Electronic – How to calculate\estimate thermocouple junction of a solder junction between a PCB and a component


Thermal junctions between two dissimilar metals on a PCB can create voltages. How do I estimate or bound (by estimating a max voltage via a max temperature) the thermocouple effect on a pcb?

Most solders don't list a number for the Seebeck coefficient. AOE does list Sn-Pb solder as 5uV/C but not lead free (Sn 97%).

How can I calculate the thermocouple effect for lead vs lead free solders? Is there a way I can minimize this effect?

enter image description here

Best Answer

SnPb 63/37 has a Seebeck coefficient of about -1.5 (relative to platinum). Copper has a Seebeck coefficient of +6.5, so you get +5 for a solder-copper junction.

Tin-based lead free solders are in the range -0.9 to -1.2 at room temperature so the net thermoelectric effect is not much different from that of leaded solder (slightly worse).

In order to avoid having thermoelectric effects affect your sensitive low-frequency measurements, you should first avoid temperature gradients as much as possible. Keep significant heat sources away from the sensitive bits. And if you must have gradients, try to keep things symmetrical so that the thermoelectric voltages cancel out.

For example, if the device in your image is a sense resistor which has a low voltage across it, but it gets hot, so there is a gradient, you can keep the copper patterns on each pin of the resistor of similar area so the resistor is the same temperature at both ends, thus canceling out the thermoelectric voltages. It's only temperature differences that matter- not absolute temperature (otherwise the laws of thermodynamics would be violated since you would be able to generate power with no heat flow).

Sometimes judiciously placed slots milled in the PCB or breaks in ground planes can be used to create islands that are more isothermal than the rest of the board, as in this article. The part shown is internally ovenized so it runs quite hot.

enter image description here

There are special low-thermal EMF solders (meaning that their Seebeck coefficient is complementary to that of copper) but they tend to contain Cadmium.

If the layout is reasonable, these effect don't tend to show up until you get into the region below 10uV. You can definitely seen an increase in low frequency noise in many measurements as a result of air currents around PCBs if you don't take care.