Heatsink calculation

[johnheath]

Hi everybody.

I have a question on how to know how large heatsink I need for a certain voltage regulator. The one I think about is the 78S12 which in my case have a dissipation of 5W.

I want it as small as possible but still useful enough.

Thanks

/John

 

[Rochey]

It's all about your Theta-JA. How hot does your chip get for every W of dissipation.
You improve you Theta-JA by adding the heatsink.

http://www.ti.com/lit/an/slva462/slva462.pdf

http://www.aavid.com/sites/default/files/technical/papers/how-to-select-heatsink.pdf

 

[johnheath]

Wow… thanks =)

A lot of calculation ahead.

Out of experience what would you say would work?….size and material and perhaps mounting?

/John

 

[JohnRoberts]

The math is relatively simple which is the good news, the bad news is some terms in that full equation, mostly on the ambient end, are difficult to discern.

It is similar to voltage drops and ohms law, but instead of current flowing in a path creating voltage drops due to electrical resistance, the power flowing creates temperature drops (or rise) due to thermal resistance.

Generally we design for a maximum 150'C for junction temp, but look at the data sheet to confirm. Most modern Vregs have built in thermal shut down at elevated temperature. 

A relatively easy input term is how much power the regulator must dissipate. Vin (unregulated) - Vout (regulated) x current draw will give you the Watts to use in heat rise calculations.

From a similar data sheet the to-220 part specs 3'C/W junction to case and 50'C junction to ambient so its apparent that it doesn't take more than a couple watts (times 50' each watt) in free air to consume the entire thermal budget. While the 3'C/watt from junction to case is insignificant compared to thermal resistance case to ambient.

While all this math is fun I have generally designed regulator heat sinks based on empirical tests (trial and error). An old school test to determine if the case is hotter than 100'C is to dab a little spittle on it. If it boils off it is hotter than 100'C, It the spittle doesn't boil you are cooler than 100'C and probably OK.

So #1 figure out how many watts, then if that times 50'C/W is more than 100' you need to look at adding some more heat sink. Even if you can get by without adding a heatsink running the regulators hot is not good for long term reliability. I've even seen 3 terminal regulators run so hot that over time they degraded their solder connections. 

JR 

Note: For higher power heat sink design like for audio amps where there could be several tens of watts, the resistance between case and heat sink are more important.

 

[PRR]

> I want it as small as possible

Always a mistake. Be generous.

> dissipation of 5W.

What temperature rise?

I like a conservative 50 deg C rise above a 25C ambient.

So total thermal resistance must be less than 50C/5W or 10 deg C per Watt.

Thermal resistance of 78L05 junction-case is 5deg/Watt.

So the sink must be less than 5deg/Watt.

As a rule-o-thumb, one square inch is 100 deg/Watt.

So you want 100/5= 20 square inches.

Both sides. So 10 sq.in. per side. 3.2" square plate with ample air around it. (Don't cramp the air!!)

The chip is rated for 150 deg C junction. 125 deg C above a 25 deg C ambient. But if you play in summer sun you can have 122 deg Fahrenheit or 50 deg C ambient, so only 100 deg safe rise. 100C/5W= 20C/Watt. Minus chip's 5C/W is 15 deg C per Watt in the sink. This gives 100/15 or 6.7 sq.in. total, 1.9 inches square. This will run HOT, the air will be boiling off so you still need ample air-space. The radiated heat will cook nearby parts.

You also need ample sink thickness to carry heat from the center to the edge, or the size does not help. Rule-o-thumb suggests 1/8" aluminum. A sharp mind can reduce this, though the mental strain is only justified for large production.

 

[ricardo]

I want it as small as possible but still useful enough.

Then you better do the calculations and allow for all the worst case stuff. 

Don't forget 'ambient' is hotter inside the case than outside.

If its DIY, it pays to use as big as you can squeeze in ...  above the calculated heatsink.

If its commercial, it pays to use as big as you can squeeze in ... above the calculated heatsink too

...  unless you're not bothered by Service Returns.  8)

 

[CJ]

try a piece of Al and touch it under full load,

if it is too hot, make it bigger,

 

[njm]

http://www.eevblog.com/2010/08/15/eevblog-105-electronics-thermal-heatsink-design-tutorial/

 

[johnheath]

Thank you all for very enlightening answers… I see that I play in the minor league here =)

A couple of things… I don't know how hot it gets since the project is in the planning and buying state aright now. And my expression "as small as possible" should be understood as just that. I understand that it could be advisable to use a bid as possible but hey… I want to keep it slim but useful.

PRR … I thank you for the "described" size. Looking at our swedish "Mouser" website I understand that there are numerous variations on design on heatsinks. When talking about square inches and so… I guess that you mean the total surface exposed to air?

Thanks

/John

 

[JohnRoberts]

Thank you all for very enlightening answers… I see that I play in the minor league here =)

A couple of things… I don't know how hot it gets since the project is in the planning and buying state aright now. And my expression "as small as possible" should be understood as just that. I understand that it could be advisable to use a bid as possible but hey… I want to keep it slim but useful.

It might be safe to leave room for a bigger heat sink just in case. You can always use less heat sink, but after a design is finished it gets awkward to add more. If you lay the parts down on the PCB and bolt the heat sink to a PCB land, you will get some extra cooling from that, while heating up the PCB. For hot regulators inside a metal chassis, bolting the regulator to the metal chassis will move a lot of heat while you will probably need to use insulated hardware.

PRR … I thank you for the "described" size. Looking at our swedish "Mouser" website I understand that there are numerous variations on design on heatsinks. When talking about square inches and so… I guess that you mean the total surface exposed to air?

Thanks

/John

Yes, total surface area... If you look at serious larger heat sinks extrusions the surface will usually not be smooth, except for the device mating area where smooth reduces metal-to-metal coupling resistance. Where the heat sink is coupling to the ambient air there will often be a surface texture (like small grooves or ridges) that can increase the effective surface area significantly. Small device clip on heat sinks are often smooth and relatively simple because they just need to be better than the tiny device heat sink pad. A secondary consideration is air flow, while inside a closed chassis without a fan, convection will be modest.

JR

 

[audiomixer]

The ballpark math would go something like this:

45 deg C max ambient inside case
85 deg C max allowable heatsink temp
delta deg C - 85 - 45 = 40 deg C

40deg C / 5W = 8 degC/W something like a Fischer SK409 heatsink would fit the bill, and you can adjust the final height size to optimize your design to the real requirements (touch under use to check ok / not ok status...)  so now you have an Ballpark number to assign space and cost to your design. 5W on a 78S12 is a 5V drop at 1A - reasonable if you don't need to ship all over the world and you don't have massive mains voltage variations to take into account.  8 degC/W is something you could get rid of thru the case, maybe...

 

[Ethan]

If you have control over the layout of the circuit, mounting it to the in-side of the aluminum case is an easy (and cheap!) way to use a big heatsink without adding a "heatsink".

 

[johnheath]

Thanks again.

I do have control over the layout since it is a p2p project =) Chassi mount is a possibility of course.

Also thank you "audio mixer"…it is very helpful to see something that would do and now I can get a bigger to be "safe".

/John

 

[CJ]

if you need to insulate the parts from the chassis, then maybe save some hassle by using these>

 

[PRR]

> "Mouser" website ... talking about square inches and so...

Factory-made heatsinks will usually have a thermal resistance listed. (Maybe not on the little sinks, but you want something not-little.)

When reading the advertised resistance, remember that they pick the best number they can justify. Not only wide-open in the middle of the room, but often at VERY high temperature (where both convection and radiation pick up a bit).

Also disregard forced-air ratings (unless you plan a fan and have a clue what the air speed will be).
___________________________________________

> after a design is finished it gets awkward to add more

Yes. I did a class-A headphone amp. I thought a scrap of aluminum under the aluminum cover would do. It did, but it ran hot enough to burn a finger. As this would sit with me on live recordings, with careless fingers and stray scores, that did not seem wise. I ended up with external finned sink and doubled the overall size of the amp.

My next A-amp was much less A and heat issues more carefully figured.

Sometimes the best thing to do is to assume you will do it again. When Benz and Ford built their first motor-buggies, they probably didn't assume they were finished forever. We know Ford worked his way through the alphabet to "T" before he got a design worth keeping in production a while.

 

[johnheath]

Yes... I ordered BIG heatsinks... a couple in fact to test my upcoming regulator conditions.

CJ - I also ordered some variations on insulation material and constructions for the same reason.

Thanks

/John

 

[JohnRoberts]

> "Mouser" website ... talking about square inches and so...

Factory-made heatsinks will usually have a thermal resistance listed. (Maybe not on the little sinks, but you want something not-little.)

When reading the advertised resistance, remember that they pick the best number they can justify. Not only wide-open in the middle of the room, but often at VERY high temperature (where both convection and radiation pick up a bit).

+1 be conservative and there is nothing wrong with extra heat sink.

Also disregard forced-air ratings (unless you plan a fan and have a clue what the air speed will be).

One of my patents is related to forced air cooling in audio power amps, yes forced air is a different ball game and even a little moving air changes heat flow dramatically. Supporting some natural convection with strategically placed vents can be helpful. Radiation for cooling is generally insignificant at the temperatures we work at.

___________________________________________

> after a design is finished it gets awkward to add more

Yes. I did a class-A headphone amp. I thought a scrap of aluminum under the aluminum cover would do. It did, but it ran hot enough to burn a finger. As this would sit with me on live recordings, with careless fingers and stray scores, that did not seem wise. I ended up with external finned sink and doubled the overall size of the amp.

Ouch... there are safety agency regulations the limit the maximum temperature that any part of the product chassis that customers can get a finger on can be. I do not recall the exact temp but probably 100'F or less.

My next A-amp was much less A and heat issues more carefully figured.

Sometimes the best thing to do is to assume you will do it again. When Benz and Ford built their first motor-buggies, they probably didn't assume they were finished forever. We know Ford worked his way through the alphabet to "T" before he got a design worth keeping in production a while.

For DIY projects, over design of things like heat sinks might save you from that extra cut.

JR

 

[ricardo]

Ouch... there are safety agency regulations the limit the maximum temperature that any part of the product chassis that customers can get a finger on can be. I do not recall the exact temp but probably 100'F or less.

This led to most domestic amps in the 70s - 90s having internal H/Ss with holes in a the case.

Alas, that made good thermal design more difficult/expensive.  :-\

FOR A GIVEN H/S SIZE, its heat dissipation to ambient is improved the hotter it is.  We want the H/S temperature to be as close to the hot chip temperature as possible.

But you don't want the hot H/S inside the case cos 'ambient' there is hotter than outside.

The same goes for resistors.  Modern 2W resistors can be small ONLY cos they can be stinking hot compared to similar rated older 2W items.

For DIY projects, over design of things like heat sinks might save you from that extra cut.

8)

JR, what the number/title of your forced air patent?

 

[JohnRoberts]

> "Mouser" website ... talking about square inches and so...

Factory-made heatsinks will usually have a thermal resistance listed. (Maybe not on the little sinks, but you want something not-little.)

When reading the advertised resistance, remember that they pick the best number they can justify. Not only wide-open in the middle of the room, but often at VERY high temperature (where both convection and radiation pick up a bit).

+1 be conservative and there is nothing wrong with extra heat sink.

Also disregard forced-air ratings (unless you plan a fan and have a clue what the air speed will be).

One of my patents is related to forced air cooling in audio power amps, yes forced air is a different ball game and even a little moving air changes heat flow dramatically. Supporting some natural convection with strategically placed vents can be helpful. Radiation for cooling is generally insignificant at the temperatures we work at.

___________________________________________

> after a design is finished it gets awkward to add more

Yes. I did a class-A headphone amp. I thought a scrap of aluminum under the aluminum cover would do. It did, but it ran hot enough to burn a finger. As this would sit with me on live recordings, with careless fingers and stray scores, that did not seem wise. I ended up with external finned sink and doubled the overall size of the amp.

Ouch... there are safety agency regulations the limit the maximum temperature that any part of the product chassis that customers can get a finger on can be. I do not recall the exact temp but probably 100'F or less.

My next A-amp was much less A and heat issues more carefully figured.

Sometimes the best thing to do is to assume you will do it again. When Benz and Ford built their first motor-buggies, they probably didn't assume they were finished forever. We know Ford worked his way through the alphabet to "T" before he got a design worth keeping in production a while.

For DIY projects, over design of things like heat sinks might save you from that extra cut.

JR

Sorry to bump this old thread but while cleaning my house I found this old graphic related to my heat sink patent (I was co-inventor with two other guys).  I am not responsible for the silly name.

JR

 

[moamps]

Way tunnels do not end at front plate? Or I don't see it correctly.  This way it looks to me  as unnecessary increase of  hot air turbulence in the chassis.
Btw, I like more that the air flow direction is from front to rear of the amp; you can easily clean the filters ( old Crowns ).  Also there is no hot air blowing in your face.  I wander is there any standard for the air direction?