DIY COB light was working great.... then failed. Why?

Hi All,
I recently completed a harvest and results were great. My DIY cob light has been functioning flawlessly the entire time. Yesterday, the light was functioning normally. Today, 1 light bar is out completely, the other two are flickering and drawing a measly 70 watts combined.

I'm confused as to what the problem is. I thought I was under driving these for efficiency; temps seemed fine- I could touch heatsink comfortably but it was warm. Wiring looked fine. I am using simple WAGO connectors for wiring. Why would all three heatsinks fail at the same time, in two different ways?

Details:
I have 18 CXB3590 cobs mounted to 3 heatsinks that are ~1.25"x6"x36". Each grouping of 6 cobs on a heatsink is supplied power from a Meanwell HLG-240H-C1050B (238V, 1.05A). Each heatsink has 2 6" CPU fans that are always on, and well as a circulating fan, all in a 4'x4'x6' tent.

Thanks for the read!

Unless you had moisture ingress, my money would be on undersized heatsinks.
Heatsinks are normally rated in C/W or sometimes K/W (Celcius and Kalvin)
I.e. for every Watt, there is a set increase of heat (there is a little more to it, as some use thermal grease/glue or even the little insulators behind power transistors).

Just some assumptions (your data was a little light), you have 6 cobs on each sink for a total of 18 COBs, you did not say how hard you are driving them so I will just use worst case numbers.
238V/6 = 39.67V per COB (shouldn't it be 36 or 72V ballpark on the 3590s?)

If you are running at 1.05A, I would assume you are using the 72V COBS. Looking at the V/I graph, the COBs would use between 68 and 72V (temperature dependant). Something is way off, could you just post a synopsis of what voltages and currents you were aiming for per COB?

ANC said:

Unless you had moisture ingress, my money would be on undersized heatsinks.
Heatsinks are normally rated in C/W or sometimes K/W (Celcius and Kalvin)
I.e. for every Watt, there is a set increase of heat (there is a little more to it, as some use thermal grease/glue or even the little insulators behind power transistors).

Just some assumptions (your data was a little light), you have 6 cobs on each sink for a total of 18 COBs, you did not say how hard you are driving them so I will just use worst case numbers.
238V/6 = 39.67V per COB (shouldn't it be 36 or 72V ballpark on the 3590s?)

If you are running at 1.05A, I would assume you are using the 72V COBS. Looking at the V/I graph, the COBs would use between 68 and 72V (temperature dependant). Something is way off, could you just post a synopsis of what voltages and currents you were aiming for per COB?

View attachment 3934064

Click to expand...

And I would assume he is running 36v COBs. No way to run 6 72v chips on a hlg-240h driver.

Could it be the driver that's causing this? If everything is is connected nice and firmly with a fan on the heatsink, there's no way that the cob could die this early!

I had an experience just over a week ago which had my room at 50c for about an hour, so you can imagine the heatsink s were like 70c+ ! My LED are all fine thank God (all cxb3590@ 50w 1.4a), so I know they can handle a bit of abuse.

Wow, that is also severely under driving them.

At 1 Amp they don't even chart it on the graph, guess it must sit at 30- or 40% driven.
To get 100% you need 1850 to 2500mA depending on the heatsink you run. Part of why I prefer the 72V units as the 1050mA drivers that are basically perfect for running them at 100% would need 1050 to 1350mA depending on heatsinking performance. Heck according to the graph you could get 85% at even105C running 1050mA..

These are 36V COBS, significantly underdriven, as you guys guessed. I used thermal grease as you would a CPU and I took the oxidation off the raw aluminum heatsink before applying grease or COB. For part of the time during grow, I had a resistor on the control leads of the power supplies to drop the current down to ~70% of the 1.05A capable. Based on the case that I am significantly underdriving these and that I have large heatsinks that are actively cooled, I don't think it was a thermal issue.

Because all three strings had a failure at the same or similar time (within a day), I am guessing that it is either environment driven or wiring related; but I am at a loss, cannot come up with any rational reason for 1 string dead and the other two at way low output.

If no one has any no brainer 'duh' moves to suggest where the problem is, I think I will take the light out of tent today and troubleshoot further.

Thanks guys!

that is strange. power surge/lightning strike?

id start by disconnecting the drivers and checking each one for voltage. if they have adequate voltage try to light up cobs in batches of 3 at a time, which the driver should do

also check your AC input voltage to eliminate variables. maybe something funky is going on with your house wiring and the drivers arent getting the minimum 95 V input