1 failed cob in a 4 cob array; how should it be replaced? (older gen 3070 vs new 3070?)

Details:
4x Cree CXB 3070 on a HLG-185H-1400B (series)

Functioning normally for years, until one day all 4 cobs in the circuit appear very dim. Adjusting the 100k pot makes no difference (as if its running in 'limp' mode and at minimum amperage?).

I tested the driver w/o load: 143vDC output. 0-~82,000 ohms from the pot (it is within 20% tolerance).
I connected the driver to another series of lights (CXB3590-that I normally run with a different driver) so I know the array is good/functioning. It powers these lights fine, full adjustment via pot. 138.6v (max), 121.6v (min); 1.42amps (max), .04amps (min).

Upon further inspection, one of the 3070 cobs is not lighting fully; as if 1 of the diodes is not firing? If a COB LED had one of its internal parallel-series diodes fail, what would happen?

I plan to do more testing.

Ultimately, how do I go about replacing this older gen >36v chip in a series circuit? What effect will a newer/lower voltage chip have on this circuit-the new chip runs cooler than the rest or a bit brighter?

Just wondering if they were mounted using pads or paste? All my testing showed the pads are worthless, but I am using the Binder & Wirth holders ( not metal/molded plastic) and they don't give enough pressure to make the pads work.

nevergoodenuf said:

Just wondering if they were mounted using pads or paste? All my testing showed the pads are worthless, but I am using the Binder & Wirth holders ( not metal/molded plastic) and they don't give enough pressure to make the pads work.

Click to expand...

This may also a consideration; I like using CPU thermal paste, because I have much experience with it from building OC'd PCs. But, this light in question is using pads. I forget which name brand holders, the plastic ones that PLC used back in the day.

I've seen dead and poorly lit diodes in some of my personal 'cheap-CN cobs' when I was initially testing COB technology. But they likely came from the factory this way.. However, I also accidentaly ran one of these cheap-CN cobs for way too long w/o enough cooling (dead fan on a cpu heatsink that was way too small to handle non-active cooling duty). The COB still works (from a lumen/visual stand point), even if its missing a few diodes.

BJB 47.319 holders
Arctic thermal pads, .5mm, 6.0W/mK (cut to fit from a large sheet)

Visual inspection indicated one cob definitely not firing fully. Only 1/3 to 1/2 of the chip was illuminating..and would flicker a bit as power was adjusted. After the circuit warmed up a bit it seemed to get a bit brighter even..but still not firing all diodes.

While removing the holder/chip from the heatsink, screws were tight and there was pressure on the pad. After peeling the pad away the chip itself looks fine; no dark/burnt marks on LES or ceramic side.

Upon omitting the [failed] chip from the series, everything else runs normally with full adjustment in power/intensity.


So, is there any sort of warranty for this type of failure? How/where to start? I don't find much about 'warranty' when I search the pdf manuals I've found.

And, finally-how do I go about replacing this single chip in a 4-chip series circuit? Mainly I'm concerned about voltage differences, and weather it matters on a reliability, longevity, or safety scale. Will a newer/more efficient chip simply 'require' less voltage to maintain the defined amperage (compared to the older chips)?
So, instead of say: 35.6v (@ 1,420mA) per chip; 35.6*4 = 142.4v. we now have three of these 35.6v chips (35.6*3) + one of these new 34.8v chips = 141.6v (@ the same amperage; 1,420mA).
(Note: 34.8v is a number I made up, I'm not even sure what the latest 3070s run at off the top of my head/without further research.)

Just looking for confirmation on this before getting something new. To me, it seems like it'd work without issues.

chip-to-chip voltage doesnt matter in series, as long as the total voltage of chips fits your driver

You would take up the warranty with the distributor/retailer.

Adding a new 3070 should work without issues, same amps and cumulative voltage are the parameters that need to be met.

So, the other COB's are still using pads? If so, I would take Tc to make sure the others aren't getting ready to fail.

Thank you all. That was also my understanding of it.. And after thinking more on it, I do remember others running COBs and monos in series to 'fill' a driver.

nevergoodenuf said:

So, the other COB's are still using pads? If so, I would take Tc to make sure the others aren't getting ready to fail.

Click to expand...

I was also thinking and wanting to do this to be sure. I have an laser temp gun and external temp probes (from a hygrometer). I'm sure there is a more accurate way; soldering a specific temp probe to the chip or something? But, not very viable at this time.