Many of Pickering Interfaces range of resistor simulation modules include a specification for Thermoelectric EMF (also known as Thermal Offset), but it is not obvious why this parameter is important. This is the simple explanation.
Thermoelectric EMF's can be generated anywhere there are different metals and different temperatures. This includes, but it not limited to, relays. In modules that have power sources on them the cooling system results in PCB's having a temperature profile, so the generation of thermoelectric voltages is inevitable. Forced air cooling also tends to create temperature differences between the top of a relay and the bottom (next to the PCB) creating more complicated temperature profiles, and if the temperature profile is associated with changes in metals then thermoelectric voltages will appear.
Depending on the design the voltages can be measured to be anywhere form just a few µV to many100's of µV, particularly if many relays are required in the design.
When the resistance is measured with a DMM the DMM works by injecting a fixed amount of current into the resistor and then measuring the voltage drop it generates. Since the current is known the resistance can be calculated. However, if a thermoelectric EMF is present it changes the measured voltage and that causes an error in the resistance measurement.
Example: A 10 ohm resistance is measured using a 1mA current source which would normally create a voltage of 10mV. If a resistor simulator generates a 100µV thermoelectric EMF the resulting measurement is in error by 1%.
To minimise errors resistor modules should have a low thermoelectric EMF and should be measured with currents that make the errors low, repeating the above measurement at 10mA would reduce the error to 0.1%. Use of four wire measurement systems does not help, though it does remove the effect of any lead resistance.
Alternatively if the thermoelectric numbers are not significantly time varying you can measure their impact by reversing the DMM polarity and taking the average of the two readings. This will establish what the true resistance value is, but a circuit in an application may not do this so users need to be aware. Some DMM's have a facility for measuring voltage offset when measuring resistance and this can be used to compensate the resistance measurement instead of having to reverse the current polarity.
The lower the voltage across the resistor being measured in an application the more significant thermoelectric effects become. So the impact of this voltage is application dependent. User hardware rarely has the ability to remove thermoelectric voltages so it should be taken into account when simulating a resistor with a relay system.
The "Precision Resistor" range from Pickering Interfaces all use design methods that ensure the thermoelectric EMF is low, minimizing errors during calibration and in use. They uses a number of techniques used to ensure these modules have a precision which is hard to emulate by other methods. The precision resistor range relay based modules with the lowest thermal EMF are 40-260, 40-261, 40-262, 40-262 and 40-265. The 40-297 has the next lowest thermoelectric EMF since it uses design methods that minimize thermoelectric effects but has more series relays. Reed relay solutions such as 40-295 have the highest thermoelectric EMF numbers and are less suited to applications where precision is required. This is reflected in the accuracy claims of the modules where the 40-260 series has the best accuracy due to its low thermoelectric effects and use of very high stability resistors.
When measuring resistor simulators based on reed relays for verification purposes we recommend using the thermoelectric compensation mode in the measuring DMM. The DMM will either using a current reversal measuring technique or will measure the voltage and compensate the reading appropriately.
PXI Chassis Dependency
Thermoelectric effects are chassis dependent, if the same module is placed in different chassis it is likely the same resistor chain will exhibit different thermoelectric effects. This arises because the cooling between chassis has different effects on temperature profiles across the PCB. There are also noticeable differences between chassis which have the cooling fans under the PXI modules and those have cooling fans on the rear panel because the air flow for the former is more turbulent.
As thermoelectric effects reflect a temperature profile across a board it should always be remembered that placing a resistor module next to a high power lower load in a PXI chassis may influence the thermo-electric numbers seen. It is usually bext avoid such arrangements if the best performance is required, particularly on reed relay designs.