A little lesson on heatsinks

[EpicHam]

Thats what Ive done with most of my lasers and it works fantastically. Due to the low thermal conductivity of most of the thermal greases and epoxies however, it probably still isnt as good as a direct press heat sink (ie without module) of Al or better still Cu. Wrt tarnishing if you are only talking about the module its not really an issue; your sweat doesnt really come into contact with it and it is hidden behind the focus adapter.

I do clean the modules before I use them though, in a salt and lemon juice mix prior to use. :beer:

Ahhh salt and lemon juice .
Simply classic and MUCH cheaper. Better than my suggestion.

 

[BShanahan14rulz]

I don't doubt that. I'm going to bookmark this thread for when that time comes.

There's always someone who says that in all of these threads, and it never seems to happen ;-)

And again, we ARE talking about dinky little laser diodes... I like Cy's comment about how 1Ga. wire is much much better than 26Ga.

And last, sure, everybody wants copper heatsink, and wants to pair a copper module with it, and put it in a copper host. I would rather have a well-fitting, tarnish-resistant heatsink that fits snug in my host, won't grow with age due to scum on the surface, and can accomodate a bare diode interference-fitted to it.

When's the last time you left your handheld laser on for 30 minutes at a time, anyways?

 

[EpicHam]

There's always someone who says that in all of these threads, and it never seems to happen ;-)

And again, we ARE talking about dinky little laser diodes... I like Cy's comment about how 1Ga. wire is much much better than 26Ga.

And last, sure, everybody wants copper heatsink, and wants to pair a copper module with it, and put it in a copper host. I would rather have a well-fitting, tarnish-resistant heatsink that fits snug in my host, won't grow with age due to scum on the surface, and can accomodate a bare diode interference-fitted to it.

When's the last time you left your handheld laser on for 30 minutes at a time, anyways?

You're going to lose service life with aluminum you know.
As the heatsink has a VERY POOR thermal conductivity.
The heat would be trapped in the core for a while before the external layer(host) warms up. By then you could very well have exceeded the operation temperature and is shortening the diode's service life.

With copper, you effectively remedied the problem.
Your choice :beer:


Performance and Long term Cost
Or
Short term Cost and Convenience

 

[Leodahsan]

You're going to lose service life with aluminum you know.
As the heatsink has a VERY POOR thermal conductivity.
The heat would be trapped in the core for a while before the external layer(host) warms up. By then you could very well have exceeded the operation temperature and is shortening the diode's service life.

With copper, you effectively remedied the problem.
Your choice :beer:


Performance and Long term Cost
Or
Short term Cost and Convenience

That thing about the heat trapped in the core isn't a true statement. My aurora SH032 has a copper HS and it goes >60ºc when the out is still barely warm.
My Solarforce L2P host heats evenly with the Al HS.

The host transferring heat is not totally related with the heatsink material.

 

[EpicHam]

That thing about the heat trapped in the core isn't a true statement. My aurora SH032 has a copper HS and it goes >60ºc when the out is still barely warm.
My Solarforce L2P host heats evenly with the Al HS.

The host transferring heat is not totally related with the heatsink material.

Of course we are assuming a host fixed ,heatsink material variable case.
When you changes the host design its only natural that the final subjective sensation will change.

However ,the fact remains. For a same host ,same heatsink design , different heatsink material case.
Copper wins.

 

[TheDukeAnumber1]

You're going to lose service life with aluminum you know. As the heatsink has a VERY POOR thermal conductivity.

I would most definitely not describe it as very poor. I do wonder at times if the service life thing it true though.

 

[EpicHam]

I would most definitely not describe it as very poor. I do wonder at times if the service life thing it true though.

So do I , but we need someone to do a comprehensive scientific research on to that if we were to find it out tho.

Go ask Spooky.
He should have some data . I think.
Don't take my word for it tho.

 

[hwang21]

Ok so in terms of the best solution overall, copper everything, from module to outer casing, is the best option, but it leaves a large maintenance job to do every few years, and costs a bit more?

 

[TheDukeAnumber1]

So do I , but we need someone to do a comprehensive scientific research on to that if we were to find it out tho.

Go ask Spooky.
He should have some data . I think.
Don't take my word for it tho.

I'm fine wondering, feel free to research it though and post your findings.

Ok so in terms of the best solution overall, copper everything, from module to outer casing, is the best option, but it leaves a large maintenance job to do every few years, and costs a bit more?

IMO, a black anodized aluminum host is the best way to go.

 

[grainde]

^^^Ahhh now were talking about the radiative emissivity of surfaces... hmm

@ hwang21 please go read this thread - almost everything has been covered there.:beer:

 

[EpicHam]

Ok so in terms of the best solution overall, copper everything, from module to outer casing, is the best option, but it leaves a large maintenance job to do every few years, and costs a bit more?

Thats pretty much it.
Copper is a bit harder to machine and is classified as a "precious metal".
They only comes in about 17 million tonnes per year.
While aluminum.... wait for it.
OVER 100 MILLION TONS OF BAUXITE!!!

So expect a metal with such an excellent performance to be on the pricey side.
They also can't be anodized without resorting to..... sigh..... aluminium copper alloy.......
If THAT option is provided , you get the BEST of 2 world.
The HARD Aluminum oxide shell for the anti-oxidation protection , and ANY color you want.

Sadly, I've yet to see such a heatsink

^^^Ahhh now were talking about the radiative emissivity of surfaces... hmm

@ hwang21 please go read this thread - almost everything has been covered there.:beer:

Ah well radiative cooling takes a VERY small part when compared to convective cooling which is the main cooling method of heatsinks.
But every bit helps

 

[ApexProxy]

@EpicHam
Aluminum doesn't corrode as fast as copper. I have many aluminum heatsink lasers and all the heatsinks are just as shiny as day one while the copper ones are all heavily tarnished and corroded. Also. I have full unanodized aluminum hosts that are not tarnishing or corroding at all.

@Leodahsan
You have too many variables with your Aurora SH032 vs Solarforce L2P comparison. Its a simple fact that copper transfers heat faster. The reason your copper heatsink may not work as nicely as your aluminum is that it likely wasn't machined to fit as snugly in its host or that the host itself isn't made to dissipate heat as well. Also, steel's thermal properties SUCK EPIC so the host itself is a ridiculous thermal insulator in comparison to the aluminum host of the Solarforce.

I'm bored so... technicality bomb time.

Thermal Diffusivity is a term we can use which is thermal conductivity divided by density and specific heat capacity at constant pressure. This gives us how far heat can travel down a 1mm by 1mm bar per second in mm²/s. This allows us to see how fast each material transfers heat in a single raw number. Here are our favorite candidates:
Aluminum - 84.18
Copper - 111

Now, here is Volumetric Heat Capacity(J·cm^−3·K^−1):
Aluminum - 2.422
Copper - 3.45

That means copper stores 29.8% more heat energy for the same temperature. Copper is also 24.16% faster at transferring heat based on Thermal Diffusivity.

This means that although Copper takes 37% longer to heat up and once at the same temperature even though it it stored 29.8% more heat it fully cooled down while the Aluminum still retained 1.35% of its heat energy.

Since I know people will not believe me here are the calculations:
Thermal Diffusivity:
84.18/111*100-100 = 24.16%

Volumetric Heat Capacity:
2.422/3.45*100-100 = 29.8%

This next one warrants an explanation because its much more complicated. This calculation shows how much longer copper takes to reach the same temperature taking into account that both it and the aluminum are gaining temperature(the laser is on) but at the same time they are also transferring some away into the air and host simultaneously. Copper is losing heat 24.16% faster than aluminum while requiring 29.8% more heat to reach the same temperature which means it loses 7.2% of its heat while increasing the 29.8% more energy required to reach the same temperature. The calculation:
29.8+24.16% = 37%

The next part is after knowing the copper now has 129.8% of the heat energy that aluminum has at the same temperature you now take into account that it will lose all its heat energy 24% faster. This means it will lose all its heat energy while Aluminum will still have 1.35% of its heat energy remaining.
129.8-24%-100 = 1.35%

In total, if you turned on two identical lasers, one with an Aluminum heatsink and one with a Copper heatsink, at the same exact time and turned them both off when they reached a set temperature the copper would have taken 37% longer to do so. Then, assuming you had the two lasers with their heatsinks at the same exact temperature the copper heatsink would cool down 1.35% quicker even tough it contained 29.8% more heat energy. Therefore, copper takes longer to warm up and then cools off in a tiny bit less time.

To me the gain of 37% more ON time and reducing OFF time by 1.35% is well worth it when I am limited on heatsink size.

 

[grainde]

Hmm interesting but you havent accounted for heat loss. You have used the thermal diffusivity ie the ability of the material to transfer heat away from a point source within itself as the means for calculating heat loss?

The heat sinks are however finite and the heat has no where to go within the material at thermal equilibrium. Your calculations should include surface area, conduction, convection, temperature differential whilst cooling and radiation to the environment.

You should also factor in, if comparing two different materials as heat sinks in lasers, that the interfaces between the module and the heat sink and heat sink and host are far from ideal and actually have an insulating effect. Most thermal epoxies and greases are around 3 Wm-1 k-1 whereas your Aluminium and Copper are 200 and 300 W m-1K-1. Not to mention if your grease wasn't perfectly applied and you have air gaps...:thinking:

Edit: From personal experience copper takes a lot longer than Al to cool down. I guess a crude test would be to heat 2 C6 heat sinks (even 3 if you wanted to compare the black anodised version..) one std Al and one Cu. Heat them to the same temp, remove them from the heat source and place them on an insulator perhaps a ceramic plate, and use temp probes (or an FLIR) to plot their cooling rates to ambient.

 

[ApexProxy]

Hmm interesting but you havent accounted for heat loss. You have used the thermal diffusivity ie the ability of the material to transfer heat away from a point source within itself as the means for calculating heat loss?

The heat sinks are however finite and the heat has no where to go within the material at thermal equilibrium. Your calculations should include surface area, conduction, convection, temperature differential whilst cooling and radiation to the environment. :thinking:

Edit: From personal experience copper takes a lot longer than Al to cool down. I guess a crude test would be to heat 2 C6 heatsinks (even 3 if you wanted to compare the black anodised version..) one std Al and one Cu. Heat them to the same temp making sure they are both isolated at the base they are resting on and use temp probes to plot their cooling rate to ambient.

Heat loss is not something you can calculate without extensive knowledge of the exact host used and the exact design of the heatsink for the host. The dissipation limits would be because of how efficient the heatsink transfers heat into the host as well as how efficiently the host dissipates the heat. The heatsinks we use are rarely designed with any real heat dissipation capabilities themselves. They simply store and move the heat. This nullifies any applicability of tests on how well each heatsink dissipates heat into the air.

In practice you may notice that Copper takes longer to cool down and that would be because it stored more heat and your host isn't capable of dissipating the extra energy quickly. Most of the time you are relying on the efficiencies of the aluminum of the host(including any powdercoat or anodizing) to dissipate the heat. If your ON cycles were kept the same for aluminum and copper, the copper OFF cycle would be faster because it is better at moving heat.

Unfortunately even testing this involves insane amounts of variables including diode efficiencies, host tolerances and also heatsink tolerances. So its not really testable without using higher grade materials and equipment than most of us have access to.

 

[grainde]

Heat loss is not something you can calculate without extensive knowledge of the exact host used and the exact design of the heatsink for the host. The dissipation limits would be because of how efficient the heatsink transfers heat into the host as well as how efficiently the host dissipates the heat. The heatsinks we use are rarely designed with any real heat dissipation capabilities themselves. They simply store and move the heat. This nullifies any applicability of tests on how well each heatsink dissipates heat into the air.

Exactly, but your calculation doesnt take any of that into account. It relies solely on a value of thermal diffusivity which tells you nothing about how a small fixed volume heat sink cools. It might be valid for a length of copper rod a few meters long, which for all intents and purposes you could assume to be infinite as the cooling over the entire length of the rod (and hence the heat dissipation to the environment) means that the other end would be at ambient.

In practice you may notice that Copper takes longer to cool down and that would be because it stored more heat and your host isn't capable of dissipating the extra energy quickly. If your ON cycles were kept the same for cluminum and copper, the copper OFF cycle would be faster because it is better at moving heat.

I realise copper stores more energy to reach the same temp and that would mean a longer on time, but Im not sure how much better copper is at getting rid of that heat to the surroundings and your calc above doesnt explain that. If you use copper as a heat sink you will probably get your 30 % longer on time (assuming heat transfer through all interfaces is ideal), but you will also have a respectively longer off time to get rid of that extra heat.
If you keep the same run times; the copper will be cooler but it will depend on how well copper performs against aluminium at giving up that heat to the surroundings. Copper probably is better, but I dont think youve actually proven it. Not yet anyway.. :beer:

Edit: Thats why I suggested testing 2 C6 heat sinks ie same volume (out of the host) and plotting how effectively they cool. You could even do a second test in the host, but you would have many more variables as I mentioned earlier.

 

[ApexProxy]

You seem to be missing that the "heatsink" doesn't do any dissipation. Its only a storage and movement conduit for the heat. All dissipation is done by the host. We use the term heatsink but in standard terms they don't actually do any by definition heatsinking. They should be called heat transferrers or heat storers or some strange combination of the two. Maybe called thermal buffers?

Granted this is very different for a host with a massive extended heatsink sticking way out of the host but who in their right mind would use copper for that anyway. It would look like crap in only a couple days of pocket time.