Encouraged by one of my visitors, CharlySays , I decided to investigate the issue further and see if anything can be done to increase the useful life of these lights. Besides, the lights have been dark for almost half a year now and I was ready to remove them from under the cabinet and possibly replace with mini CFL lights of a similar form factor.
Keep reading for the results of the hacking exercise:
Basically, there are several design characteristics that, combined together, lead to horribly bad reliability of the light fixture despite the fact that each individual LED is capable of burning in excess of 30,000 hours. I’m hesitant to call them design flaws because a differently designed light would most certainly cost several times what L of A 7200 costs.
One of the basic problems is that all 24 LEDs are connected in series and therefore each burnt LED breaks the circuit and the entire fixture goes dark.
Another problem I see is that the LEDs are very nicely insulated by the plastic body of the fixture so they just cook themselves while on.
Lastly, the problem that we can actually do something about is that the LEDs are allowed to run in a current that exceeds that of the standard white LED spec. I don’t really have a spec for the particular LED used by Lights of America but most one-crystal white LEDs (you might call them “older type” because currently popular SMD-type LEDs usually have 3 LEDs connected in parallel, so they need three times the current) are spec’ed at 20mA whereas Lights of America 7200LED-BN runs at above 26mA (see pic below)
The voltage between the ends of the LED cluster coming from the rectifier diodes and the RC circuit on the other PCB is approximately 65V (see pic below)
The voltage drop PCB has an empty position for a current limiting resistor marked as R3. I am assuming the designers wanted to be able to limit the current through the LED cluster but then did not like the brightness at the lower level. My wild guess would be that they used good LEDs in their prototypes yet for production “standard” LEDs were purchased and that killed reliability.
The modest hack I’m doing here is to install the resistor and bring the current through the LEDs to within the spec limits yet to the point that the LEDs are not too dim to be useful as a light fixture.
We need to use a sharp knife to cut the PCB trace on the copper side of the PCB that originally shortens the R3
I’ve tried a couple of R3 values and it appears that using a 1KOhm resistor brings the current down to 22mA – better but not quite the spec and 2K2 (2.2KOhm) resistor (pictured below) gives 16mA – 36% less than in the original light and within the spec.
Update: CharlySays was kind enough to provide couple more current readings for R3 values: 13mA@3K3 and 10mA@4K7
Here is the current reading I was getting with 2K2 R3 installed:
I have also repaired two of the lights in a regular way (just replacing the burnt LED) and installed the hacked one and the regular one side by side, pictured below. It is notoriously difficult to photograph a light fixture properly when it’s on, so you’ll have to trust my judgment that the light output of the 16mA version is only very slightly less than that of 26mA one (almost imperceptible) and it runs cooler enough to feel the difference by hand.
Conclusion? It’s probably too early for one. One of the repaired original LEDs is already dead again. Another original is working along the hacked one just fine so far but I totally expect it to fail soon. It’s been something like 16 hours and the original one is hotter than the hacked one, so as soon as it fails, I’ll post an update here.