help with voltage regulator heatsink calculation

[podoljanko]

hi, im trying to find out if i need a heatsink, and if so how big, for the following component: http://www1.elfa.se/data1/wwwroot/assets/datasheets/07306781.pdf

In figure 6, on page 11 it states that i have to take the ground current into consideration, however i have no idea what that current may be.

I am interested in reducing 9 volts to 5V, running at 0.8A, the component is in a SOT-223 package.

Hope you guys can help:=)

 

[ronv]

You can't quite get there with that package. You can pretty much ingnor the ground current becaue it is only 120 ua. But to drop 4 volts at .8 amps is 3.2 watts. With no heatsink the temperature rise is 136C/Watt which is way to hot. Even with 1 inch of copper area on the PCB for a heatsink it is still 66C/Watt or 211 C. Still way to much. If you really need the full .8 amps you probably want the TO220 package and a heat sink of 10 C / Watt.

 

[bountyhunter]

You can pretty much ingnor the ground current becaue it is only 120 ua. But to drop 4 volts at .8 amps is 3.2 watts. With no heatsink the temperature rise is 136C/Watt which is way to hot. Even with 1 inch of copper area on the PCB for a heatsink it is still 66C/Watt or 211 C. Still way to much. If you really need the full .8 amps you probably want the TO220 package and a heat sink of 10 C / Watt.

Actually, the ground current spec is max of about 10 mA. But I agree, you need the TO-220 for power ov more than 3W,

If you solder down a TO-220 with about one square inch of copper (or more), you should get a theta of around 30 -35 C/W which should be OK.

 

[podoljanko]

Thank you very much for your reply! so the ground current, is the adjustment current... makes sense..
But i do not seem to understand how you calculated those values..
as i understand the power dissipation Pd is approximately 3.2W, the Tj(max) is 125C, and I set the Ta(max) to 50C. thereby Tr(max) = 75C

Øja = Tr(max)/Pd
= 75C / 3.2W
= 23,4 C/W

this value is lower than the 136 C/W than the max specified for the SOT-223 package..
Am I doing something wrong?

 

[bountyhunter]

Thank you very much for your reply! so the ground current, is the adjustment current... makes sense..
But i do not seem to understand how you calculated those values..
as i understand the power dissipation Pd is approximately 3.2W, the Tj(max) is 125C, and I set the Ta(max) to 50C. thereby Tr(max) = 75C

Øja = Tr(max)/Pd
= 75C / 3.2W
= 23,4 C/W

this value is lower than the 136 C/W than the max specified for the SOT-223 package..
Am I doing something wrong?

Yes. You are trying to use a SOT-223 in an application where it can not possibly handle the power. You must use either the TO-220 or the TO-263 (which is a TO-220 with the tab snipped off). Solder either of these to the PCB with at least 1 square inch of copper and it will work. Use a SOT-223, and it can not possibly work.

 

[ronv]

The goal is to keep the temperature well below 125C at the junction. With your set up the temperature would be 3.2 X 136 or 435C plus lets say 25C ambient or 460C. To hot. Table 1 shows the temperature rise with 1 sq. in of copper on the same package. 66C/watt. 66X3.2 = 211C + 25 for ambient = 236C. Still to hot.
If you use the TO-220 package you can use a better heatsink that the copper on the board. 10 degree C per watt heatsinks are fairly small. So in that case you get 3.2X10 or 32C + 25C ambient for about 57C. It will be a bit hotter but well below the 125C junction temperature.