0
\$\begingroup\$

I have one bridge rectifier operating at the constant junction temperature of ~72°C, (GBPC3510W datasheet), which is rising to ~80°C in some specific cases.

I'm trying to find a kind of "aging curve" which shows the lifetime of the bridge rectifier in function of the junction temperature. Since there's no such thing specified in the datasheet, I thought it would have been easy to find a "general curve" for PN junctions on the internet. It seems to be a common thing to specify in LEDs but unfortunately the only thing I could find related to "normal" diodes is the 10°C rule of thumb explained here. This rule requires you to know the lifetime of the component, and is not useful in my case.

So my questions is: does a general curve which specifies aging in function of junction temperature exists? Is its existence feasible, or there are too many variables involved not allowing to define it? And why is not common to include such information in the datasheet?

\$\endgroup\$
5
  • \$\begingroup\$ can you install a small fan, and cool the leads and the case? \$\endgroup\$ Nov 14 '19 at 16:20
  • 2
    \$\begingroup\$ You could do worse than calculate from MIL-HDBK-217. I did that for a power rectifier, temp = 80C, Stress (reverse) of < 0.3 (the applied reverse voltage is less than 0.3 rated reverse voltage), metallurgically bonded construction, plastic packaging and a benign environment and it gave me a FIT rate of 32.4 (FIT = failures per billion device hours). There are a number of copies of the standard available - try barringer1.com/mil_files/MIL-HDBK-217RevF.pdf \$\endgroup\$ Nov 14 '19 at 16:24
  • \$\begingroup\$ yes this is the backup idea, but I wanted first to see if I should really be concerned about those 70°C. @analogsystemsrf \$\endgroup\$ Nov 14 '19 at 16:25
  • 1
    \$\begingroup\$ When in doubt I pick up my trusty copy of MIL-HDBK-217F. \$\endgroup\$
    – Andy aka
    Nov 14 '19 at 17:17
  • \$\begingroup\$ wow that handbook is definitely a goldmine \$\endgroup\$ Nov 15 '19 at 9:50
1
\$\begingroup\$

does a general curve which specifies aging in function of junction temperature exist?

Not directly, no. It really depends on how you define / measure "aging", and on what specific PN junction you're looking at:

Is its existence feasible, or there are too many variables involved not allowing to define it?

The exponential function usually cited essentially describes the speed at which (unwanted) chemical reactions happens. That happens to have something to do with how likely it is that certain impulse or position space pertubations occur – and that's an exponential function of temperature, in general. Sadly, reality isn't that easy, because, as you'll notice, a junction at 540 K isn't just "much faster aging" than a junction at 270 K, it's probably instantly broken, since probabilities that you could neglect (like simultaneous disintegration of the crystal lattice in very many places) suddenly become likely.

Also, external factors influence the likelihood of you undergoing damage than just the temperature of the junction: For example, temperature of a bridge rectifier under load will undergo variations, and if you look at these, you'll notice that for a given load you don't get the exact same temperature every time – you get something that has variance. (That's because a) you can't measure anything without variance, that's how physics is, and b) well, environment temperature, wind, position, thermal resistance of wire, dust, …)

And with that variance, you get a probability that "something with a probability that very overproportionally increases with temperature" and immediate failure modes become vastly more likely. And so, component aging is a multi-causal thing that can only in very complex models be related to temperature.

And why is not common to include such information in the datasheet?

Because you typically find something like a mean time between failure in there, if you pay for that qualification. Semiconductor producers might be a bit reluctant to publish data that allows the competition to make conclusions on their doting processes – and a diffusion speed vs temperature curve would most definitely.

\$\endgroup\$
1

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.