TEC and current driver chips for laser?

True, but more heat does increase the chances of failure. So why should I accept higher temperatures when I can come up with a solution that runs much cooler?

Watch it on take-off (or at rotation, as pilots say) when the wings get loaded. They bow upwards a whole lot. But that's meant to be. However, the same goes here: The more flex you design in the faster fatigue can set in. For airplanes the time frames are usually in excess of 30 years. Come to think of it, same for electronics. Some aircraft are more prone to structural failures than others. The worst one out here (California) was a CDF plane where both wing struts gave up after decades of service. Unfortunately the crew did not survive that. The other (larger) accident of a jetliner was more likely due to maintenance issues. With that one the threads for the stabilizer screw gear head had stripped out.

AD is an excellent company IMHO. I trust them and they have never disappointed that trust. Can't say that for some other companies. But there are some parts with every manufacturer where I feel a bit uneasy and decide to go another way.

It

package.

Chips have a 10 year lifetime. While people think this is related to nonvolatile memory, the same 10 year rule is used in fine geometry logic chips. There is a threshold shift as the gate is toggled. It is very small, but take 10 years, 365 days per year ...blah blah and then toggle it at a gigahz, and you can get an amazing number of toggles. The "weak" file used in simulations should take this threshold shift into account. The threshold shift is a worst case that could happen over 10 years at maximum toggle. Of course, most chips will last much longer than 10 years.

You might want to talk to apps at AD about the chip. They may put it in a different package if there are enough inquiries. Each semi is different, but many companies have a few product engineers that take existing chips and change the layout for an alternate package. It is a low risk venture since the initial design is debugged. Not that this will be done in time for your project, but there is always the next project.

Most of my designs must live a lot longer than 10 years. More like aircraft where 30 years is almost an expected minimum. Even there I only remember one that had to be retired, AFAIR because the manufacturer or the aeronautics agency in Belgium would not allow a cert extension.

I have done that in the past but nearly always for full-custom chips where I wanted another package than the standard. In this case the volume would be too low. Often it is easier and less time consuming to design around it and come up with a solution that works with standard parts. Which do have lifetimes of several decades.