Connect with us

Diode Failure Mechanism & Mode

Discussion in 'Electronic Design' started by RST Engineering \(jw\), Jun 8, 2006.

Scroll to continue with content
  1. Suppose I have a popcorn glass diode like a 1N4148. I will slowly advance
    the temperature so as to not thermally shock the glass.

    At what temperature (or range of temperatures) would you expect the diode to

    What is the mechanism for the failure (thermal expansion, solder melt,

    What is the mode of failure (short, open, gross deterioration, etc.)?


  2. Guest

    Depends whether you are applying a reverse voltage or not. If the diode
    is back-biased, the leakage currents doubles every 10C as you raise the
    temperature, and eventually the current reaches a level where the heat
    dissipation within the junction is high enough for the junction
    temperature to run away.

    The diode then fails as a short circuit. With a suffiiciently potent
    power supply, the diode can be vapourised - I did this once and all
    that was left of the diode was two rounded ends on the bits of wire at
    either end of where the diode used to be.

    With no reverse voltage, you can probably get the diode hot enough for
    the dopants that create the N- and P-type areas in the diode to diffuse
    across the junction, wrecking the semi-conducting behaviour that we
    rely on. Less exciting and much slower.
  3. tlbs101

    tlbs101 Guest

    According to empirically compiled data over many years (decades?), Mil
    Handbook 338 lists 3 modes of failure for diodes (such as the 1N4148)
    and the likelyhood of each:

    Short Circuit -- 75%
    Intermittent -- 18%
    Open circuit -- 6%
    I suppose "other" would be the remaining 1%, or the numbers were
    truncated after the decimal place.

    These are failures under normal operating circumstances. At higher
    temperatures, I don't know how the probabilites would change, but I
    don't believe the percentages would change that much.

    As for mechanisms, you would have to get a materials expert to answer
    that. I have personally seen overheated diodes "go" both ways, most
    recently being a zener that shorted out.

  4. Approximately 1 milliampere of forward current. 5 volt supply with 4.7k
    resistor in series, with the diode far away from the current source.

  5. All that can be said about this IS: Whoever believes that components
    fail in "certain" fashion does live in science fiction.
    The best we have is some statistical numbers because you simple cannot
    predict what happens in field and Mr. Murphy can always proove you wrong.

    Have fun

    Slack user from Ulladulla.
  6. Joerg

    Joerg Guest

    Hello Tom,
    More than 1% of the bad thru-hole era diodes I found had cracks in the
    glass. Some of them were missing chunks of glass.
  7. Jim Thompson

    Jim Thompson Guest

    Same as the TO-220 lead bending failures. Somewhere around here in
    one of my tool boxes I have a red plastic strip that you drop
    resistors (or diodes) into a slot with the lead spacing you want and
    bend without stress.

    ...Jim Thompson
  8. Pooh Bear

    Pooh Bear Guest

    Mine was blue. No idea where it is now.

  9. John Fields

    John Fields Guest

    Even that's no guarantee. The 100% right way to do it is to grab
    the lead with a pair of longnose pliers in one hand and then to bend
    the free (non-diode) end of the lead where it exits the pliers.

    Same way with T0-220s; grab the lead where it necks down on the
    package side of the lead and then bend the free end of it down.

    Start with the middle lead. :)
  10. Jim Thompson

    Jim Thompson Guest

    Same here. I was going to post a picture, but I couldn't find it :-(

    ...Jim Thompson
  11. OOOOOOOOOkay. Let me try again. The diode is being operated forward bias
    at about a milliampere. It has been handled with kid gloves from the day it
    left the factory until it was carefully installed with no stress on the
    leads into the circuit. I put the sucker into an oven and start slowly
    bringing the temperature up.

    At what temperature (or AROUND what temperature) would you expect the diode
    to reach an elastic limit of some sort, either physical or electrical, and
    not come back to approximately the same forward voltage at room temperature
    than it had before the thermal stress test? What do you SUSPECT might be
    the mode of the failure.

  12. Robert Baer

    Robert Baer Guest

    Define "fail".
    Leakage greater than a specified current at specified voltage?
    Avalanche breakdown voltage change more than x percent WET room themp
    value when measuerd at a specified current?
    Dump the diode into an oven pre-heated to 200C and find that it is
    still useable 10 seconds later when it is more than thermally stabilized.
    Note i did not define "useable".
  13. Robert Baer

    Robert Baer Guest

    Under those conditions, 200C is just as good as 20C or -75C.
  14. Robert Baer

    Robert Baer Guest

    ...and so, if you want to decrease collision with Mr Murphy, then revise
    the design for Justin Case...
  15. Robert Baer

    Robert Baer Guest

    And that was due to added stress in the glass-to-metal seal; the
    extrastress caused by: 1) socketed type testing (proven in MIL
    reliability tests at FSC in the 80's), 2) assembly (no stress relief
    when leads were bent for board insertion, and 3) no stress relief added
    in the shaping of the leads (see NASA and other documents showing a "U"
    type "hook" on one end).
  16. Robert Baer

    Robert Baer Guest

    I also have one of those, and they are very handy.
    They do *not* eliminate stress, in fact, one could show that they do
    little to reduce stress at the lead-body interface, which is critical
    for glass encased devices.
  17. Robert Baer

    Robert Baer Guest

    Excellent start, but better if a gap is left between the part and the
    Do not remember the recommended minimum gap.
  18. Robert Baer

    Robert Baer Guest

    The picture, the post, or the horse tied to the post?
  19. Jim Thompson

    Jim Thompson Guest

    "The horse may talk" ;-)

    ...Jim Thompson
  20. In addition to the other info already presented, I suspect it would
    vary (maybe greatly) with the *brand* of diode used as well.

    Dave :)
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