A good 20-months ago, I settled on a Philips 13w E27 LED globe for my room. At the time, it seemed like a solid buy, given that I had no RF problems and it was as bright as expected with a solid build that precluded a teardown.
Fast forward a year and a bit later, and unfortunately this one globe began to be an issue. When turning on, the globe would buzz and flicker for a bit and then operate normally. Over time, it got worse, and had me wondering if it was my socket contact. After some alterations, it didn’t improve, so I was then convinced it was the crimping of the shell connection on the ES connector. Squeezing on that seemed to make the problem better for a while.
Eventually it got so bad that it outright failed. At best, you get a short flash, and then nothing. It seems it may have succumbed to a bad solder joint inside – so I decided a teardown was in order for a repair mission that was ultimately unsuccessful.
Taking off the dome was a bit of a hard task. Prying with screwdrivers and gouging at the dome was necessary to get enough leverage to pry it off.
Once taken off, we can see that a rectangular MCPCB is used with a cluster of white LEDs arranged in a rough oval shape. A piece of white plastic border is secured around to work as a diffuser/reflector of sorts. Removing the screws allows the diffuser to be lifted off.
The MCPCB is fixed into place with a yellow thermal adhesive. A total of 30 LEDs can be accommodated on the board in a single series string configuration, although through the use of a jumper, three of them were omitted. This is likely to accommodate different wattages and different bins of LEDs – less efficient LED chips will need more chips to achieve the same brightness. The printing on the MCPCB implies it was made in Week 23 of 2013, and the board design was for Nichia 757 LEDs which are a high quality Japanese LED.
Sadly, there isn’t anywhere to go on this side. Prying on the MCPCB managed to get it to release to reveal the rest of the metal “cap” at this end. Trying to pry open the metal cap at the seams was fraught with difficulty and ultimately failure as the soft aluminium gave way at every attempt to lever it apart.
Eventually, out of frustration, I decided to attack the globe at the other end, hoping a collar of plastic might slide off in some way … but it didn’t.
It took some work with the side-cutters and wires were cut, but the cap came off and revealed just the top section of the current driver PCB. Nothing unusual was visible, although the shape of the PCB confirmed it wasn’t going to come out of this end anyway, so it should be taken apart from the other end. Hoping there would be a way, I cut the plastic and tore it away but all it revealed was just more metal. It’s as if it’s a plastic coated onto the heatsink, which makes it a lot less efficient than if it was just painted metal.
In anger, I got a tapered reamer and started reaming the screw holes. What I found was telling – a lot of dark coloured silicone glue which must have been applied inside as a form of “potting” compound. This explains the difficulty in taking it apart – it was literally glued together so well there wasn’t much of a chance to service it anyway.
So unfortunately, this was a failed teardown – we didn’t get very far, and ultimately, weren’t able to fix what may have been a very simple cause of failure. Oh well. That being said, I suspect this is a one-off incident, as the other globe and the 14w globes seem to be doing just fine for now.