Thermal conductivity of solder

21 posts / 0 new
Last post
Bort
Bort's picture
Offline
Last seen: 1 week 5 days ago
Joined: 06/01/2012 - 17:15
Posts: 7460
Location: Holding the proverbial flashlight
Thermal conductivity of solder

After reading this
http://budgetlightforum.com/node/17466#comment-304439
I am curious if there is higher heat conductivity solder out there, has anyone ever thought to test different formulations, is method of application a determinant of heat transfer?

The Journal of Alternative Facts TM

"It is critical that there is a credible academic source for the growing and important discipline of alternative facts. This field of study will just keep winning, and we knew that all the best people would want to be on board. There is a real risk in the world today that people might be getting their information about science from actual scientists"

 

NightCrawl
Offline
Last seen: 4 years 8 months ago
Joined: 01/22/2012 - 08:20
Posts: 3071
Location: Karlsruhe, Germany

The thinner, the better heat transfer. And there are better conductors out there than silver. For example liquid metal pads which are sometimes used on CPUs. Those are around 80W/mK while solder is about 40W/mK. Still an order of magnitude better than the dielectric on regular stars or any thermal glue/paste. Those range from <1W/mK (Fujik) and range to (I think) 8W/mK on some Arctic products.

saabluster over at CPF is using some exotic and extremely good glue: http://www.icproto.com/typeroom/assets/uploads/pdf/DIEMAT%20DM6030.pdf

60W/mK, that should be better than solder but worse than liquid metal. For him its better to glue the stuff because he sands the emitter base down.. seems to work great, because the emitter consists of some kind of ceramic with poor thermal conductivity.

 

I'm still waiting for a flashlight with heatpipes.

PilotPTK
Offline
Last seen: 6 years 10 months ago
Joined: 09/04/2011 - 21:53
Posts: 1736
Location: Michigan, USA

100% Sn (Tin) solder has, if I remember, about twice the thermal conductivity as 63Sn37Pb (Standard eutectic tin/lead). It’s a bit of a bitch to work with though..

PPtk

Edit:
Originally typed Sn/Ag – completely wrong on that account. Also wrong on the difference.. Just wrong, wrong and more wrong tonight Smile
Correct Info is:
Not quite twice.. 73W/M*K compared to 50W/M*K

Here’s a nice char done by Indium showing the thermal conductivity of many of their alloys. Note that it’s in W/CM*K, so multiply by 100 to get W/M*K
http://documents.indium.com/qdynamo/download.php?docid=453

I am currently extremely busy with work. Please do not expect a response from me quickly. I will be dropping in as time permits, but the amount of time I can dedicate to responding to topics and PMs is very limited.

NightCrawl
Offline
Last seen: 4 years 8 months ago
Joined: 01/22/2012 - 08:20
Posts: 3071
Location: Karlsruhe, Germany

According to this:

http://www.electronics-cooling.com/2006/08/thermal-conductivity-of-solders/

Its 78 vs. 50.. not twice, but a lot more. Why is it hard to work with? Because of the high melting temperature or is there more to it?

PilotPTK
Offline
Last seen: 6 years 10 months ago
Joined: 09/04/2011 - 21:53
Posts: 1736
Location: Michigan, USA
NightCrawl wrote:

According to this:

http://www.electronics-cooling.com/2006/08/thermal-conductivity-of-solders/

Its 78 vs. 50.. not twice, but a lot more. Why is it hard to work with? Because of the high melting temperatuer or is there more to it?

I was wrong about SnAg – It’s actually far worse. I trust Indium FAR more than I true electronics-cooling.com

Sn or SnAg is a pain to work with because of the high melting point and because it doesn’t wet the same as Tin/Lead

I am currently extremely busy with work. Please do not expect a response from me quickly. I will be dropping in as time permits, but the amount of time I can dedicate to responding to topics and PMs is very limited.

f42
Offline
Last seen: 4 years 5 months ago
Joined: 08/14/2012 - 21:15
Posts: 234


Nationalism is an infantile disease. It is the measles of mankind.

Albert Einstein

NightCrawl
Offline
Last seen: 4 years 8 months ago
Joined: 01/22/2012 - 08:20
Posts: 3071
Location: Karlsruhe, Germany

Nice document PPTK, thanks. Pure gold would be great..

What is #4 in that list? 86W/mK is also nice. I've got some tin rods meant to make tin soldiers.. if those are pure, I could use them. But it does need a lot of heat until it melts. I think I'll stick to regular solder, I'll just keep improving the PCBs. Wink

PilotPTK
Offline
Last seen: 6 years 10 months ago
Joined: 09/04/2011 - 21:53
Posts: 1736
Location: Michigan, USA
  1. is 100% Indium. It’s a rare metal, and it’s not cheap.

I am currently extremely busy with work. Please do not expect a response from me quickly. I will be dropping in as time permits, but the amount of time I can dedicate to responding to topics and PMs is very limited.

Manual Man
Offline
Last seen: 2 years 11 months ago
Joined: 08/29/2011 - 01:02
Posts: 411
Location: New Zealand
NightCrawl wrote:

According to this:

http://www.electronics-cooling.com/2006/08/thermal-conductivity-of-solders/

That is where I got the figures in my post linked to above… there are quite a few better then 50W/m.k however they also have higher melting points and I am not sure how good it is for the LED to be pushed up to >220°C or so even if only for a short period.

That copper solder does look promising, I don’t actually have a clue about the type of solder I use… apart from that it melts at around 160°C, I’m going to have to find myself some 63/37 stuff.

My guess is that with good thermal conduction away from the LED it will be the LED itself which becomes the limiting link / bottleneck.

NightCrawl
Offline
Last seen: 4 years 8 months ago
Joined: 01/22/2012 - 08:20
Posts: 3071
Location: Karlsruhe, Germany

Yes. Once you solder the LED directly to copper, the bottleneck is the ceramic under the die. Thats why saabluster grinds it off. But honestly.. once the dielectric is eliminated, you are already on a really good way. Its like with anything.. the last few % of performance require the most effort. So lets keep it simple.. Wink

Bort
Bort's picture
Offline
Last seen: 1 week 5 days ago
Joined: 06/01/2012 - 17:15
Posts: 7460
Location: Holding the proverbial flashlight

NightCrawl wrote:

Yes. Once you solder the LED directly to copper, the bottleneck is the ceramic under the die. Thats why saabluster grinds it off. But honestly.. once the dielectric is eliminated, you are already on a really good way. Its like with anything.. the last few % of performance require the most effort. So lets keep it simple.. Wink


simple is good, but if its not too much harder to get better then thats also very nice

The Journal of Alternative Facts TM

"It is critical that there is a credible academic source for the growing and important discipline of alternative facts. This field of study will just keep winning, and we knew that all the best people would want to be on board. There is a real risk in the world today that people might be getting their information about science from actual scientists"

 

PilotPTK
Offline
Last seen: 6 years 10 months ago
Joined: 09/04/2011 - 21:53
Posts: 1736
Location: Michigan, USA

Taking an LED above 220C for a few seconds to solder it is nothing to worry about. Modern LEDs are designed to be reflow soldered using lead-free solder. A typical lead-free solder profile will take the boards as high as 240C-260C for 30 seconds.

PPtk

I am currently extremely busy with work. Please do not expect a response from me quickly. I will be dropping in as time permits, but the amount of time I can dedicate to responding to topics and PMs is very limited.

Old-Lumens
Old-Lumens's picture
Offline
Last seen: 3 years 11 months ago
Joined: 11/04/2011 - 11:39
Posts: 7478
Location: Tyler, TX, USA

What about the copper bearing solder paste and silver bearing solder paste, or the stuff jewlers use, like the copper wire solder? Never mind, way too hot.

Just asking, I don't know sh** from shinola when you get technical on me...

My PayPal address: oldlumens (insert the @ sign here) gmail.com

My YouTube Flashlight Video Channel

The BLF Modding Links Thread 

http://imageshack.com/a/img922/1374/jQ2wdL.jpg

 

NightCrawl
Offline
Last seen: 4 years 8 months ago
Joined: 01/22/2012 - 08:20
Posts: 3071
Location: Karlsruhe, Germany

Bort wrote:
simple is good, but if its not too much harder to get better then thats also very nice

Thats the problem: grinding off the bottom of LEDs is far from being easy, and the stuff he uses to glue them down costs a fortune.

You can of course use better solder, but regular solder is at 50W/mK while the better stuff gets to 73W/mK.. and the latter is hard to work with, according to PPTK. I wouldnt (and I dont) mind, a few days ago everybody still used PCBs with dielectric and was really happy..

Bort
Bort's picture
Offline
Last seen: 1 week 5 days ago
Joined: 06/01/2012 - 17:15
Posts: 7460
Location: Holding the proverbial flashlight

NightCrawl wrote:

Bort wrote:
simple is good, but if its not too much harder to get better then thats also very nice

Thats the problem: grinding off the bottom of LEDs is far from being easy, and the stuff he uses to glue them down costs a fortune.

You can of course use better solder, but regular solder is at 50W/mK while the better stuff gets to 73W/mK.. and the latter is hard to work with, according to PPTK. I wouldnt (and I dont) mind, a few days ago everybody still used PCBs with dielectric and was really happy..


yes, but to push the envelope and be extra happy with our accomplishment of what was not thought possible yesterday we have to be open minded
however i am also practical, i wouldn’t spend $100 on a better conduction XM-L2 to sinkpad, but i would spend $10 more

The Journal of Alternative Facts TM

"It is critical that there is a credible academic source for the growing and important discipline of alternative facts. This field of study will just keep winning, and we knew that all the best people would want to be on board. There is a real risk in the world today that people might be getting their information about science from actual scientists"

 

NightCrawl
Offline
Last seen: 4 years 8 months ago
Joined: 01/22/2012 - 08:20
Posts: 3071
Location: Karlsruhe, Germany

Thats exactly what I was talking about. Silly

Another cheap but imo hard to accomplish approach would be a metal heat pad and then somehow apply permanent pressure on the LED.. that stuff can do 82W/mK. I think the stuff saabluster uses costs 300$ for the MOQ..

NightCrawl
Offline
Last seen: 4 years 8 months ago
Joined: 01/22/2012 - 08:20
Posts: 3071
Location: Karlsruhe, Germany

PilotPTK wrote:
#4 is 100% Indium. It's a rare metal, and it's not cheap.

Indium didnt let me go.

I found it for about 1.25€/g which seems fair. 99.9% pure, melting at 175°C, thermal conductivity of 86W/mK, better than solder. It seems you can solder it to aluminium and even ceramics. That means I could grind off the bottom of LEDs and solder them with indium?

If one could manage to solder directly to aluminium, would that be better than copper?

Rufusbduck
Rufusbduck's picture
Offline
Last seen: 1 year 8 months ago
Joined: 04/04/2012 - 15:34
Posts: 10389
Location: Golden state

Just guessing but I think the first contact should be copper to funnel the heat away quickly to the radiant surface which could be aluminum to reduce weight. I think a copper star on aluminum sink would be better than an aluminum star on a copper sink.

Three Tanna leaves to give him life, nine to give him movement. But what if he eats the whole bag?

Scott

PilotPTK
Offline
Last seen: 6 years 10 months ago
Joined: 09/04/2011 - 21:53
Posts: 1736
Location: Michigan, USA
NightCrawl wrote:

PilotPTK wrote:
#4 is 100% Indium. It’s a rare metal, and it’s not cheap.

Indium didnt let me go.

I found it for about 1.25€/g which seems fair. 99.9% pure, melting at 175°C, thermal conductivity of 86W/mK, better than solder. It seems you can solder it to aluminium and even ceramics. That means I could grind off the bottom of LEDs and solder them with indium?

If one could manage to solder directly to aluminium, would that be better than copper?

That is very reasonable pricing for pure indium.

Yes, you technically can solder indium directly to ceramic.. even glass..

Soldering to aluminum isn’t really that difficult. The problem is that the flux you have to use to cut through the aluminum oxide layer is nasty stuff (acid), and electronics do not get along well with it.

PPtk

I am currently extremely busy with work. Please do not expect a response from me quickly. I will be dropping in as time permits, but the amount of time I can dedicate to responding to topics and PMs is very limited.

NightCrawl
Offline
Last seen: 4 years 8 months ago
Joined: 01/22/2012 - 08:20
Posts: 3071
Location: Karlsruhe, Germany

Does technically also mean that I can do it in home application or are there some issues like with aluminium?

I've read that it works to put some oil or flux on aluminium, sand the layer off and then solder to it? If it really is that easy, wouldnt that be a simple way to make really good LED PCBs?

SeanHatfield
Offline
Last seen: 8 months 2 weeks ago
Joined: 11/22/2012 - 11:56
Posts: 65

Here is a video that shows soldering to aluminium:
http://youtu.be/FmjdXKyDEWY

Seems to be as easy as you say. The creator of the video wrote that its better to put the oil on the aluminium before sanding it down, but he didn’t want
his glass fiber pencil to get into the oil.