1W LS, how MUCH can it be overdriven?

[moses]

Ok, for you pros who have been modding for a while or have purchased modded items which you have tested. How far can we overdrive the 1 watters? For example, with items such as Mr. Bulk's creations, how high does the current run without ruining the 1 watters assuming proper heatsinking?

Thanks!
Mo

 

[FalconFX]

My LGI goes up to 1.7Amps in fresh-off-the-charger NiMHs without much fanfare... And my Cyclops digs up to 1.8Amps on fresh-off-the-charger C NiMHs...

Anything above 700ma for a 1W LS is probably a waste, figuring that it's proper wattage is only 1.25W, and some of us are driving it at 6.0+W DD!!! (3.75V x 1.7A) /ubbthreads/images/graemlins/shocked.gif

Needless to say, the LGI for me has become a barnburner if I didn't resistor it down when DDing it off of NiMHs... I'd be scared to see a 1W LS being driven over 1.0A constantly; can't be good for the life of the LS... /ubbthreads/images/graemlins/naughty.gif

 

[evan9162]

I think you hit diminishing returns after about 700mA. One night, I played with a royal blue 1W, driving it from 100mA to 1.3A. I also had the heat sink it was attached to immersed in ice water, so the LS never got much above room temperature. After about 800mA, I saw absolutely no brightness increase with increasing the current. Given the extremely good thermal conditions, I think I was replicating (at least for this sample) the best-case conditions.

-Darin

 

[moses]

THanks FalconFX and evan9162. Wow, 1.7 AMPS!!?? Goodness, that's 5 watt territory in terms of consuption.

I wonder if the 700ma stands true for the white LS as it is for the blue. I assume it is since the LED is basically the same.

Looking at the LS chart, light produced is almost linear to current passing through it, at least at the rated capacities. With 700ma max, then that's about twice the light of a normally driven LS. However, with people like Mr. Bulk, it seems that he is getting up to 3 times more light than a normally driven unit? SO the current beyond 700ma seems to be producing additional useful light?

Thanks again for the interaction. Very valuable.

Mo

 

[Doug S]

[ QUOTE ]
THanks FalconFX and evan9162.
Looking at the LS chart, light produced is almost linear to current passing through it, at least at the rated capacities. Mo

[/ QUOTE ]

Mo,
This is a case where you need to watch the fine print. Those output vs current graphs on the datasheets are for the junction held at a constant 25C. This is not possible in most typical applications. You need to adjust the output downwards using the output vs junction temperature graphs. You will note that different colors respond differently with rising temperature. The ambers are the worst. With amber, if the heatsinking is not good it is actually possible to reach the point where output drops with further current increase even before reaching the max rated current! I hope to cover this subject very soon in my long-running thread in the Mod forum on thermal design issues.

 

[Slick]

My "hardest driving" 1W Luxeon sucks down 1.879 amps...

That being said, I've found the optimal results to be 700 - 900mAh. Using my monster copper heatsinks in the 2AA size lights seems to keep them comfortable. I've built dozens like that, so far [knocking on wood] without failure [/knocking on wood].

 

[moses]

Doug - thanks for the reminder on heat. Yes, I have forgot to take that into account. So I take (without looking at the specs again) that higher temperature reduces the efficiency? Obviously with a metal light, I'll try to make the entire body a heat sink.

Slick - Yes, I'd like to use copper too. Do you know of any source that can custom cut copper disks for heatsinking for affordable prices?

Thanks for the comments.
Mo

 

[Slick]

[ QUOTE ]
moses said:
Slick - Yes, I'd like to use copper too. Do you know of any source that can custom cut copper disks for heatsinking for affordable prices?

[/ QUOTE ]

I have my heatsinks punched by a local machine shop. I buy them by the 100 to keep prices a bit lower. Unfortunately, using a punch limits the sizes of heatsinks you can get. I'm looking into having some cut with a laser tho...

The ones I currently use in AA size lights are .655 diameter and .125 thick. I expect that when I have a little more time on my hands that I'll post a sale add on them in the B/S/T forum.

If you're building an AA size light, feel free to email me on the heatsinks and we'll discuss off-line.

 

[Entropy]

It's pretty well known that LEDs are more efficient at lower currents. (Probably partly due the the fact that junction temp is so hard to regulate) In most cases (except for EXTREME heatsinking), two 1W LEDs driven at 300 mA will usually have more light output than a single 1W @600.

 

[Peter Atwood]

You might consider another material rather than copper for the heat sinking. Copper is heavy and holds onto the heat whereas aluminum tends to dissipate it quicker and is still a fine conductor.

 

[Slick]

[ QUOTE ]
Peter Atwood said:
You might consider another material rather than copper for the heat sinking. Copper is heavy and holds onto the heat whereas aluminum tends to dissipate it quicker and is still a fine conductor.

[/ QUOTE ]

I agree copper is definitely heavier (in weight) than aluminum.

Could you explain how aluminum is a better heatsink than copper, so that I can better under stand this?? /ubbthreads/images/graemlins/confused.gif

 

[Deleted member 684]

Slick,

Copper conducts heat quickly away from a heat source quickly, but doesn't transmit the heat to air as quickly as aluminium. Aluminium doesn't conduct heat so quickly, but loses it better than copper. In the end, the overall balance is that aluminium does transmit and looses heat better than air, unless you use a denser medium of conducting heat away such as water or use a lot of air.

1 watt + Alu + Air cooling = good
1 watt + Cu + air cooling = a lil less good depending on how you sink it
1 watt + Alu + water cooling = very good
1 watt + Cu + water cooling = Very Very Good

Copper needs a good medium to conduct the heat away. Air is bad medium

 

[Slick]

Well if that's true, I'm thinking that someone better tell all those engineers that are designing copper heatsinks (to upgrade from aluiminum) for computing applications that they are wrong and just wasting their time...

At least now I know that blindly following the advice of thermal engineering experts is foolhardy.

 

[Bill.H]

Slick,

I think the computer heatsinks aren't directly comparable because they are fan (or even water) cooled. There was an article at tomshardware explaining more about the Al/Cu thing a while back, I don't remember all the details but you can still find it there if you wish.

Now a cooling fan in a flashlight would be, well, cool, but it would tend to hurt battery life. /ubbthreads/images/graemlins/wink.gif

 

[tvodrd]

Just $.02 on the Al vs Cu thermodynamics. Al has about 2 and a half times the specific heat of Cu. (Takes 2.5 times the watt-hours to raise the temperature of a pound of Al 1deg as for a pound of CU.) When you factor in Cu's density is about 3.3 times that of Al, a given size ("cubes") Cu sink is the better choice. Cu has about 2 and a half times the thermal conductivity of Al. The increased thermal conductivity is just a bonus.

Larry
(Us water-computer idiots mostly do it for the noise reduction /ubbthreads/images/graemlins/grin.gif .)

 

[PsycoBob[Q2]]

Well, and bragging-rights. (watercool everything, including the southbridge. /ubbthreads/images/graemlins/buttrock.gif )

Back on topic- The hardest I've driven a 1w LS was 4v @ 3.22A, on a monster copper block, with a 0.75" water-cooling line in it, going to an ice-chest full of ice/water. Didn't get much brighter after the first 600-700mA, unfortunately. Anyone want to make a finned silver heatsink, and drop it in a bucket of liquid nitrogen, so we can try powering the poor LS with a car-battery? /ubbthreads/images/graemlins/twak.gif

 

[Slick]

[ QUOTE ]
PsycoBob[Q2] said:
Anyone want to make a finned silver heatsink, and drop it in a bucket of liquid nitrogen, so we can try powering the poor LS with a car-battery? /ubbthreads/images/graemlins/twak.gif

--------------------
Damn. Oh well, on to plan B I guess... >sets fire to head<

[/ QUOTE ]

This distinctly sounds like a project for someone with a name like "psyco" Bob! /ubbthreads/images/graemlins/grin.gif /ubbthreads/images/graemlins/grin.gif /ubbthreads/images/graemlins/grin.gif

- plus you'll wind up having something nice to put your head out in /ubbthreads/images/graemlins/wink.gif

 

[Rothrandir]

hey bob, why don't you tell everyone what actually killed the poor thing? /ubbthreads/images/graemlins/grin.gif

/ubbthreads/images/graemlins/yellowlaugh.gif


also, about the "heat-suckage" thing...

if i were to make a heatsink for a minimag drop-in, whould it be better to use copper, or aluminum.

obviously, one would think copper, but being as the copper "sucks" heat better, would it hold a little bit from going into the al? or would that be offset by the better overall conductance of copper? would al work better because it is better mated to the al body of the light?

i asked my teacher this several weeks ago, and she said copper would be beter because it is a better conducter, but i still wonder...

 

[PsycoBob[Q2]]

yeah... That one was the FIRST one- the Greenie was on a Golden Orb CPU-cooler. I think that one only got to 1.6A

The second one was a cyan 1w I got to see if I wanted to spend the loot on a pair of cyan 5-watters. (warning: Cyan seemt to be the perfect color to go thru welding-lenses. Don't bother.)

As for the Cu/Al question, the Cu should perform better... by a few percentage points, asuming identical size, and perfect thermal contact. You can always make it a 'Special Edition' by using silver. /ubbthreads/images/graemlins/bowdown.gif

Silver isn't as expensive as you might think- but more than I intend on paying for a hobby. Anyways- Aluminum won't have solder stick to it, Copper will. Both have their advantages (Al is easier/cheaper to work).

 

[hotfoot]

I thought copper was used only in the <i>core</i> of many of those CPU heatsinks. The copper then conducts heat away to the secondary finned aluminum heatsink that makes up all or the bulk of the air-exposed surfaces. The supposed logic of using copper at the heatsink core (or as a primary heatsink) is because its conductivity is better than Al, so it wicks heat away from the CPU faster and then dissipates it to a larger immediate surface area for "relay" to the lighter/cheaper/easier-to-work-with finned Al heatsink. Bottom line is to distribute heat away from the source ASAP and copper seems to be the best common metal to do this.

Oh - back on topic - I too agree that most luxeons don't fare much better when overdriven more than 80-100%. For eg, a Q3J driven at 500ma isn't much visibly brighter than one driven at 700ma (at least to my eye). But it *is* a LOT hotter.