Blog Post Actions
Group Actions
Engagement
  • Author Author: genebren
  • Date Created: Date Created
  • Views 2256 views
  • Likes 0 likes
  • Comments 7 comments
Related
Recommended

Kitchen Lighting System

genebren

I have over the past few years blogged about my kitchen lighting system, starting more than five years ago with Kitchen lighting system - take 2 .  Then in 2019 I had started a project to update the lighting system, which culminated in winning a 1st place finish in the project14 Home automation contest (Kitchen Lighting System Phase 2 - Wrap up ), but I never fully finished the project or installed it (future 'Off the Shelf' project?). Fast forward to 2021 and recently I noticed that the initial system had failed.  My initial thought was that the 5 Volt wall wart power supply had failed.  I dug through my project boxes and found a backup for the power supply (I always tend to buy more than I initially need, leaving me with backups or extras for other future projects) and swapped out the suspected back supply.  No joy, the system stayed dark.

 

This past weekend, with the switch from Daylight saving time, my wife reminded me that the kitchen would be getting dark again and that I should get the under-cabinet lights up and running if I wanted to continue to enjoy tasty homemade meals.  It doesn't take much more than the threat of missing a meal to prompt me into action, so I dove in and started figuring out went wrong. Here is a look at the (working) lighting system:

imageimage

I climbed up to access the Master Controller, located on the top of the cabinets, over the microwave oven.  Checking the power connections (PIR and I2C connectors), there was clearly a lack of power.  Having previous tested the power supply, I figured that I had a short somewhere in the system, so I removed power and disconnected the PIR sensor and the two I2C cables.  After reapplying power, I verified power at all of the outputs, everything looked good.  So I reconnected the PIR sensor and checked again, all good.  I reconnected the right hand side of the system, and checked again, all good, and the systems was properly handling the PIR outputs and turning on the LEDs when motion was detected.  Next I checked the impedance between the power and ground on the cable to the left hand side of the system and saw a very low impedance.  Unsure of whether it was a issue with the cable or the light pucks, I disconnected the cable from the first punk on the left hand side and rechecked the cable.  The impedance was high, so the cable was good.  I then disconnected the cable between the first and second light pucks and reconnected both ends of the cable to the first light puck.  Reapplying the power clearly showed that the problem was at the first light puck.

 

I removed ans disassembled the light puck and looked at the board under the microscope.  At first nothing looked bad, but on later inspection the filter capacitor (10uF) on the power supply lines look a little dark and the solder on the +5 Volt side of the capacitor looked black and powdery.  I removed the capacitor and recheck the power line impedance, which looked good.  I then replaced the capacitor and reassembled and reconnected the light puck into the system.  I then powered everything up and all the lights came on as I triggered the PIR sensor.

 

This is the first shorted ceramic capacitor that I have experienced.  Not sure exactly what took it out, but I did install a higher voltage capacitor in case the part was damaged on a surge.  I guess another possibility was a bad solder joint that continued to heat up and eventually damaged the capacitor.  I will have to keep an eye on this and see if anything else happens.  In the meantime, maybe I should get back to work on the improved kitchen lighting system and upgrade the whole system.

 

But the immediate good news is that there once again is enough light for my wife to safely continue producing tasty homemade meals image.

 

Thanks for reading along!

  • I find it interesting that not many vendors talk about ESD damage to MLCCs.  Most don't even have ESD ratings.  Maybe because in general it is a low occurrence failure.  We used to do extensive ESD testing of our high speed pin electronics because they were exposed to ESD at the test sockets.  I got surprised when I was doing characterization to set the ESD spec for pins in a new design.  We were getting some pins that started failing at relatively low voltage.  I thought we were having clamp failures in our ASICs which were normally the weakest link, but it turned out that it was the broadband MLCC caps that we were using to AC couple the pins.  Higher value MLCCs have very thin dielectric layers and we were getting variations in the ceramic particulate size in the slurry used to manufacture the parts.  We were getting early punch through when the particulate was large enough to create gaps in the dielectric.  Long story short - we started sample ESD testing and returned any lot that didn't meet our requirements.  So there are other failure modes than mechanical damage and low voltage overstress.

  • in reply to fmilburn

    Crazy, I was just reading through this.

     

    I rescued the capacitor that I removed from the board (I had wiped it off the soldering iron tip onto the steel wool-like tip cleaner).  There is no visible signs of damage, but the contact on one of end of the capacitor was significantly darker (heat discolored) than the other.  This discoloration is likely due to heat buildup on the capacitor after it failed, as the circuit was power-up, likely in current limit, but still likely dissipating ~1/2W. The capacitor is shorted (very low resistance). I don't know if I could polish the part enough to see any internal cracks, but it might be interesting to try.  Also, since the capacitor is off the board, I have no good references as to the parts orientation while it was attached to the PCB. And again, also, I am not sure if I could rule out any stresses applied during the desoldering and removal process.

     

    With cracking being a very likely source, I am reminded of an issue that has bothered me for some time.

    imageimage

    My libraries for Eagle have pretty narrow chip outline (seen in Eagle and a similar part on the board).  The result is that the mask and even more so the silkscreen are higher than the component pads.  This results in some of the components not sitting completely flat on the PCB during the hand soldering process.  I usually end up re-heat the joints and pressing the component flat to the PCB.  I imagine that I might be adding some extra stress on the components by doing this.  I think that I might want to update some of these silkscreen outlines, making them wider than the pads, so the components will not bump on these and hopefully set flatter during the soldering process (less reworking during the solder process).

     

    On the subject of stress, these documents talk about vibration and board flexing as a contributor to stress related fractures. These PCBs are pretty securely mounted into their housings (4 screws for a 1 sq. inch board might be overkill), so I am not sure there is much vibration or flexing (pretty difficult to flex a 1" x 1" board).

  • in reply to Jan Cumps

    I think I’ve seen a manufacturer’s post on this at some point here on e14. It is an interesting topic for me because at one time I used to do stress analysis and testing (but not on electronic parts). This Kemet paper is interesting:

     

    https://sh.kemet.com/Lists/TechnicalArticles/Attachments/89/2006%20CARTS%20Flex%20Improvement.pdf

     

    They give ppm failure for different chip sizes and it shows smaller chips less likely to fail.

  • The failure mode for (mechanical) defects on a ceramic SMD is often: Shortcut.

    Both repair channels of Louis Rossmann and Paul Daniels spend a decent part of their time looking for broken SMD ceramic caps in MacBooks - except for laptops with liquid damage: there they look for corrosion.

     

    Possible reasons:

    • already slight cracks before placed
    • damage during solder process
    • pcb bending
    • heat cycling causing stress.

     

    (in some circuits, this is mitigated by putting 2 higher value caps in series, so that one cap can't cause havoc).