Diode Failure

I can't answer your question. If you are exceeding the absolute maximun rating of the part, you're asking for trouble.

Got a schematic?

Best regards, Spehro Pefhany

Where did you ever get such an idea? Your design is faulty.

You may not be able to see the damage from the outside. A brief very high current pulse can have enough energy to damage the P-N guard ring or the junction. That particular part does not have a reverse avalanche energy rating on the datasheet.

Got a schematic?

Best regards, Spehro Pefhany

I don't operate them outside the design voltage, I am just trying to think of any possible reason for the failures.

Ralph

Then I'm confused. If you never exceed the absolute maximum specs, they should never fail. Your first post implied that you thought you could get away with exceeding the reverse voltage specification and get away with it if there were no heat damage.

Correct me if I'm wrong, but if you apply a high enough reverse voltage to cause them to breakdown, aren't you exceeding the abs-max reverse voltage specification?

Time to ask a better question.

it?

a) defective part b) incorrect design that failed to account for something such as in-rush current c) environmental damage

d) (possibly included under hazards of c)) something unforseen that this particular customer is doing to the product. Never underestimate the creativity of customers in finding design or instruction manual weaknesses.

Best regards, Spehro Pefhany

think

Perhaps: Diode broken, why? Is that a good question?

I don't knowingly run the diode in reverse break down or exceed the voltage spec of the diode. I do know that all of a sudden the diode is failing and so am trying to figure out why. My understanding leads me to believe that reverse breakdown is non fatal to a diode.

Ralph

customer

under

will

it?

It's pretty simple. Used for an automotive supply.

--------~~~------|>----------~|~--- | IGN | |

--------~~~------|>-------

The always on and a switched are feed into one via the diodes. It's the ign diode that is dead. The fuses before the diodes are 5A and the polyswitch is 1.3A.

The only other thing not shown here is a high impedance voltage divider before the diodes on each inputs used to measure the incoming voltage.

That's about all there is to it.

Thanks for any thoughts

Ralph

I'm not surprised. The IGN circuit is feeding something inductive, and depending on dI/dT effects to generate large voltages. You probably need kilovolts of PIV for that diode. Automobiles are noisy places.

ign

Is this connected to the ignition relay, perchance?

I think this should really belong to b) -- a good design should take "creative" (or stupid) customers into account, too. ;-)

Maybe a field-decay transient. These can have a lot of energy. You can't stop this sort of thing, just put in parts with enough rating to avoid the issue (or suppress it, but that may be more expensive). On automotive supplies it's possible to see ~ -400V /+150V transients with various amounts of energy depending on the source. You can't get Schottky diodes with that high reverse breakdown voltage.

If you're going to be making thousands of these things whoever is doing the design might want to get ahold of the relevant standards so they won't be guessing as to automotive requirements.

Best regards, Spehro Pefhany

I am not sure about the details, but car supplies are notorious for very dirty voltages. Big transients that last longer than your usual spike etc.

Did all the failures happen on the same type of car or even the very same car? Might be that this particular car has a badly filtered ignition or something as simple as a weak battery that is not as good at absorbing spikes as it used to. Also cable routing and bad grounding (rust) are common problems.

But as you probably can not control how the device is installed and in which car, your safest option is to use more rugged diodes en maybe even some filtering and/or transzorbs.

IIRC, I read somewhere in this thread that de reverse rating of these diodes is only 40V, which seems very low for this application.

"Should", maybe, but that would neglect one of the oldest theorems of engineering: trying to design things to be fool-proof is futile --- nature's response to such efforts is to evolve better fools, and you can't beat evolution in the long run.

It's impossible to make anything absolutely foolproof. Somewhere out there is a fool who's smarter than you are...

Regards,

-=Dave

Also, by the time you've figured it all out, you're overqualified and should have moved on..

Best regards, Spehro Pefhany