Thermoelectric voltage over a solder joint of two identic copper wires My background: I'm engineer of electronic HW engineering and embedded computer systems. During repair of a vintage digital voltmeter (resulution 1µV) I discovered an effect that probably only a physicist can answer. I isolated the problem and condensed it to a simple experiment:


*

*Take a piece of copper wire (about 20cm) and cut it in two halves.

*Join the two halves again by soldering. But don't let the two wires come in direct contact. They should be connected via the solder bead alone.

*Measure the voltage across the joint with a 1µV resolving voltmeter.

*Apply heat via hot air (about 100 degrees Celsius) to the solder joint.

*There is a voltage of about -20µV.

*Swap the wire ends at the voltmeter side.

*The voltage is about 20µV (to be expected).


My question: I assumed, the thermoelectric voltages from copper to solder and solder to copper would cancel out. But they don't. Why ?
Thanks !
P.S. The solder is regular solder, leaded, out of the pre-lead-free age.
 A: Mario, welcome to physics SE. You are right, the effects should cancel because of the symmetry of the junctions. In fact, a similar technique is commonly used to null out the effects of ambient temperature at the measuring gauge when a long run of thermocouple wire is used between the junction and the gauge. So you have uncovered a subtle and interesting effect! Here are some thoughts. 
Zeroth, try the same measurement with a DIFFERENT voltmeter and leads to see if the meter itself is properly functional. 
First, it is always good & prudent to check the null-point calibration of the meter, when making measurements of order ~1 mV- especially when the device has been recently repaired. 
Second, how are the meter leads attached to the ends of the copper wire? with toothed clamps? If so, are the clamps metallurgically identical?
Third, how are the meter lead plugs connected to the input jacks on the meter housing? Same commentary applies. 
Try these things and then write us back here with your results so we can ponder the imponderables some more, as needed. 
Good luck!
