How to determine thermal resistance from junction to ambient (for transistor) ?

Thread Starter

nqchanh194

Joined Jul 5, 2013
27
I've read datasheet of some NPN transistors. In some cases, datasheet barely gives me the value of thermal resistance from junction to case, but without the one from junction to ambient. So how I can determine the thermal resistance from junction to ambient.
Here let's take SA1943 as an example.
 

Attachments

Thread Starter

nqchanh194

Joined Jul 5, 2013
27
I think these two following articles might be helpful.
1.Thermal resistance
2.List of thermal conductivities
In some case, datasheets only mention thermal resistance from junction to case, and I want to determine thermal resistance from junction to ambient, so how can I do ?
Please take a look at the equation at the end of the first article. And my question is how we determine x, the length of the material (measured on a path parallel to the heat flow) (m), in case we don't use heatsink?
 

ronv

Joined Nov 12, 2008
3,770
It will be a science project to get it very close, but to get pretty close you can use:

Junction to case plus 50 / square root of the exposed surface area in cm

For some reason it is never quite this good but you can try it with a few packages to get a rule of thumb for yourself.

PS.
Works for heat sinks too.
 

crutschow

Joined Mar 14, 2008
34,470
...............................
In some case, datasheets only mention thermal resistance from junction to case, and I want to determine thermal resistance from junction to ambient, so how can I do ?
................................
Thermal resistances simply add.
So you take the junction-case thermal resistance and, if it's free air, you add the case-ambient value given in my reference in post#3.
If it's case to heat-sink then you add junction-case resistance to the thermal resistance from case to heat-sink (quite small if it's properly mounted with a thin layer of thermal grease) and then add the heat-sink to air thermal resistance.
 

ronv

Joined Nov 12, 2008
3,770
Crutschow, Am I reading something wrong in your chart? It shows TO220 as 35C/W. I thought they were almost always 62C/W.
 

Thread Starter

nqchanh194

Joined Jul 5, 2013
27
Crutschow, Am I reading something wrong in your chart? It shows TO220 as 35C/W. I thought they were almost always 62C/W.
So do I! It seems Mr.Crutchow's reference belongs to Linear Tech.Coporation, so it might be not true for other ICs from other manufacturer!
Please take a look at my attachments ! Both are the TO 220 type !
 

Attachments

Top