diode had made its way between the contacts.

    I start looking at other causes.
     
  15. Jim Thompson

    Jim Thompson Guest

    THAT is NOT a correct conclusion. It just took that long to cook the
    diode. The relay (diode) IS in backwards.
    ...Jim Thompson
     
  16. How do you know that the diode didn't fail, immediately, but
    the contacts didn't get contaminated enough to fail until 3
    minutes later?
     
  17. Jim Thompson

    Jim Thompson Guest

    The OP needs to learn how to use an Ohm-meter ;-)

    ...Jim Thompson
     
  18. The Phantom

    The Phantom Guest

    Use the relay which has already failed to perform further tests.

    Clip out the damaged suppression diode, and solder in a new 1N4148 or
    equivalent, in series with a 1 ohm resistor. Clean the debris from between
    the contacts.

    Operate the relay in the 2 second on, 2 second off regime. See if the
    diode overheats; feel it with your finger but don't burn yourself. It
    shouldn't heat up at all.

    For a more sophisticated test, get a 2 channel (or 4 channel) oscilloscope
    which can do trace math. Connect one channel across the diode to monitor
    diode voltage. Connect the other channel across the 1 ohm resistor to
    monitor diode current. Set the scope to display the product of the two
    measurements. Monitor the V x I trace for exceptionally high spikes. If
    they occur, find out why.
     
  19. Mark P

    Mark P Guest

    I guess its possible, This is a high rel application. (Military
    relay) The diode is contained inside the relay can. It passed in
    house tests at the manufacturer and was then screened by an
    independent QA house before being installed. It also passed
    functional testing, although I don't know the duration of those tests.
     
  20. Jim Thompson

    Jim Thompson Guest

    Take an un-mounted relay in your hand. Ohm the coil... both
    polarities... to determine diode direction. Operating direction
    should be cathode to plus.

    My bet is that you have it backwards.

    ...Jim Thompson
     
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

-