Current shunt temperature coefficient

[splin]

Searching on Digikey finds some surprisingly cheap Stackpole BR series (< $0.2) 5mohm 3W 20ppm/ºC shunts. Cheap for 20ppm/ºC that is. But the datasheet actually says:

 "±20 ppm/ºC element wire
 ±200 ppm/ºC at lead stops"

Hmm so does that mean you use the 200ppm/C figure as an overall TCR? That seems very pessimistic. TT electroncs OAR series shunts are similar and also cheap. The OAR data sheet only states 20ppm/ºC for the wire - it doesn't provide an overall TCR value.

I guess its complicated by the relative heating of the resistance wire and the leads - the OAR datasheet shows some pictures of the temperatures of the resistive elements and the copper leads under load. E.g. the 3W, 5mohm version running @ 3W shows the element at 160C and the solder joint at 70C.

So what TCR would you use? I assume that the overall TCR varies with power levels so what if it's running much below its rated power, say less than 10% with a temperature rise of perhaps only 20 or 30C above ambient? Yes I know there is no substitute for actually measuring some in your own circuit usage but I assume someone is familiar with these type of shunts and their characteristics.

[T3sl4co1l]

If what matters is the total voltage dropped across it, you're kind of stuck.

If what matters is the voltage drop across the Kelvin sensed section (assuming it comes with wires, or you can add them), then you're fine.

Tim

[Zeranin]

I haven’t looked up this Digikey part, but I presume these are 2 terminal resistors, in which case you will never achieve anything like 20ppm for a 5mohm shunt. Why so?

Copper has a TCR of ~3900 ppm/K, As a 2-terminal device, this resistor must have some ‘lead resistance’. Let’s assume that the copper ‘leads’ contribute only 0.25 mohms, that’s a very small resistance indeed, but it’s still 5% of the total resistance of 5 mohms, so the copper will contribute (5%x3900) =195ppm of TCR! Enough said? If you repeat the calculation for a 10K or 1K resistor then you find that the TCR effect of the leads is rarely an issue, but for 5 mohms it’s a different story, and you will never get decent TCR without going to 4-terminal construction.

[Marco]

I think you are meant to solder separate wires if you want a Kelvin connection, the pinched legs make it awkward to drop it on the PCB, not very convenient.