Improving on HeatSyncing STARs (an IDEA)

[evan9162]

[ QUOTE ]
VidPro said:
[ QUOTE ]
evan9162 said:
A Luxeon I running at 350mA on the star board with no heat sinking in free air will stabilize to 60C board temperature. This puts the junction temperature at 80C. Increasing the current will both increase the stabilized temperature, and the junction temperature (both by increasing the stabilized temp, and increasing the thermal differential between junction an board due to increased power dissipation).

It sounds like you just need more heatsinking period.

The vast majority of power in an LED is converted to heat. Consider that the best lab samples hit 25% efficiency. Only about 10-15% of the power dissipated in a Luxeon gets turned into light. So 3W of power in means 3W of heat to dissipate.

[/ QUOTE ]

well that answers 2 question.
so is a 80* junction temp a bad thing?
would you say that something like these 3W drop-ins (pr2 replacements) run low enough to achieve 80*?

does 3 LUXIIIs in a Mag body, with a 2" heat sink stay at 80* ?

does anybody have lights that are achieveing the 80* when they are encased , and 100% sealed, when they are run for a full hour?

better question, would lots of these lights melt plastic if it was put on the heat syncs they use?

[/ QUOTE ]

80C is okay...lower is obviously better. With the heatsinks I make for Mag mods, I keep my Luxeon V mods get to a MAX of 85C Tj, after 45 minutes of sitting still. My Lux I at 580mA Mag gets to a maximum Tj of 55C. The 1.4A R/O I just built, max Tj of 65C.

Some of the Lux III drop-ins are terribly designed - I'm mostly talking about the Diamond one. Craig (LED Museum) reviewed one, and the temp of the drop-in got to 95C! It was running at 700mA, so the Tj there is just under 130C, or right at the thermal limit...a poor design indeed.

On the other hand, the EverLed has built-in thermal management, that prevents the luxeon from overheating - that's a well designed drop-in with proper thought given to thermal issues.

Keeping a Luxeon III running at 1A with a Tj of 80C or less is quite difficult - you would have to keep the heatsink at around room temperature. I'm sure most if not all tri-Lux III mods end up with a Tj of 100C or more.

I have a plastic body light that I put a luxeon into - but I keep the current to around 250mA, otherwise the heatsink inside just gets too hot.

I'm not sure if these would get to melting plastic temps before some kind of catastrophic failure happened to the Luxeon...

 

[VidPro]

Great thanks for all the answers.

well, i have now tested this theory of small copper transfer of heat , leaving the star exposed to the air.
and it works and it doesnt.

the copper transfer area was done using a 12Guage wire flattened slightly, it runs up and down, and is the length of the stars sink.
I used the wire, because i knew it was real copper and probably highest in actual copper content.
it also raised the height of the GAP, more than i was going to originally.

the copper transfer, covers mabey 1/8th of the star base, and transfers enough heat to the base to get it almost hotter than the stars own heat sink.
Meaning, it moves at least 1/2 of the to air transfer off the star.
it goes through the CENTER, where the emitter is, so that is probably why it is very efficient still at transfering the heat.
its hard to tell if there is any or enough convection or air movement between the 2 sinks , there is a Gap, big enough for air, what it accomplishes i dont know.

about the only thing that i can weed out from what i am feeling is, the star was better aluminum, because its almost colder than the base.
and the SMALL transfer area must be moving an extrodinary ammount across for its size , because the base is getting loads of heat.
i have effectivly split the heat across all the areas that it could depart into the air.

overdriving it, the bracket (base sink) gets very hot to the touch, touchable but barely, its scarry hot, but usable still.

so i conclude that i have lost very little in 'to the sink' transfer, and i am not at all sure what i gained in air contact, other than it is working.
i think it would have worked the other way also.

 

[VidPro]

[ QUOTE ]
n_den said:
[ QUOTE ]
VidPro said:
[ QUOTE ]
n_den said:
Currently, the best is to use the whole star surface.

The data presented below was designed for worst case scenario for 1 LuxIII.

[/ QUOTE ]

so via those stats, are you saying that a good sync would have 7 square inches of to the air dissipation?

[/ QUOTE ]



A MINIMUM of 7 sq. in. of vertical surface area,

[/ QUOTE ]

I have one LIGHT that is 10 LUXIII stars on a 10 Foot x 1 inch aluminum bar, the real heat barely makes it 2inches from each star before the rest of the metal is cool.
THAT thing , is probably properly synced , it works great, but it rarely gets overdriven.

Everything else i got gets really warm, the flashlights get HOT.

 

[HarryN]

Part of the confusion is a problem with semantics. What many people are referring to as a "heat sink" (this metal attached to an LED or star) is actually a "heat spreader". It has essentially no ability to "sink / destroy" the heat except for the first minute of operation when it is warming up.

Heat "sinking", which might be better called "heat dissipation" in a flashlight is mostly performed by the heat entering your body and being removed by sweating, which puts its upper limit somewhere around 3 - 5 watts continuous.

Heat dissipation by metal to air in quiet air is somewhat complex, but ends up at requiring at least 3 square inches of exposed metal area per watt to be readily dissipated.

 

[VidPro]

Harry
ya well now you tell me
when messing with them, i often run them in the air, using the star as a sync.
its aluminum, its exposed to the air, that makes it as viable a actual heat sink as a 1/2" piece of aluminum
while it might not meet the capacity for actually sinking the whole thing 100% , ad therfore a "spreader" it doesnt make it any less capable of being a heat sink.
say for example you run a luxIII at 100ma, its a sink, at 1000ma its barely a spreader /ubbthreads/images/graemlins/smile.gif

my IDEA was , that it could come in contact with AIR, and now it does, so its a sink now /ubbthreads/images/graemlins/smile.gif

if they REALLY wanted it to only be a spreader, then they shoulda used copper. why dont they use copper? would that 12c extra add to the cost so greatly?
some of the others are using high conductive ceramics in places.

passive air transfer is not so great at getting the heat out, but convection can turn passive into partly active.
with enough heat, and a chimney you can , , , even cool a small cpu /ubbthreads/images/graemlins/smile.gif they do that with some chepo quiet computer.

they are building a big arse chimney solar thing in austrailia, and it moves so much air, the "turbines" are put into the air.
so ideas on chimney convection , might be ways to increase air transfer .
like that big 3inch head for the mag light, with its sink rings.

 

[asdalton]

Natural (unforced) convection doesn't work well when things are small, or when the temperature differences are modest. Under these circumstances, the viscosity of air is too much of an impediment. What works inside of a hurricane, industrial smokestack, or power plant cooling tower won't work very well inside something the size of a computer CPU case.

Natural convection is particularly ill-suited to dissipating heat from a computer processor or LED--since the goal in these cases is not simply to transfer heat, but also to keep the source as cool as possible. Natural convection works best when the heat source is very hot relative to the ambient air.

There are ways of estimating when natural convection will be important for a given system. To initiate any convection at all, you need a sufficiently high Rayleigh number. For turbulent flow, which is much better at transferring heat than laminar flow, you also want to have a high Grashof number. The math looks uglier at first than it actually is. /ubbthreads/images/graemlins/smile.gif

 

[VidPro]

right, but working well, and increasing cooling ANY ammount at all, is different.
look at the CRT monitor, and 1/2 of the components of the home, including the light bulb they all use convection for cooling.

if it wasnt for the heat moving the air, the stagnation of the heat, in things like stereos with passive sinking would croak after a limited ammount of time.

 

[asdalton]

[ QUOTE ]
VidPro said:
right, but working well, and increasing cooling ANY ammount at all, is different.
look at the CRT monitor, and 1/2 of the components of the home, including the light bulb they all use convection for cooling.

if it wasnt for the heat moving the air, the stagnation of the heat, in things like stereos with passive sinking would croak after a limited ammount of time.

[/ QUOTE ]

Light bulbs are cooled mainly by thermal radiation, not convection. And to do that, the filament temperature is about 3500 K.

Other electrical devices have *far* less power to dissipate per area than a CPU processor or a Luxeon LED. To see how vastly the power/area ratio can vary, consider the following:

Human body--
Area: 1.8 m^2
Thermal power: 100 W
Power/area: 56 W/m^2

Luxeon III driven at 3 W and attached to 2 cm x 2 cm heatsink--
Area: 0.0002 m^2
Thermal power: 3 W
Power/area: 7500 W/m^2

So even though the Luxeon produces roughly 1/100 the heat of the human body, this particular Luxeon/heatsink arrangement has to dissipate over 100 times the power per area.

Power/area matters because heat transfer typically follows this type of equation:

Q = h*A*(Ts - Ta)

Q = total heat load
U = heat transfer coefficient
A = surface area
Ts = surface temperature
Ta = ambient temperature

So ...

Q/A = h*(Ts - Ta)

So if Q/A (power per area) is high, then either the surface temperature Ts must rise to a very high value, or else we have to find a way to increase h. In air, the only practical way to do this is through forced convection. We can't allow Ts to rise too much because the whole idea is to keep the component cool in order to avoid damage.

Yes, natural convection could increase h a little bit. The problem is that we need more than incremental improvements; we need hundred-fold improvements. Otherwise, the LED overheats and lets out its magic smoke.

 

[VidPro]

and
that is exactally what i did, increased the surface area, by exposing the star itself.

 

[twentysixtwo]

Based on what little I learned of heat transfer when I wasn't sleeping, and based on your description, you're better off without the air gap. The reduced contact with your sink will reduce the heat transferred to the sink. In general, you want the largest sink with the best contact, and you're sacrificing both. Taken to the next step, why isn't it better to go with the star alone?


If it was a choice between a star with lots of convection and a sinked star with none, well, you'd have to do the math.

Seems you're pretty convinced that your idea has merit. It's possible, though I'd want to see some pictures and data. One easy way to test would be to drill through the PCB off to one side, next to the emitter. Put a thermocouple in there so you're (almost) taking the temp of the emitter slug.

 

[VidPro]

there are other ways of testing convection, other than math physics and actual science
i just did the "smoke convection test" this very scientific test of reality , will show if there is an increase of passage of air past a particular area.

and this o so scientific method /ubbthreads/images/graemlins/smile.gif shows that there is lots of turbulance and at least 2x the speed of the smoke around the heated components, blasting through the gap (chimney) quite effectivly, as effective as one can get in a little gap like that.

why put small serations on a heat sink. because they look good? not likley, heat sinks with serations, have better to air capacity, not a lot but better. they have entire cpu heat sinks that are effective because of the serrations (small gaps) in thier RODs they have comming off the copper base.

i am never saying this is ENOUGH sink,ITS NOT, you all proved that. in this project there will not be enough, like other projects before it.
it will get hot, and it will still run like other that have come before it, that have run for years.

i have run it for 4 hours, and its working, it might be toasting the thing, but i see it lasting longer than a 5W lux /ubbthreads/images/graemlins/smile.gif

 

[twentysixtwo]

You asked whether it would give a cooler emitter slug, not whether your design will live - it sounds as though it does, good luck in your future "experiments."

 

[VidPro]

[ QUOTE ]
twentysixtwo said:
Based on what little I learned of heat transfer when I wasn't sleeping, and based on your description, you're better off without the air gap. The reduced contact with your sink will reduce the heat transferred to the sink. In general, you want the largest sink with the best contact, and you're sacrificing both. Taken to the next step, why isn't it better to go with the star alone?


If it was a choice between a star with lots of convection and a sinked star with none, well, you'd have to do the math.

Seems you're pretty convinced that your idea has merit. It's possible, though I'd want to see some pictures and data. One easy way to test would be to drill through the PCB off to one side, next to the emitter. Put a thermocouple in there so you're (almost) taking the temp of the emitter slug.

[/ QUOTE ]

and if i had a thermocouple , i probably would have a 7in heat sync too /ubbthreads/images/graemlins/smile.gif
i mean if i KNEW the kind of torture this thing was getting, i probably would throw it in the trash and start all over again, with a 10inch plate on my head , to be a proper heat sink.
i have a few thermal devices laying about, but they certannly would not provide any correct readings of any sort, unless the speed of a thermal adjustable fan would be usefull data /ubbthreads/images/graemlins/smile.gif

i think i would need a real thermocouple, that had some excelent way of making complete contact, and reading accuratly. or better yet a camera that is thermal capable.

all they give me here is fingers lets see what temperature does fingers smell like moths burning?

 

[VidPro]

mine cant be Nearly as bad as this "3W" headlight, Wheres the beef /ubbthreads/images/graemlins/smile.gif

http://www.candlepowerforums.com/ubbthreads/showflat.php?Cat=&Number=981626
(about 1/2 way down you can see the huge heat sink they put on this headlite