Post Go back to editing

LTC3769 boost converter becoming damaged

Category: Hardware
Product Number: LTC3769

We are using LTC3769 in a 5-20 boost to 24V configuration (very similar to Fig 1 in DS). In our first pilot run using these, we have seen 3 failures of the IC that produce 38V instead of the 24V. This does not seem to be related to the components surrounding the chip (feedback resistors, etc) as all seem to fine. Replacing the LTC3769 is the only remedial measure that is effective. I have also taken the trouble to measure the reference voltage on a good chip versus a bad chip and have found that the reference voltage on both is within spec. Also I have probed around pins with a DSO to look for any nasty spikes either at startup or in operation but I don't see anything approaching maximum specified values.

The 38V is of interest as we have downstream regulators with max inputs of 35V. So far none of these have been damaged but I would like to know why 38V and why not something higher as the chip is capable of producing 60V. Has anyone come across this issue? Any theories as to why 38V on these failed units?

Thanks 

  • Hi JKK,

    Have you verified your design using LTpowerCAD then LTspice? I would suggest to double check the layout checklist found on pages 24-26 and the layout debugging guidelines on page 26. You could also check your board for manufacturability issues such as wrong component values or poor solder.

  • Hello and thank you for your reply. I have checked the pcb layout against the recommendations in DS. I find a few potential areas of difference, one is that the output bulk electrolytic is placed quite a way from the output  (about 3 inches), output ceramics are close though. This impacts recommendation 2. Recommendation 7 seems tricky to implement. Our design has the Vfb, Freq and Ith resistors tied together  in an AGND puddle on an internal plane and single point connected to EP via pin 4.  
    As far as simulation is concerned, LTspice doesn't show any issues like overshoot or overvoltage. LTPowerCad did indicate a red PM parameter for 5V input operation. 
    I don't believe our issue has anything to do with wrong components as we replaced the faulty chip with a good one and it worked successfully without changing anything else. I guess it's possible that a pin or the EP was not soldered adequately but it did seem to me that the chip itself has gone faulty. 
    I could make the PCB changes in line with the above but would be somewhat reluctant to make a change without having good evidence that it would fix the problem. 

  • If you are reluctant about the PCB changes. You could try getting hold of the demo board DC2173A and make the necessary changes from there. It is good to have a working reference board when debugging.