My 1st 2D Mag SST-90/SSR-90 build...I have ??'s

Hi folks,

I am getting help from a buddy to build me non-pedestal heatsinks. It looks all that is available is raised on a pedestal. I will make 3 heatsinks; copper, aluminum, and brass. I will run the SST-90 direct drive with stock 2D Mag Switch. Powered by 3 NiMH Tenergy C cells or 4 SubC NiMH cells. Then I can see which heatsink material works best. Oh, and 24gauge wire will be used.


I wanted to purchase one of the SST-90 sold at 4sevens, but it doesn't have a star. If I will only use this direct drive; do I still need the SSR-90???

My goal is to get 1800 OTF lumens and still have 1500 OTF at 3~5 minutes So, far not even the mighty SSR90 from Olight can do this.

Thanks,
bigC

Hi Big C,

Some quick thoughts on your idea (I'm sure TJ will chime in since he has had vast more experience)

Use 20 - 21g wire (24g is only rated to 3.5A according to this chart)

SST-50 wo/ star is ok, but make sure it's firmly attached to the sink. As I'm sure you know, heatsinking is critical as this sucker produces massive amounts of heat.

Isn't the stock mag switch 5A max?? You may want to check on that.

Not sure about those battery combos. I know others have gone dd with NiMh. I want to say you may need to have 4 x SubC to get max output.

Good luck!

Hill said:

Hi Big C,

Some quick thoughts on your idea (I'm sure TJ will chime in since he has had vast more experience)

Use 20 - 21g wire (24g is only rated to 3.5A according to this chart)

SST-50 wo/ star is ok, but make sure it's firmly attached to the sink. As I'm sure you know, heatsinking is critical as this sucker produces massive amounts of heat.

Isn't the stock mag switch 5A max?? You may want to check on that.

Not sure about those battery combos. I know others have gone dd with NiMh. I want to say you may need to have 4 x SubC to get max output.

Good luck!

Click to expand...

I will go for the 20~21g wire then.

The stock Mag D size switch I been reading is good up to 10A.


I figured that I will need 4 NiMH C cells to run the LED at full 9~10A since 3 of them might give me just 3~5A:thinking:

Now I still not sure if I can purchase the sST-90 without the Star or not:candle::candle:


EDIT: I am going with an SSR-90 from Photon Fanatic and at $60 dollars those 65k color ones are not cheap.

I am doing a heatsink similar to this one from Lambdalights.com, but I plan on putting a lip on mine and making it 1.5in~1.8in in total lengh.





Thanks,
bigC

This will definitely be interesting :thumbsup:, but my money is on the aluminum, then again, I'm not a betting man and there's a reason for that.

How are you planning to attach the star to each heatsink? Will you drill and tap each heatsink so you can tighten the star down with screws? If so, I'm assuming you will apply AS5 under the star for each heatsink.

Also, I have very little experience with direct drive applications, but I do know that you can fry one of these emitters very quickly if you're not careful. I fried one pretty darn fast with a Li-ion, but I gained some knowledge from it; when the tint starts shifting to blue, see just how fast you can cut power, then let the emitter cool for a minute.

I don't really have much experience with NiMH cells either, but if they have a nominal voltage of 1.2v per cell, 4 cells would be 4.8v. The SST-90 datasheet says a typical vf of 3.87v at 9A and 3.9v max. I would also keep in mind the "burn-in" that Techjunkie discovered which lowers the vf of the emitter. Click here for the thread. If you lower the vf, the current will likely go up. Like I said though, my experience with direct drive applications is limited, so I'm just thinking out loud here.

bigchelis said:

I will go for the 20~21g wire then.

The stock Mag D size switch I been reading is good up to 10A.


I figured that I will need 4 NiMH C cells to run the LED at full 9~10A since 3 of them might give me just 3~5A:thinking:

Now I still not sure if I can purchase the sST-90 without the Star or not:candle::candle:


EDIT: I am going with an SSR-90 from Photon Fanatic and at $60 dollars those 65k color ones are not cheap.

I am doing a heatsink similar to this one from Lambdalights.com, but I plan on putting a lip on mine and making it 1.5in~1.8in in total lengh.

Thanks,
bigC

Click to expand...

Hey bigC,

That's a sweet heatsink!! You do that yourself? All my custom sinks are painstakingly had cut and glued. They work, but are a difficult to make and never look pretty.

I'm betting on the copper sink to be the best.

Using my knowledge from PC CPU cooling, the aluminum will be worse than the copper. The CPU cooler people never use brass so either it's too expensive or isn't a good enough conductor to compete with brass.

The evolution of PC CPU cooling stared out with aluminum, then the better companies started using copper baseplates that transitioned to aluminum fins/pins, then all copper, then heat pipes came on the scene.

When absolute max performance is required, all copper is employed regardless of the cooling technique. So my vote goes for the copper. Make sure you use your AS5 when install that slug in the tube.

Travis

Hi Bigchelis,

Cool project. I will be interested to see how this works out.

In my xp 20 awg wire has perfect for this type light. I use Accupower C cells in mine and on only 3 cells DD current draw was about 11-12 amp. Depending on what you use to measure the current you may need the extra cell during your test. I made a set of leads for my DMM using some thick copper wire ( 12 AWG ish maybe) to more accurately simulate the output with the tailcap on.

Good luck and thanks for sharing

If you're going to use 20g wire, I would get some silicon wire from here:


http://www.radicalrc.com/shop/?shop=1&cat=64 $3 for 1 meter of red & 1 meter of black

You could always get 4 sub-c cells, then drive it with 3 sub-c's first, and check what the voltage drop is, and how much current you're putting to the LED. A 4th cell may make the LED go , which I don't think you want!

~Brian

Thanks for the link to the wire.

FroggyTaco said:

When absolute max performance is required, all copper is employed regardless of the cooling technique.

Travis

Click to expand...

I know very little about thermal transfer rates with different materials, but I do know that copper tramples aluminum in thermal transfer. I just wonder if the heatsink will get saturated, or too hot, too quickly, but maybe the copper will transfer the heat to the aluminum body quickly enough that it will keep the heat moving. My choice of the aluminum heatsink winning in this showdown was really just an uneducated guess , and I thought I'd root for the underdog.

ti-force said:

I know very little about thermal transfer rates with different materials, but I do know that copper tramples aluminum in thermal transfer. I just wonder if the heatsink will get saturated, or too hot, too quickly, but maybe the copper will transfer the heat to the aluminum body quickly enough that it will keep the heat moving. My choice of the aluminum heatsink winning in this showdown was really just an uneducated guess , and I thought I'd root for the underdog.

Click to expand...

You hit upon the the crux of the issue. Can the entire heatsink system ie: the emitter heatsink & the flashlight body cope with the heat dump of the emitter.

The copper is better not only in heat transfer, but also heat absorption capacity. Think of it as a bigger sponge if relating to absorbing water.

Travis

p.s. I like rooting for the underdog, but since there is no heart in the matter were screwed! :buddies: :lolsign:

Three heatsinks from different metals. This should be interesting. I'll be curious to see the results of that comparison. Do you have a DMM with a thermistor probe? I have a really cheap DMM. When I put together my SST-90, I thought about buying a small SMT thermistor. I was going to make a chart of the resistance vs temperature for say every 5°C from the mfgr. data. I had nothing to compare it to. I just wanted to see how long a Mag-D with an SST-90 can run before it gets too hot. I was in too much of a hurry to see a 2,000 lumen flashlight though.:devil:

I use the same table Hill pointed to for wiring. Maybe you can make some good use of wire that's "too" small. It would act like a built in resistor. The 4/5SubC Sanyo NiCad's I bought have a 5.5 milliohm internal resistance @ 50% charge according to Sanyo. They only drop about 1/8 volt under the 10.25 amp load my DIWdiver regulator puts on 4 of them. Under load they measured 4.96 volts. The Vf of my LED is 3.88 volts at the moment. Three of these high drain batteries would only be about 6 amps. And 4 of them would definitely blow the LED. I think if you get the highest capacity SubC you can find would be the best for a 4 battery direct drive setup. High drain and high capacity are to some extent, mutually exclusive. So high capacity cells and thin wire might work very well together. The bad thing about direct drive testing with a DMM on the amp scale is it will take more power when you remove the meter. Unless of course you are lucky enough to own a clamp meter. Keep an eye out for an "angry blue" LED.

I'm betting that a solid hunk of copper will not work as well as aluminum or brass. Brass and aluminum are about the same. Since brass is so much more expensive, the only reason to use it is if you want to solder to it. The specific heat (amount of heat required to raise the temp) of copper is much lower than aluminum. So copper gets hotter faster. Even though copper has more than double the thermal conductivity of aluminum. Specific heat as a measurement involves the weight of the material as well. Even after you factor in copper's higher density, aluminum still has a better specific heat characteristic than copper.

I wouldn't have picked copper (read stole the design:naughty for my heatsink core if I didn't think it had awesome thermal properties. As FroggyTaco said, modern day CPU coolers all use copper. I just think it depends as much on good design as on the properties of the metal.

The best material for heasinking is silver (430 W/mK) ... coper is 401 W/mK, gold 314 W/mk (!) Al is 230 W/mk and brass is 120 W/mk .... coper is able to store lot more heat compared to Al. I always use coper for my SST/SSR builds -> In Germany we call it Wiedemann-Franzschen Gesetz. It explains the reason for this behavior (siehe auch: Pauli Prinzip)

It always impresses me just how much knowledge one has to aquire when going into something like this which covers so many areas. As many people have found, it's not just a case of whacking it all together. Current carrying capacities, and accumulated resistance in the circuits all start to get individual attention. I'll be watching this thread to see how things develop. Good work all!

Al Combs said:

I'm betting that a solid hunk of copper will not work as well as aluminum or brass. Brass and aluminum are about the same. Since brass is so much more expensive, the only reason to use it is if you want to solder to it. The specific heat (amount of heat required to raise the temp) of copper is much lower than aluminum. So copper gets hotter faster. Even though copper has more than double the thermal conductivity of aluminum. Specific heat as a measurement involves the weight of the material as well. Even after you factor in copper's higher density, aluminum still has a better specific heat characteristic than copper.

Click to expand...

Yes, the specific heat of Al is about 2X that of copper. But that isn't the relevant spec here. Since the heat sink is going into a fixed volume (a 2D Mag body/head), the relevant spec is probably thermal volume. Copper is superior to aluminum on a per volume basis by about 50%. Of course, one could always make the relevant spec to be specific heat by making the max acceptable mass your constant instead of volume. But then the heat sink using a more dense material would get shortchanged because it would be smaller in volume that the heat sink using a less dense material. There would be little point, IMO, to make a heat sink undersized for the available volume, especially when you are looking at something as high-powered as an SST-90/SSR-90.

BigC, if you can max out the power delivered to the SST-90/SSR-90, you will be looking at around 32W of power draw by the LED, or perhaps about 25W of waste heat generated that has to be dissipated by your heat sink and Mag host. That is a lot of heat. In some ways, it is fortunate that you aren't going to use a driver because between the driver's waste heat and the LED's waste heat, you'd have a real thermal management challenge to keep the system happy.

The heatsink is actually something Lambdalights has been working on to help with his heat issues on the Vara2000 SST-90. He gave me his specs and pictures of his for me to attemp to duplicate it, since he already saw improvements with his heatsink.

The heatsinks will have holes big enough for 14g wires to fit and I think I should try both 14g and 20g just for kicks.

I will try SST-90 with 3 NiMH cells 1st and if I get sub 5A curret then I will try 4 NiMH cells. I hope I get full power with just 3 cells, but so far it looks like the 4 NiMH cells will win.

Two words: liquid cooled.


Subscribed.

kito109654 said:

Two words: liquid cooled.


Subscribed.

Click to expand...

If we look to PC CPU cooling, the next evolution is heatpipes. No moving parts, no maintenance, & amazing ability to literally move heat to a different part of the host.

How that would work in a ergonomic flashlight is beyond my comprehension though.

Travis

p.s. I am on my 5th water cooled PC so I am all for it but.....

FroggyTaco said:

If we look to PC CPU cooling, the next evolution is heatpipes. No moving parts, no maintenance, & amazing ability to literally move heat to a different part of the host.

How that would work in a ergonomic flashlight is beyond my comprehension though.

Travis

p.s. I am on my 5th water cooled PC so I am all for it but.....

Click to expand...

I'm thinking of a single unit head and heatsink/heatpipes w/reflector and glass held down by bezel. Of course extending to the PC world I'd like to see where the LED attaches via pins or a ball grid array and becomes upgradeable like a CPU.

Justin Case said:

Yes, the specific heat of Al is about 2X that of copper. But that isn't the relevant spec here. Since the heat sink is going into a fixed volume (a 2D Mag body/head), the relevant spec is probably thermal volume. Copper is superior to aluminum on a per volume basis by about 50%.

Click to expand...

Pure copper has about 40% more volumetric heat capacity than pure aluminum. Funny all the sites I Googled for specific heat were a little different. Anyway I think the relevant spec is a type of aluminum someone would actually use in a flashlight. Here is a data chart showing T6061 aluminum vs C110 copper. Not the last word, but typical of what I saw. The adjusted volumetric specific heat put copper in the lead by only 1%. If you use draw copper then T6061 is the winner by slightly under 3%. Too close to call I'd say. My mistake guys, I thought aluminum was the clear winner.