Luxeon flashlight heat temperature calculations: DIY formulas.

dat2zip · Sep 18, 2002

Here are my calculations and extrapolated data for the minimag and badboy running 400mA through a 1W Luxeon with a Vf of 3.4V.

The minimag thermally stabilized to 115F on the case in ambient 25C room temperature.

From this and from the total power being drawn we can now compute the Thermal Resistance of the flashlight body to Air (Rfba).

The total power is (V * I) / efficiency = (3.4V * 400mA ) / 0.85% = 1.6W total.

Using the equation Rfba = (Tfb -Ta) / Total_power = (115F - 77F) / 1.6W = 23.75 F/W = 13.19 C/W.

Where:
Tfb is the Temperature of the Flashlight Body (we measured 46.1C (115F))
Ta = Temperature of Ambient or 25C (77F)
Rfba = Thermal Resistance of Flashlight Body in degrees Centigrade per Watt.

(This means if there was 1W of power the flashlight body would rise 13.19C above ambient. Two watts the flashlight body would rise twice this. etc etc etc...)

Armed now with 13.19 C/W Rfba we can now compute our worst case Luxeon die temperature for the Badboy minimag mod.
The three thermal resistances are:
Rfba = Thermal resistance of Flashlight body. 13.19 C/W.
Rbbfb = Thermal resistance of badboy board to flashlight body. (Assume 15 C/W)
Rslg = Thermal Resistance of Luxeon emitter aluminum slug. (15 C/W)

For the 400mA badboy we know we have 1.6W so the die temperature is the ambient plus the sum of the thermal resistances of each junction.

Ta + TRba + Tbbfb + Tslg = 25C + (1.6W * 13.19) + (1.6 * 15) + (1.6 * 15) = Tdie = 94 C

This is the nominal. Now worst case would be on a hot day at 110F or 43.3C.

Ta + TRba + Tbbfb + Tslg = 43.33C + (1.6W * 13.19) + (1.6 * 15) + (1.6 * 15) = Tdie = 112C

So, worst case we have not exceeded the 125C of the die of the emitter and we are operating the die more than 10C even under the worst case of 110F.

==============================================
Now let's do some quick computations for 500mA or our Badboy_500 mod.

The thermal resistance model is similar with the exception that we are now pushing 2.05W in total.

I'll only compute the worst case as an example.

Ta + TRba + Tbbfb + Tslg = 43.33C + (2.05W * 13.19) + (2.05 * 15) + (2.05 * 15) = Tdie = 131.86C

This means that we are overtemping the Luxeon by 6.86C. TO keep from exceeding specifications we would need to drop our maximum temperature specifications of 110F by at least 6C or more.

Subtracting 10C for safety from the 125C die specification I calculate that the Badboy_500 in the minimag I have will be safe up to 26.46C or 79F.

NOTE: These calculations are for zero air flow. Moving the flashlight increases the air flow across the flashlight and significantly lowers the Thermal resistance of the flashlight body to ambient air.

==============================================
Conclusion:

I hope this helps modders who are now working with the new 5W to be inspired to make the effort to ensure that they don't over drive and shorten the 5W Luxeon dies expected life.

This now confirms my belief that driving more than 500mA of constant current into a typical 1W luxeon is doing more harm overdriving the Luxeon for more light. The current 500mA can push the die past the LEDs die temperature rating and the Badboy_violet would definitely be doing some serious harm to the LED and overtemping it.

BTW: My calculations for the 5W greenie in the Surefire M2 housing is 10.88F/W (6.05 C/W)

This is based on 135F flashlight body temperature, 666mA Luxeon current, 6.8V Vf, and 85% efficiency.

Maybe, Doug S can verify the correctness of this post.

Doug S · Sep 18, 2002

Originally posted by dat2zip:
Here are my calculations and extrapolated data for the minimag and badboy running 400mA through a 1W Luxeon with a Vf of 3.4V.

The minimag thermally stabilized to 115F on the case in ambient 25C room temperature.

From this and from the total power being drawn we can now compute the Thermal Resistance of the flashlight body to Air (Rfba).

The total power is (V * I) / efficiency = (3.4V * 400mA ) / 0.85% = 1.6W total.

Using the equation Rfba = (Tfb -Ta) / Total_power = (115F - 77F) / 1.6W = 23.75 F/W = 13.19 C/W.

Where:
Tfb is the Temperature of the Flashlight Body (we measured 46.1C (115F))
Ta = Temperature of Ambient or 25C (77F)
Rfba = Thermal Resistance of Flashlight Body in degrees Centigrade per Watt.

(This means if there was 1W of power the flashlight body would rise 13.19C above ambient. Two watts the flashlight body would rise twice this. etc etc etc...)

Armed now with 13.19 C/W Rfba we can now compute our worst case Luxeon die temperature for the Badboy minimag mod.
The three thermal resistances are:
Rfba = Thermal resistance of Flashlight body. 13.19 C/W.
Rbbfb = Thermal resistance of badboy board to flashlight body. (Assume 15 C/W)
Rslg = Thermal Resistance of Luxeon emitter aluminum slug. (15 C/W)

For the 400mA badboy we know we have 1.6W so the die temperature is the ambient plus the sum of the thermal resistances of each junction.

Ta + TRba + Tbbfb + Tslg = 25C + (1.6W * 13.19) + (1.6 * 15) + (1.6 * 15) = Tdie = 94 C

This is the nominal. Now worst case would be on a hot day at 110F or 43.3C.

Ta + TRba + Tbbfb + Tslg = 43.33C + (1.6W * 13.19) + (1.6 * 15) + (1.6 * 15) = Tdie = 112C

So, worst case we have not exceeded the 125C of the die of the emitter and we are operating the die more than 10C even under the worst case of 110F.

==============================================
Now let's do some quick computations for 500mA or our Badboy_500 mod.

The thermal resistance model is similar with the exception that we are now pushing 2.05W in total.

I'll only compute the worst case as an example.

Ta + TRba + Tbbfb + Tslg = 43.33C + (2.05W * 13.19) + (2.05 * 15) + (2.05 * 15) = Tdie = 131.86C

This means that we are overtemping the Luxeon by 6.86C. TO keep from exceeding specifications we would need to drop our maximum temperature specifications of 110F by at least 6C or more.

Subtracting 10C for safety from the 125C die specification I calculate that the Badboy_500 in the minimag I have will be safe up to 26.46C or 79F.

NOTE: These calculations are for zero air flow. Moving the flashlight increases the air flow across the flashlight and significantly lowers the Thermal resistance of the flashlight body to ambient air.

==============================================
Conclusion:

I hope this helps modders who are now working with the new 5W to be inspired to make the effort to ensure that they don't over drive and shorten the 5W Luxeon dies expected life.

This now confirms my belief that driving more than 500mA of constant current into a typical 1W luxeon is doing more harm overdriving the Luxeon for more light. The current 500mA can push the die past the LEDs die temperature rating and the Badboy_violet would definitely be doing some serious harm to the LED and overtemping it.

BTW: My calculations for the 5W greenie in the Surefire M2 housing is 10.88F/W (6.05 C/W)

This is based on 135F flashlight body temperature, 666mA Luxeon current, 6.8V Vf, and 85% efficiency.

Maybe, Doug S can verify the correctness of this post.

Jumping right in by invitation! Hooray! Someone addressing thermal issues rigorously. Nicely set up. Here is my critique:
There is one more small thermal resistance to be considered. That is the emitter slug to board. As best I can tell, Lumileds acheives 2C/W for this resistance when they mount emitters on their MCPCBs. There is no reason to suspect that we can do any better so I suggest using that value by default.

Originally posted by dat2zip:

Rbbfb = Thermal resistance of badboy board to flashlight body. (Assume 15 C/W)

This assumption is very important to your conclusions and needs to be tested. The way I would do it is to use your BB emitter board and instrument for temperature. Install board in MiniMag in manner that replicates the normal BB mounting. Power directly [not using switching regulator] with measured current and voltage. After thermal SS is reached, take appropriate data to calculate actual board to body thermal resistance.

Originally posted by dat2zip:

I'll only compute the worst case as an example.

For the worst, worst case, I seem to recall that your BB efficiency at low battery voltage is not as good as assumed in your calculations above. Lower efficiency of course results in higher heat input.

Originally posted by dat2zip:

BTW: My calculations for the 5W greenie in the Surefire M2 housing is 10.88F/W (6.05 C/W)

I don't understand what you mean here. Per the 5W Luxeon datasheet, the thermal resistance of just junction to board is 11C/W.

Originally posted by dat2zip:

Conclusion:

I hope this helps modders who are now working with the new 5W to be inspired to make the effort to ensure that they don't over drive and shorten the 5W Luxeon dies expected life.

.

Yessss! Amen

dat2zip · Sep 18, 2002

quote:
--------------------------------------------------------------------------------
Originally posted by dat2zip:

BTW: My calculations for the 5W greenie in the Surefire M2 housing is 10.88F/W (6.05 C/W)

--------------------------------------------------------------------------------

I don't understand what you mean here. Per the 5W Luxeon datasheet, the thermal resistance of just junction to board is 11C/W.

Doug,

I'm just computing the Rfba in this equation. I only know the power, the flashlight temperature and the ambient. The remaining numbers were not computed and for the greenie I'm not going to do it since it gets real complicated real fast with the number of thermal transitions layers that are in this design.

You are correct. One of these days I should compute the board to flashlight housing thermal, but, in reality, it a mood point since it is the weak link in the drop in flashlight sandwich module. This is just to evaluate what margins there are. This is no way a product I could really stand by since the contact pressure, oxidation and contamination between the board and the flashlight housing is not a constant.

As such, I tossed in a conservative larger than average number as a guesstamate and that's how it'll stand.

I forgot the junction between the slug to board. Thanks for the pointer.

WayneY

Slick · Sep 18, 2002

Originally posted by Doug S:
Here is my critique:
There is one more small thermal resistance to be considered. That is the emitter slug to board. As best I can tell, Lumileds acheives 2C/W for this resistance when they mount emitters on their MCPCBs. There is no reason to suspect that we can do any better so I suggest using that value by default.

Doug, I'm curious about this. The Luxeons I have worked with (LS/o's) have all been "set" over a thin layer of the fiberglass-like PCB board material (with aluminum underneath).

Wouldn't affixing an emitter directly to a copper slug (with Artic Silver) be a distinct improvement over the "LS/o-type" of attachment?

If the emitter base were in direct contact with a copper heatsink, I would expect the thermal resistance to be lower than "stock." The glue that Lumileds used to mount these emitters to their boards doesn't appear all that thermally conductive either.. What do you think? - Thanks..

Doug S · Sep 19, 2002

Originally posted by Slick:
Doug, I'm curious about this. The Luxeons I have worked with (LS/o's) have all been "set" over a thin layer of the fiberglass-like PCB board material (with aluminum underneath).

Wouldn't affixing an emitter directly to a copper slug (with Artic Silver) be a distinct improvement over the "LS/o-type" of attachment?

If the emitter base were in direct contact with a copper heatsink, I would expect the thermal resistance to be lower than "stock."

I think that we are unlikely to do better than Lumileds. BTW in the "stock" arrangement it is copper to copper as the soldermask does not extend under the slug. There is an insulating layer under the copperclad but it is epoxy, not fiberglass, and is only 100um thick per [EDIT] page 3 of AB10. Worthwhile reading on this subject is AB05, AB10, and AB23, all of which can be downloaded from the Lumileds site.

Originally posted by Slick:
The glue that Lumileds used to mount these emitters to their boards doesn't appear all that thermally conductive either.. What do you think? - Thanks..

While my skills don't extend to determining thermal conductivity visually, it is inconceivable to me that Lumileds would not choose as conductive an adhesive as possible that still met their mechanical requirements. In AN01, they describe it as "heat conductive adhesive". They devote considerable space to discussing thermal conductivity issues so I don't think they would overlook this consideration.
Link to Luxeon data:
http://www.luxeon.com/products/documentation_index.html

hotfoot · Sep 19, 2002

Doug is right about the adhesives. In their downloadable datasheets (the one intended for manufacturers - can't remember which), Lumileds even go as far as specifying the specific thermal adhesives recommended.

BTW, the emitter slug is not electrically neutral and I can tell you from firsthand experiments that you really don't want the emitter slug forming an electrical path with the metal base you mount it on.

For single luxeon purposes, you'll probably never notice the diff, but you'll positively run into weird grounding problems if you have a multi-luxeon app, where several emitters connected in series share the same metal (or other electrically conductive) mount.

dat2zip · Sep 19, 2002

Lumileds now only has AB05 the others seem to have been taken offline.

Just FYI, check out some overclockering web sites.

Just for reference here are some thermal conductors greases and tape. The higher the number the better.

ASIII 4.65 to 5.15 W/m^2K
Circuit Works Silver... 4.167 W/m^2K
Radio Shack Heat sink Compound 0.4187 W/m^2K
3M thermall Conductive Adhesive #9885 0.43 W/m^2K

As I recall the last time I did a heat sink design calculations I did not find the standard white heat sink compound and tapes to be much different.

But, ASIII is about 10 times better than the radio shack Heat sink compound which is probably as good as the standard Wakefield Compound that is the everyday standby.

Most tapes, white goop, and rubber goop are pretty poor compared to the high tech ASIII and the Arctic Alumina. The other note is the Circuit works stuff rated pretty high on the one web site overclocking that I looked at.

As far as I'm concerned the tape stuff is junk, hard to apply, hard to manage. ASIII can be sqeezed down to a much thinner layer and get even better performance than almost anything I know of.

It's a testamont that using ASIII in my minimag that the aluminum disc and the barrel temperature are unmeasurably at the same temperature. If there was a 2C/W gradiant I would have noticed that by now.

Head to head comparson. (Overclocking site)

WayneY

Doug S · Sep 19, 2002

Originally posted by dat2zip:
Lumileds now only has AB05 the others seem to have been taken offline.

WayneY

My reference to AB03 was in error. I meant to say AB10. Yes, PN01 seems to be AWOL. For those interested, the link to the Luxeon data is here:

http://www.luxeon.com/products/documentation_index.html

dat2zip · Sep 19, 2002

I'm going to stick my neck way out on this one and say, that when I initially read the one that referered to the different thermal tapes and stuff I brushed it off.

It's my personal opininion that that was written by a newbie or fresh out of college person and the content to me didn't jive with my long established experiences. As such, I tossed the info into the trashbin...

I believe it is being totally re-written to address the issues that are more prevalent and that is addressing the 5W. Now this is a beastie with many dragon heads to be tamed. I say that since the die temperature max rating is reduced, 5 times more power is being applied, the total contact surface are of the aluminum die slug is the same. I've already shown that the 1W can be exceeded very easily. Taming the 5W compared to the 1W is like a kitten and a tiger.

Well, maybe not that bad...

hotfoot · Sep 19, 2002

If you check out Dan's Data, you'll find a comparo he did between some hi-tech thermal compounds and stuff you'd never expect to use in a computer or a flashlight.

What were his findings? According to him, ASII, vegemite (a yeast-based sandwich spread), and toothpaste(!) all yield similar thermal performance when fresh.

I'm just regurgitating something interesting here - all inferences you draw from this are entirely your own!

Doug S · Sep 19, 2002

Originally posted by hotfoot:
What were his findings? According to him, ASII, vegemite (a yeast-based sandwich spread), and toothpaste(!) all yield similar thermal performance when fresh.

At last! Something useful to do with Vegemite.

Doug S · Sep 19, 2002

Originally posted by dat2zip:
I say that since the die temperature max rating is reduced, 5 times more power is being applied, the total contact surface are of the aluminum die slug is the same. I've already shown that the 1W can be exceeded very easily. Taming the 5W compared to the 1W is like a kitten and a tiger.

Hmmm... My version of the datasheets for 1 and 5W dated 07/02 list a *higher* max Tj for the 5W vs 1W [135C vs 120C].

hotfoot · Sep 19, 2002

Originally posted by Doug S:
<blockquote>quote:<hr />Originally posted by dat2zip:
I say that since the die temperature max rating is reduced, 5 times more power is being applied, the total contact surface are of the aluminum die slug is the same. I've already shown that the 1W can be exceeded very easily. Taming the 5W compared to the 1W is like a kitten and a tiger.

Hmmm... My version of the datasheets for 1 and 5W dated 07/02 list a *higher* max Tj for the 5W vs 1W [135C vs 120C].<hr /></blockquote>Right you are Doug. But the same datasheet also states that in order to maintain or keep within that die temp limit, the metal core PCB must not exceed 75C, as opposed to the 105C allowed for the 1W. I think that's what Wayne was referring to.

dat2zip · Sep 19, 2002

Originally posted by Doug S:
<blockquote>quote:<hr />

Hmmm... My version of the datasheets for 1 and 5W dated 07/02 list a *higher* max Tj for the 5W vs 1W [135C vs 120C].[/qb]<hr /></blockquote>You're right. I was going by previous discussions and had not actually look at the datasheets.

135C for the 5W and 120C for the 1W. Sorry bout that.

Since the derating is for the metal core PCB it would be prudent to not use the anything but the bare emitter and maximize heat transfer to the outside world. Starting with the star or the LS/O version already puts you at a significant disadvantage.

If you look at the derating of light output it might actually be better to run it a reduced power since there would be less heat and the efficiency would be much higher. You might find that running full power generates no more light output than at some significantly reduced current since the efficieny goes way down as temperature goes up.

Vegamite?? What's that?

hotfoot · Sep 20, 2002

Originally posted by Doug S:
<blockquote>quote:<hr />Originally posted by hotfoot:
What were his findings? According to him, ASII, vegemite (a yeast-based sandwich spread), and toothpaste(!) all yield similar thermal performance when fresh.

At last! Something useful to do with Vegemite.<hr /></blockquote>Hey... I actually LOVE that stuff in sandwiches!

It's heavenly with butter on toast... *drool..*

Doug S · Sep 20, 2002

Originally posted by dat2zip:

135C for the 5W and 120C for the 1W. Sorry bout that.

Since the derating is for the metal core PCB it would be prudent to not use the anything but the bare emitter and maximize heat transfer to the outside world. Starting with the star or the LS/O version already puts you at a significant disadvantage.

Vegamite?? What's that?

Click to expand...

Wayne, I think you may be misinterpreting the spec for the 5W. The MCPCB *materials* can handle higher temps. That 75C limit is based on keeping the junction below 135C if operating at max current with a sample that has the max datasheet Vf. See note 3, page 4, of DS30 for clarification.

Vegemite: a food-like substance that seems to be popular in parts of the former British Empire but hasn't caught on much in the US. There are a couple of vaguely similar products, Marmite(sp) and Boveril(sp) that I mentally lump together with Vegemite though I think they are made from different starting material. One of these may be South African. All three are a consistancy to be used as a spread.

hotfoot · Sep 20, 2002

Doug is right again about the specs.

And, Doug! Wow - you certainly *DO* know your foods!

Vegemite: Made from vegetable extracts (though I suspect vegemite and marmite really are the same stuff)

Marmite: Made from yeast extracts

Bovril: Made from beef (BOVine)ril - get it?

All have a ton of salt added and -..erm, have a strong character on the palate. Rich in B vitamins, so its good for you. Just like luxeons are.

Doug S · Sep 20, 2002

Originally posted by hotfoot:
Marmite: Made from yeast extracts

Well, I'm relieved, I thought maybe it might be made from ground up Woodchucks

Caution: regional humor, may not play well internationally.

hotfoot · Sep 20, 2002

Originally posted by Doug S:
<blockquote>quote:<hr />Originally posted by hotfoot:
Marmite: Made from yeast extracts

Well, I'm relieved, I thought maybe it might be made from ground up Woodchucks

Caution: regional humor, may not play well internationally.<hr /></blockquote>Well, yeah - you got me there. I'm a little stumped. But I'm sure that brought the house down in your neck of the woods. Your very *tiny* eetsy bitsy neck of the woods

Luxeon flashlight heat temperature calculations: DIY formulas.

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Flashlight Enthusiast

Similar threads