About a year ago, I have converted the old Halogen turned CF fixture to LED for better colour temperature. A few days ago, it fails to start. What makes it hard to debug is that it worked a bit after I tried a whole bunch of things. I tried googling for failures, but no one have documented anything. They probably just replaced the driver and move on.
I initially tried measuring the electrolytic caps on the driver board as the cheap ones they use tends to die. There are 3 capacitors - one for the bulk cap (10uF 400V) after the rectifier, one for some startup/bootstrap circuit (4.7uF 50V) and one for smoothing the LED (4.7uF 400V). I desoldered them and tested them on the multimeter capacitor scale. Their values all came out to less than 1% to their labels. So I went looking for other causes - resoldered the wires, checked the connectors etc.
Each of these trials worked a bit until the slightly cooler temperature next evening or early morning.
I had my bench supply wired in series so that I can run the driver from DC. The driver starts up around 80V, but turns off above 90V. The supply is DC, so the test ruled out the bulk cap. The smoothing cap is also out which leaves the start up cap. So I swapped in a 10uF 63V cap as it is the only new good cap I have. It takes a bit longer to startup and seems to fix the issue.
What I think happen is that the cap starts to dry out increasing its ESR slightly. On a cold day, the ESR was high enough to cause it to fail. The startup cap is much smaller in volumes, so it contains less electrolytic and thus more prone to drying out first.
By the way, the old CF electronic ballast is a good source for replacement bulk caps if/when they fail. The ones I have are only rated to 200V as the AC is only 120V here which gives around 170V DC after rectified.
I ordered an extra 24W LED driver module for $2.75 for a spare just in case.
Found the bad cap that I have replace from my LED supply. It is supposed to be a 4.7uF/50V cap. Ironically, the marking also say "Low ESR".
While the value is slightly lower (-8.5%), it is well within the +/-10% or even 20% of these types of capacitors. It is hard to tell if it is bad just from the capacitance value alone. It ESR value tells a different story.
Here is a generic used cap with the same value, rating and similar package. This gives you a bit of an idea what the ESR should be.
The electrolytic capacitor in question is likely dried out due to heat. The oxide thickness that determines it values doesn't change in the process, so it is within tolerance. The lack of electrolyte increases its internal resistance and is reflected in its current ESR value.
This is the reason why an ESR meter is useful for fixing old circuits.
09 Nov. 2018
It failed again. This time LED on the inner failed. A couple of LED blew up and an additional 4 LED are shorted. The bad ones turned black. The LED next to the screw cracked. I replaced all 6 and reflow the ring in my oven to make sure that the heatsink connection under the package get soldered. I bought 1 reel of 100 for $2 as spares.
I recently found out that they no longer sell this style of 120V LED rings.
More LED blew up. They are all in the same inner ring. There is a higher density of LED there and more heat is trapped there.
They didn't connect the heatsink tab (center) that the LED die is bonded to! The two connections on the sides are wire bonded so they don't conduct much heat. The center heatsink tab is electrically connected to the Anode of the diode. I removed all the LED from the ring and solder in a small piece of copper foil connecting up the Anode and the heatsink tab hoping that it would help to conduct some of the heat via the copper track. I hope this would extend the life a bit.
The other way of fixing this is to shift the LED such that the heatsink tab is on top of the Anode pad.
They could have done a much better job and it would have costed them nothing. :(
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.