3 wire RTD, 4W read mode, 4 wires connected

[itsbiodiversity]

Just gathering some thoughts from the community.  Reference the attached picture.  I noticed that a 3 wire RTD is much more accurate when connected this way, with the two HIs connected to the reds and the two LOs connected to the singular white wire.  The Fluke 8846 or Agilent equivalent do not have 3 wire RTD reading functions.  I have been in contact with Fluke and they are looking in to this, but I am interested in what you guys/ladies have to say.  When using the 8846 to read a 2 wire RTD there is over 1°C of error, any troubleshooting tips?  Thanks.

[The Soulman]

Your temp sensor is measured like a low-ish value resistor, in two wire mode any additional wire and connection resistance is also measured.
For best accuracy and repeat ability use four wire mode/connection, three wire is a compromise between the two. 

[itsbiodiversity]

I appreciate that.  I am hoping for some explanation as to why I can or cannot connect and use 4 wire like the picture I attached. 

[Jorn]

3 wire RTD only gives good results used with a bridge circuit where the current in the sense lead is zero and the resistance in the two remaining leads cancel out.

4 wire RTD with 4 wire ohms measurements is of cause the best solution but if 2 or 3 wire RTD is what you have you should at least use 4 wire resistance measurement to eliminate contact resistance between wires and multimeter. Contact resistance will vary over time, temperature, humidity and vibrations. Contact resistance tend to jump suddenly from one value to another after long periods with small variations.

The wire resistance to the RTD is relatively stable compared to contact resistance and it can thus be subtracted from your measurement.

If you need absolute precision you will of cause have to use a reference thermometer to calibrate your RTD.


best regards
Jorn

[Jorn]

And to answer your question: Yes you can use the setup you showed (and subtract the resistance of the white wire)

Hint: The resistance of the white wire can be estimated as half the value of the resistance between the two red wires.

Maybe your fluke multimeter can do the math for you real time...

[wraper]

Basically you could attach your 4W setup to just 2 wires of RTD and leaving 3rd wire unconnected, take measurement. Then take measurement in 3W as you just did. Calculate the difference between measurements, and then subtract that difference  from your 3W measurement. Then you will get a fully wire resistance compensated result.

True_temp = 3W_temp - (2W_temp - 3W_temp) = 3W temp * 2 - 2W_temp

[HighVoltage]

For your Fluke 8846A you should really use a true 4 wire RTD, because it is not setup for accurate measurements on a 3 wire RTD.
Reason: you are missing the resistance of one return wire.

3 wire RTDs are made for precision automation equipment that calculated the proper value for the missing wire and adds it mathematically.

When I use a RTD on one of my bench meters, I always use a true 4-wire RTD.

[Dave]

3 wire RTDs are made for precision automation equipment that calculated the proper value for the missing wire and adds it mathematically.

Not quite. You just need to cancel the resistances out. You can do this by injecting the same current in two leads or by clever subtraction, either works fine.
This can be done entirely in the analog domain, no need for any computation. The only thing you're relying on is that the resistances of the wires are going to be equal (because they are in the same cable).

[HighVoltage]

Ok, I understand your reference to the correction in the analog circuit.

But how can I do that on a Fluke 8846A or similar instrument?

[Dave]

Mathematically, it could be done by connecting it like OP did and switching back and forth between 2- and 4W mode and subtracting the difference from the 4W measurement. However I'm not familiar with Fluke bench meters and if they allow for fancy scripting that would facilitate this sort of measurement directly on the instrument. It could definitely be done through GPIB (or similar) on the computer.

The second idea that comes to mind is a circuit that mirrors the current injected by the force terminal and also shoves it into the second lead. That way you could measure things directly on the meter (possibly ).

[itsbiodiversity]

I have a four wire RTD with really good uncertainty.  I know the "standard" temp value.  With this setup I have gotten results like
-19.93std -19.38uut ::: 30.22std 30.74uut ::: 134.882std 135.092uut

Does my connection seem "acceptable"?  I have validated the results - I am looking for some validation from an engineering / theoretical principle standpoint.

[HighVoltage]

That looks like 2 to 3% difference and that is really high for lab RTD.
A good quality RTD is more in the range of 0.5 % or better.

Really good RTDs come with a calibration certificate and values for Alpha and 0 degree value.