Operating temperature of integrated circuits

[Flyback]

Hi,
The UCC3817A PFC controller IC has an operating temperature range of 0-70degc.
However, its junction temperature is said to be rated to 150degc

Do you know what would happen if it was operated at say 100degc for 8 hours per day for 30 years?

UCC3817A
https://www.ti.com/lit/ds/symlink/u...l-digikeymode-df-pf-null-wwe&ts=1593441712447

 

[Grossel]

Depending of the material quality. In general, you should expecting a reduced lifetime. Also, if that component have a high operating temperature then one should assume that the surrounding components would also operate under high temperatures.

Also - the number of transitions from cold (room temperature) to heat and back again, combined with the temperature difference between hot and cold condition would contribute to reduced life span (or probability of so).

Unfortunately I don't sit on data that let me compute a final expected lifetime for one particular component.

 

[ChrisP58]

The UCC2817A extends the temperature range to –40°C to 85°C. Still not what you're looking for, but it won't be as far over stressed.

 

[Diver300]

Hi,
The UCC3817A PFC controller IC has an operating temperature range of 0-70degc.
However, its junction temperature is said to be rated to 150degc

Do you know what would happen if it was operated at say 100degc for 8 hours per day for 30 years?

Is the 100 °C the ambient temperature or the junction temperature? How much heat does it generate in you application?

 

[Flyback]

100degc is ambient, the chip only generates ~100mW . (as their is an external gate driver)

 

[Diver300]

100degc is ambient, the chip only generates ~100mW . (as their is an external gate driver)

The data a sheet indicates 50 - 100 °C/W so that 100 mW would put another 5 - 10 °C on the chip.

I doubt it would cause a problem, however I don't have any proof to back that up.

You should be avoiding really rapid temperature changes for the IC, but it is overloads of the IC that cause that.

 

[crutschow]

The 70°C (or 85°C) is the maximum rated operating temperature where the chip will meet its specs.
Above that some chips may meet the specs and some may not.
One of the reason military rated ICs are so expensive is that they are tested to operate at up to 125°C, it's not that the chips are constructed any differently.

Military rated chips are also generally packaged in metal or ceramic packages to withstand the higher temperature.
So there is also a concern that the typical plastic package of commercial rated devices could deteriorate at 100°C.

I would contact the manufacturer about operation above the rated temperature.

 

[simonbramble]

crutschow has got it right. I work for a semiconductor manufacturer and come across this problem quite often. Commercial grade parts are exactly the same silicon (and mostly the same packaging) as military grade parts. However military grade parts are tested (and thus guaranteed) to work over a wider temperature range.

Now for the lifetime... there used to be a rule that you should never use a part above 100 degC or its lifetime will be greatly reduced. This was the case back in the 90's. That is not really the case now (at least not with the devices my company make). Failure rates are so low that, even at high temperatures, the part's lifetime can run into thousands of years (statistically). The manufacturer will not stand by any guarantees and you are definitely on your own, but it should still work. You need to look at the MTBF data for the part and extrapolate using the Arrenhius equation. That will tell you, statistically, how long the part should last. However, this is only applicable to the silicon. it does not allow for stress on the solder joints as a result of temperature cycling. These will go open circuit long before the silicon gives up.

Higher temperatures also causes the package to deform placing stresses on the die which means some of the characteristics might wander (or just fail). I have seen power supply parts oscillating at high temperature because the stress on the die caused the internal reference to oscillate

I posted a question on this forum back in December last year about the effect of lifetime on components, saying it should run into thousands of years... and got hounded for my answer. I have looked into it since and spoken to many packaging guys, chip designers and QA people and indeed it is thousands of years as far as the silicon is concerned. Mechanical failures is another matter though