Did you try dipping the copper to just under the LED? The water should provide much better cooling then the fan and as such I would think that at least some of those gains are from keeping the LED base cooler.
Yes, I did.
The output stopped climbing at approx. the same current at ~3A
You are entering new testing territory with underwater tests! :partying_face: I love it!!!
Like this Tex?
Don’t worry, I thought about it too. It’s not the base cooling, it’s definitely the phosphor cooling problems
Sad it is so limited, kinda strange with such a low thermal resistance in theory. I think something might be up with those resistance numbers.
Yes, same here. If only Nichia use AlN as the phosphor support, things would be different. The soft polymer used doesn’t conduct and spread heat properly thus the high delta temp between phosphors and the die. And probably, is this because the Ra9080 uses special thicker phosphor layer mix?
You’re born with the magic. Muggles can’t simply acquire it by being given a wand.
Really interesting. I never thought about the phosphor temperature, rather than the junction, as the limiting factor.
Do you have a place these LEDs can be purchased, or can you source them? I’d like to buy a handful to play with in a few different tints. I do plan to purchase a bunch of 219C, 319A, and 144A from you once the customs stuff is sorted if that makes it easier.
Yes I will “probably” have them all.
Just a question I thought of: with phosphor cooling you see output increase up to around 6A. Without phosphor cooling output tops out at around 2.5A. Is this difference directly due to the phosphor temperature/efficiency itself or actually still a result of junction temperature since the junction probably is significantly heated by the phosphor.
NOTE: I use not so accurate apps and medium tech smartphone (Sony XperiaV). The accuracy tops out in +/- 8lux.
In watercooled test I used the same LED and didn’t really took note about the result because it’s impractical and I didn’t want to use my new LED for such test. At least I might do the same test later after I got a feel of what the test should be performed.
- Normal DTP cooled: output max @ 3-3,4A (flat curve)
- Watercooled DTP (dipped up to LED base depth): output max @ ~3-3,5A. I just use my eyes to check if there was a gain in brightness (ceiling bounce method), stopped when there’s no more gain. Even at 2A the water near the LED was boiling, upsetting my low tech camera readings (the meter jumped).
- Watercooled (completely submerged) DTP: output max @ ~6A (some tiny parts of the edges already burnt)
……cleaned the whole remaining phosphor….
- Watercooled (submerged) DTP without phosphor: output max @ ~7A. The die started to show heavy burnt marks after 1 minute running. That showed that it’s unusable anymore. Actually the die still lit up until 9A but severely damaged.
The next step should be done with 3 different healthy LED to start with. Because each tests seems to severely strained the LED. Or to make it more economical, do the test backward: submerged - dipped - dry.
The submerged & dipped test was indeed improved base cooling and output but not that much.
- Clemence
As a side note:
- from my earlier test you can see that the thicker phosphor in 3000K showed more damaged than the 5000K (both were 9080). To me it’s the sign that thicker phosphor absorbs the photon energy more - thus more heat. And worse, the phosphor also act as a blanket to the juction.
- junction temperature should be higher with the phosphor on it. We’ll find out later with three separated tests using the LED from the same reel. I got these LED in almost as expensive price as the 219C, I have to plan ahead
Double side cooling is nothing new in LED industry. A manufacturer already use this idea to cool their LED bulb using silicone liquid filled diffuser.
Link: http://www.billfrovichelectric.com/ugetlight-produces-new-liquid-cooled-led-bulb/
Thanks for the response. I found a couple papers that measured the phosphor efficiency as a function of temperature. Depending on the phosphor type they measure a 2-10% efficiency drop at 100C.
https://www.osapublishing.org/captcha/?guid=850BEBAF-DEF4-4D51-0D76-467C817716F5
Those validate my suspicion :sunglasses:
I guess high CRI mix further reduce the phosphor efficiency
Hmmm… Topping the reflector off with distilled water before putting on the lens and bezel…
You could get brighter output by using optical grade silicon oil. The refractive index is closer to the LED primary lens. A chinese scientist did an experiment to 30+% more light extraction from any LED sources using multi layered silicon gradually from high index to lower (inside out). And he did that before the flip chip era.
- Clemence
Where can we get high quality (no cheap knock-off) optical-grade silicon oil? Wait, is it silicon or silicone?
Silicone of course hahaha.
Try Dow Corning if you’re in US. It has many different indexes available to public. For Asian there many japanese manufacturers such as Shin-etsu, Asahi rubber, etc…
Water is quite unbeatable though when it comes to heat absorption. (isn’t it?)
It also conducts heat better than oil(s) typically.
(I’m not worried about the refractive index.)