Thermal EMF for soldered components

[justanothername]

Interesting how people try to improve thermal conduction in a metal PCB using a pillar. Maybe one can learn something from high power LED designs.

In my case, for the application I'm designing this, there will be blind vias on the uppermost layer anyways. So I simply will make them Type VII (filled and capped). The next layer is a maximum of 100µm below, so I have high hopes for very good thermal homogenity.
You can stitch gnd planes with type VII vias as much as you want, there will never be solder paste drain. Even when the stitching covers a thermal pad.

[justanothername]

Hi!
Here is a follow up.
I tried to replicate the setup from the Vishay video at home. I recently got a HP419a and restored it, see here:
https://www.eevblog.com/forum/other-blog-specific/great-meter!-the-hp-419a-dc-null-voltmeter-(restoration)/msg3588691/#msg3588691
My setup consists of a 419A, a resistor (DUT) on a stripe of prototype PCB, and a 100W face warming lamp shining on the DUT from a 50° angle. The bulb is 8cm air distance to the DUT.
I used several random SMD current sense resistors I found on my desk.





The following steps were done for each measurement:

* mount DUT on the stand with double sided sticky tape
* let settle for 5 minutes
* zero scale
* turn on the lamp for exactly 30 seconds
* record maximum needle position, watch 30 more seconds

Firstly the used wires and pcb material:
I cut several stripes from hasl tinned prototype material and put a solder blob as short.
For the solder wire, I used ISO-Core Clear Sn100Ni+, as much tin as possible. I tried the following shorts:

1) Tinned wire - hasl tinned pcb - tin solder blob as short:
+0.2µV

2) Tinned wire - copper pcb - tin solder blob
+0.5µV

3) Tinned wire - manually tinned pcb - tin blob
+1.6µV
This is somehow unexpected

3) solder wire as connection wire - manually tinned pcb - tin blob
+1.2µV
Totally unexpected, since there is only one material along the conduction path.

So for the tests with the resistors, I used Setup 1).



Results:
I can only compare the results. There is no value like µV/°C, since I have no clue of real temperature gradient across the resistor. But I tried my very best to make the setup as consistent as possible. I marked all the positions, used the same length wire and so on... Results are best performance first.

Vishay WSL0805R2000FEA, 0.2 Ohm 0805
+0.5µV
This part is marketed as low thermal EMF. I can confirm.

Welwyn LRCS0603-0R5FT5, 0.5 Ohm 0603
+1.7µV

Panasonic ERJ3RSFR20V, 0.2 Ohm 0603
+4µV
Thick film resistor, but pretty good

TE RLP73N2BR082FTDF, 0.082 Ohm 1206
+5.8µV

Yageo PE0805FRF070R2L, 0.2Ohm 0805
+6.3µV

Stackpole CSR1206FTR500, 0.5 Ohm 1206
+7µV

Yageo PE0402FRF7W0R2L, 0.2Ohm 0402
+7.2µV

Ohmite KDV06DR180ET, 0.18 Ohm 0603
+8.2µV
This is disappointing for a metal film 200mW current sense resistor!

If you want any resistors checked like this or if you have improvement ideas, contact me.
M.

[Kleinstein]

The tined copper trace is a mixed material. Much of the conductivity is still he copper trace and the composiotion in the solde can vary, as some copper transfers to the solder and some tin can go into the copper. The fresh solder wire can be different from the material of a solder joint. To some degree also the cooling speed of the solder or cold working of copper wire can make a difference to the Sebecke coefficients.

In the configuration for the shorts it is also a lot about the symmertry. Even if the materials are a poor choice, with symmetry there would be no resulting thermal EMF. The same applies to the situation with the shunt: if symmetric and thus no temperature difference between the ends of the shunt, there would be no thermal EMF. So for testing the shunt quality one should use an intentionally asymmetric setup. With a nearly symmetric setup like in a real application it is hard to judge the actual shunt quality.

[justanothername]

So for testing the shunt quality one should use an intentionally asymmetric setup.

I heated the shunts as well as the shorts from the side. See attached picture.

[dietert1]

The name is Seebeck, not Sebecke. A german physicist who lived from 1770 to 1831. Amazing how he discovered those small voltages. And the link with the voltmeter restauration is: https://www.eevblog.com/forum/other-blog-specific/great-meter!-the-hp-419a-dc-null-voltmeter-(restoration)/msg3588691/#msg3588691.

Nice measurements! Did you check the effect of the lamp on the nearby meter?

Regards, Dieter

[justanothername]

Nice measurements! Did you check the effect of the lamp on the nearby meter?

Thank you for reminding me.
1) for the actual measurements, the meter was 20cm to the left.
2) See attached picture. I shorted the inputs, set the range to 3µV, zeroed, turned the lamp on and took a picture 30 seconds afterwards. Looks good.

[Kleinstein]

The name is Seebeck, not Sebecke. A german physicist who lived from 1770 to 1831. Amazing how he discovered those small voltages. And the link with the voltmeter restauration is: https://www.eevblog.com/forum/other-blog-specific/great-meter!-the-hp-419a-dc-null-voltmeter-(restoration)/msg3588691/#msg3588691.

Nice measurements! Did you check the effect of the lamp on the nearby meter?

Regards, Dieter

The inital measurements of the thermoelectric effect where not measuring the voltage, but measuring the current in the very low resistance circuit. The current was enough to turn a compass needle.

[dietert1]

Seebecks report is here: https://archive.org/details/abhandlungenderk182233deut/page/n311/mode/2up?view=theater (German language). Nearly 200 years ago!

Regards, Dieter