# Interpreting datasheet thermal information

#### AlbertHall

Joined Jun 4, 2014
10,226
Below is an extract from an SMD diode datasheet. I want to be able to calculate temperature using the copper on a PCB as a heatsink. It gives a value for thermal resistance from junction to lead of 25C/W. If I put an area of copper for each lead can I then use half this value (as there are two leads) for the calculation?

#### Jony130

Joined Feb 17, 2009
5,180
I think that the value given in the datasheet already includes two pad areas. So, they already gave you the "total" Rth_JL resistance for given PCB pad areas.

#### BobaMosfet

Joined Jul 1, 2009
1,175
That value is per junction. Whatever leads are attached to that junction will thermal at a rate based on power through the junction, minus dissipation and compensate for derating. Art of Electronics has some great information for calculating the thermal curve, and if you are sensitive to heat, you should calculate the curve to determine how much power you can actually run through the component. The sweet spot will not be anywhere near the maximum.

I find Excel is great for graphing load-line and power-dissipation curves:

Ideally, the load-line should be to the left of the dissipation curve, if it isn't, you have a thermal junction overload.

Last edited:

#### AlbertHall

Joined Jun 4, 2014
10,226
Thanks guys. That means I will need a heatsink (or two!) not just the copper on the PCB.

#### crutschow

Joined Mar 14, 2008
25,247
Here are some references for determining PCB copper thermal resistance.

#### Hymie

Joined Mar 30, 2018
846
Semiconductor device manufacturers invariably devise performance figures that flatter.

Consider the frequency value ‘fT’ quoted for transistors – many might believe that this is the upper useful frequency at which the transistor can switch. But it is the typical frequency at which the transistor gain is unity (in other words the device is providing no gain).

The data sheet value of 25⁰C/W is almost certainly based on the leads having a theoretical infinite heatsink. You would be well advised to use the 550mW* figure together with 226⁰C/W and check that with your projected power dissipation the device does not turn to toast.
(* even this value is based on a 25⁰C ambient)

My understanding of the data sheet 3W power dissipation figure is applicable where the device body is maintained at 75⁰C by some magic.