Post Go back to editing

Rsense accuracy for LTC2986 and other questions


I'm looking to use a LTC2986 to measure Pt100 RTDs and wanted to know the accuracy requirement for the Rsense to meet the temperature error specifications. I couldn't find the error contribution anywhere in the datasheet but I guess since it's ratiometric a 0.01% error in the Rsense would be a 0.01% error in the RTD resistance?

If the chip is powered and no measurements are active I assume the inputs are high impedance and I could measure the Rsense directly with a 4 wire measurement? I just see that the data sheet is always listing the Rsense resistance in ohms to two decimal places, so I assume it needs to be very accurately reported. 

I also couldn't find an ESD tolerance specification in the datasheet. Is there any integrated protection?

Lastly I assume the higher excitation current the better for signal to noise and leakage current compared to excitation current, so I was intended to use a 1k Rsense with .5mA excitation, but the specified error for Pt100 RTDs mentions using a 2k Rsense, requiring a lower excitation current. Am I missing something?

Thank you!


  • Hi Matt, 

        You are correct that the RTD measurement is performed ratiometrically, the LTC2986's voltage reference error nearly disappears and you are left with the Rsense resistance being the reference. There is a scaling of the error term due to the ratio between the Rsense and the RTD - for example, a 0.01% error on a 1k Rsense with a PT-100 at 0C would effectively be only a 0.001% error in the RTD reading since there is a 10:1 ratio between Rsense and the RTD. Keep in mind that the Rsense tempco will contribute to the error budget as well. 

    Download the evaluation software if you have not yet, it includes an error estimator that will let you play with values of Rsense, tolerance, tempco, etc. and see an error band. Note that the RTD itself has a tolerance and accuracy specification, the GUI will let you enter this as well. The manual has good information on using the error estimator. 

        Yes, you should be able to measure the Rsense while the device is idle - do not exceed ABSMAX as this will cause current to flow in the ESD structures. Rsense can be configured in the part with 1/1024 ohm resolution. 

         The part does have ESD protection up to 2kV HBM - current limiting resistors or any RC filtering will improve this robustness.

         Yes, higher SNR is better, we typically recommend 250uA or 500uA. The 2k Rsense was an arbitrarily chosen example, it is used on our demo boards to give decent ratios even with 10k NTC thermistors. Choose any value of Rsense such that Iexc * Rsense does not exceed 1V and has the best tolerance and tempco you can find for your budget. Typically values between 1x and 5x of the nominal RTD value work well. 

    Hope it helps,