How can I prevent premature aging in IR emitter/receiver components?

[mdjensen22]

Hello All

My company has several products that use an IR emitter & receiver in line and are used as a sensor to detect when something 'breaks the beam'. The problem we are running into is that several of these components are aging prematurely, but they work perfectly at the time of our products' manufacturing.

Does anybody know of any good ways to screen for this premature component degradation at an incoming material level?

I've already changed all the manufacturing processes to prevent the parts from taking damage during assembly (removed thermal and mechanical stresses). Otherwise, I don't know any good ways to check for this - looking for parts that will fail down the road, but currently work.

 

[dlgoff]

Here's an old thread on https://www.physicsforums.com/showthread.php?t=179706".

 

[jsgruszynski]

Hello All

My company has several products that use an IR emitter & receiver in line and are used as a sensor to detect when something 'breaks the beam'. The problem we are running into is that several of these components are aging prematurely, but they work perfectly at the time of our products' manufacturing.

Does anybody know of any good ways to screen for this premature component degradation at an incoming material level?

I've already changed all the manufacturing processes to prevent the parts from taking damage during assembly (removed thermal and mechanical stresses). Otherwise, I don't know any good ways to check for this - looking for parts that will fail down the road, but currently work.

Define "aging" for me. Are you saying you are getting mechanical failures of the package itself or electrical failures of the semiconductor components?

BTW this is my bread-and-butter professionally. I've experienced situations with exactly this kind of weirdness before. Screen testing components is something my company sells solutions and consulting for.

PM me if you want to take the discussion private, off the list.

 

[mdjensen22]

Thanks for the information Don - not exactly what I'm looking for, but much appreciated anyway.

When I say aging, I mean electrical failures of the components themselves. Their electrical responses are prematurely degrading. For example, let's say that with nothing in the way of the 'beam' , I get an output voltage of 0.3V and with a single piece of media between the elements, I get a voltage of 4.2V.
What is happening is that after a few months of use, the readings with nothing in the 'beam' start to yield voltage readings of ~1.4V. The components are not being driven anywhere close to their maximum ratings (4mA on 30mA emitters & 5V on 35V parts) and environment is not a factor.

We are seeing this type of result on about 10% of our devices. While I have fixes for this issue already in place, I have been asked by the head of the company what we can do to screen for these components going forward. My biggest hang-up is that all components work as expected on receipt/initial build and I do not see any correlation with any of the standard electrical parameters (forward voltage drops, gain, etc.).

 

[dlgoff]

Do you do any kind of http://en.wikipedia.org/wiki/Burn-in" for failure/degradation?

 

[mdjensen22]

Not typically - we have in the past, but this was challenged by the powers at be.

This was mostly stopped because there were still issues in the field even after the burn-in. This was all before my time, so lucky me, I inherited this issue.

 

[dlgoff]

Maybe not so lucky.

Anyway, my experience says Burn-In. Hopefully others here may have more input.

 

[jsgruszynski]

Are you the original manufacturer of the semiconductor devices?

If you are, you probably need to do some failure analysis work on the failed units to look for a physical cause. Also if they are yours, you absolutely need to start doing burn-in again.

I'm a bit shocked you've (not personally, your company) stopped. Not wise because all semiconductor manufacturing processes have manufacturing created defects that need to be screened out with burn-in. I'd also insist on reliability testing if I were running the show. End-of-life field failure is a possibility also - though I'd look at s infant failure first unless the FA suggested a reliability problem.

If you are OEMing them, you need to talk to the supplier and get their quality/reliability folks involved.

 

[mdjensen22]

We are not the original manufacturer by any means - we purchase the components and then put them in our product.

I am told that burn-in is done on the supplier's side.
Previously we had tried working with the supplier without much luck...so we switched suppliers. So far everything has been good with this supplier (recent transition though), but the challenge was still set before me to see if there is anyway to electrically screen these components.

 

[jsgruszynski]

Ok. Well if you need to test for this, what you're going to want to do is some type of sampled accelerated life testing (basically burn-in on a few parts). This will accelerate any defects to failure using heat and/or electrical stress. if you get failure rates above the vendors expected in-use failure rate (the flat part of the bathtub curve) then you know you've still got some infant failures passing through the vendor's screens. This is all statistical so the exactly number of parts that have to be tested is a "statistically sufficient number". I'd recommend working with your vendor's quality/reliability group as much as you can.

http://en.wikipedia.org/wiki/Bathtub_curve

Do you have a budget for doing this or is your boss expecting you to present a proposal for a budget?

 

[mdjensen22]

I'm not worried about budget for this - to eliminate this (or reduce this) issue, we have deep pockets.

Thanks for the information guys - so far it seems that my initial thoughts were correct though. There isn't really any good way to electrically screen these parts and that some sort of life/burn-in would be needed.