1

We've had a number of optical transceivers fail recently and I am trying to track down the cause. It has happened with two different models: HFBR-53A5VSEMZ (now obsolete), and the TSP-D1CH2-C11.

Some have failed after 18 months in the field, and some after 1 month. We have also seen some fail after our burn-in procedure. All of the failures I've examined behave the same way: the VCSEL integrated into the transceiver is no longer emitting anything. The receiver end always looks fine.

I suspected some kind of electrical overstress (EOS) or electrostatic discharge (ESD) damage, and according to a couple of sources I've found (3 and 4), this type of damage should cause an increase in reverse bias current leakage. So, I measured reverse leakage current in both a working and failed VCSEL. Unexpectedly, the leakage in the failed unit was less than the leakage in the working unit:

Reverse leakage current comparison

I also measured the forward I-V characteristics:

Foward I-V

Does anyone have any insight about what kind of damage could cause data like this?

sbell
  • 1,571
  • 12
  • 20
  • 1
    Have you given a part to a failure analysis lab for analysis? After basic DC measurements are made (like what you've done), they would remove the lid of otherwise remove the encapsulant so that the die can be examined. – SteveSh Dec 07 '22 at 01:12
  • You're certain it's the VCSEL that's damaged and not the driver circuit? What is your burn-in procedure (How many hours at what temperature)? How are you controlling the supply voltage during burn-in? – The Photon Dec 07 '22 at 04:31

1 Answers1

0

As it turns out, I made the lazy mistake of measuring the VCSEL in-circuit. When I removed the VCSEL from the surrounding circuitry and ran the reverse-bias test again, I got a result that more closely matched the pattern for ESD damage:

out-of-circuit leakage measurement

sbell
  • 1,571
  • 12
  • 20