Would people buy a heatsink that allows water cooling?
- Thread starter ketchup45685224
- Start date
-
- Tags
- heatsink
ketchup45685224
Well-Known Member
Maybe, but if it cost to much. you'll never get started
wietefras
Well-Known Member
That isn't really true though. The efficacy still drops when you increase the power.
Unless you employ an expensive and inefficient chiller to cool the water I doubt the the temperature of the leds would be lower than with air cooled systems.
I "use" a radiator cooled water cooling and the temperatures of the leds are similar. The reason for me to use water cooling is what Randomblame mentioned, to be able to optimally use what little heat the leds produce.
CobKits
Well-Known Member
"many times" is a stretch. *maybe* up to 10% improvement if you significantly drop the temperature 50C or more and maintain same current.
you can drop heatsink temp significantly, but you still have the case-to-heatsink thermal resistance and the junction-to-case thermal resistance to deal with. these are bottlenecks so as soon as you start increasing current your case temp is rising above what your heatsink temp is and now amount of delta T on the heatsink will reduce the amount of heat that needs to flow through those bottlenecks as its directly proportional to input current
Here is a graph of case temp vs flux at a fixed current for CLU048 .keep in mind in the grow world we are generally already operating in the left portion of the curve where it is flatter. this is 1080 mA or ~35W for a citi 1212. on a decent sized heatsink youd have a heatsink temp of ~50C and case temp 60-70C. If you aggressively liquid cool and drop heatsink temp to 0 C you might get a case temp of 10-20C which is an efficiency bump of a little less than 10%
really the only way to know would be to do some direct measurements of par or lumens of a single chip on a passive vs an active vs a watercooled setup, in the latter case starting with room temp water and dropping the temp down to 0 C or less (with glycol below 0 C.). then you could have a direct measurement of the % of efficacy increase theoretically achievable. In practice if you built a rig that was able to vary temp (with increased cooling load on an increased deltaT), youd want to do measurments of light intensity with the entire system including lights and cooling system on a single killawatt plug to determine overall energy input for overall light output
keep in mind that unless you live in a nice alpine climate (where cooling is less of a benefit anyway), you are paying money to produce cold water. Even a basic liquid-to-air heat exchanger uses a fan, you have pumps to pump the water around, etc.
A large temp drop such as 50C drop would require a chiller. so now you are investing money on hardware to get an efficiency gain. so say you have 1000W of lights to cool. by aggressively liquid cooling them you may be getting up to 8% more efficacy out of them, but any energy used by pumps, fans chillers etc will be counted against that. chilling water isnt cheap and you can suck up that 80W pretty easily and be right back where you started.
as a point of reference most residential hot water hydronic heat systems (boiler on the heat side and baseboard radiators, domestic water heater coils, or in-slab/in-floor pex on the load side) are designed to operate with a 20F(10C) temperature difference. so water enters your boiler at 140F/60C and exits at 170F/70C. the 170F/70C water is carried to the load where it gives up heat and again is 140/60F on its way back to the boiler.
depending on the size of your cooling system, water slightly warmer than room temp (30C) going back to your lights is probably practical. look up at the chart and see what 30C heatsink temp vs 50C heatsink temp is - its 5% bump at most so now you only have 50W of savings to work with to cool your lights. a circ pump and simple passive cooler can easily be 50W even without active chilling so youre always working against diminishing returns
in most cases its just as simple and cost effective to increase number of emitters in a given area to drop the current and case temp of the chips and achieve same results
Last edited:
nevergoodenuf
Well-Known Member
I tested the PAR output of a CLU058 3618 @ 250w on an Ice cold day (almost need glove to hold it). From start up to fully warmed up, I only saw a 10 umol difference (1400 to 1410). The reason I quit playing with watercooled COBs.
CobKits
Well-Known Member
that is commensurate with my experiences. starting in a chilly warehouse with a pinfin heatsink at 10-15C, the first par measurement you get while everything is cold is only nominally higher vs steady state. temperature is definitely a factor in efficiency but really only significant when you have a lot of it and are not dissipating it effectively, not where most of us operate our cobs in the firstplace. its rare to see people operating chips higher than 50% of their max current. even the 3618 at 250W in nevergoodenufs example is about half of max current for that chip
ketchup45685224
Well-Known Member
The efficiency drop tracks closer to an increase in temperature than a increase in current. With air cooling, the degree of cooling is limited by the size and shape of the heatsink and the airflow over it. Water cooling gives you the opportunity to have a lot more cooling. How big would an air cooled heatsink have to be to compete with this? https://www.ebay.com/itm/Outdoor-Wood-Furnace-Boiler-Water-to-Air-Heat-Exchanger-36X36-american-royal/151702549444?ssPageName=STRK:MEBIDX:IT&_trksid=p2060353.m1438.l2649 with a 5 ton air condenser fan pulling air thru it. Or water coming out of a water well that is rarely over 55 degrees.
It is true that LEDs are more efficient the softer you drive them. In an air cooled system there is very defined relationship between drive current and efficiency. With a well designed water cooled system the relationship is much less dramatic. There are limits but it is much more forgiving.
I think it's more important to keep upfront cost down. I'm never going to spend over a thousand dollars to replace a 1000 watt HPS.
CobKits
Well-Known Member
yes its one of several limiting factors including Rj-c and Rc-h
if you want to discard water, which is not very green
agreed. also apart for cost a passive heatsink cannot fail. ask @ttystik about the perils of a complex watercooling system when fittings leak, you have condensation issues, pumps go down, etc etc
ketchup45685224
Well-Known Member
Rj-c is defined in the datasheet. I'd hope it is accurate. I've done some tests on vero29 with theromcouples on test point on LED and Heatsink to test thermal paste, Rc-h was not a problem. It's been awhile but I remember thinking "what was I worried about".
Not very Green? nothing about growing plants indoors is Green.... If we were trying to be green we'd be growing in a greenhouse or outdoors.
complexity. You win that one, without a doubt. I have a micro controller monitoring my room for temperature and humidity with an app on my phone to tell me if their is a problem. I could add some sensors to the water cooled system.
It's like I'm going to have to build something and test it. I have all the test equipment already to measure results.
Slinging PAR
Well-Known Member
Ah, but if it is worth doing then it is worth doing right!
Everyone will find their own starting point and continue their journey from there. I just know that at the end of the journey it will be as many LEDs as possible.