When reading an RTD temperature using an ADC, the following topology is usually used:

RTD ADC reading

Most guidelines insist that a precision resistor (0.1%) be used for RREF (inside red rectangle). I believe the reason for this practice, is to use the full range of the ADC device (correct me if I'm wrong). Would any problem arise if I don't need to use the full range of the ADC and I use a normal resistor with 5% accuracy instead of a precision one? Keep in mind that a full scale system calibration will be performed and I have also performed all calculations to make sure nothing goes outside the acceptable limits of the ADC.

  • \$\begingroup\$ ADC range usually means the difference between highest and lowest voltages it reads. Are you using 'range' to indicate the precision with which you are expecting to read it, so the number of bits you can have confidence in? A calibration will take out initial error in the resistor, but a 5% resistor may have much worse resistance tempco than a precision one. \$\endgroup\$ – Neil_UK Mar 2 '16 at 13:41
  • \$\begingroup\$ @Neil_UK Now that you mention it, I probably got something wrong about the 'range'. What is resistance 'tempco'? \$\endgroup\$ – hadez Mar 2 '16 at 15:27
  • 1
    \$\begingroup\$ It's temperature coefficient of resistance. Typical 1% resistors may be 200ppm per degree, 5% may be worse. The best precision resistors may be quoted in the 50ppm to 15ppm range depending on cost. For comparison (why you don't use tracks for current shunt resistors) copper has a tempco of 0.4% or 4000ppm per degree! \$\endgroup\$ – Neil_UK Mar 2 '16 at 16:47

The designator of the component should be a clue-\$R_{REF}\$.. it's the REFERENCE resistor to which your RTD value is compared.

The calculation of the RTD should use the ratio of the ADC readings of the sensor and thee reference resistor.

You MUST have an accurate and stable reference resistor if you want an accurate and stable reading. On the other hand, the ADC reference VOLTAGE drops out of the equation so as long as it is stable enough that readings of the reference resistor and RTD are made with essentially the same reference voltage, then it can be inaccurate and/or drift around slowly with temperature and/or time.

Yes, you could calibrate out the initial inaccuracy of a 5% resistor, but a 5% resistor is highly likely to severely compromise the accuracy of reading even a cheap thin-film RTD due to temperature shift and drift over time and other factors.

| improve this answer | |
  • \$\begingroup\$ If had the impression that precision resistors had a very precise value with very small tolerance among different samples. According to your response, they are superior over other resistors in other factors including drift and and shift over time and temperature. Am I correct and is this the reason that precision resistors must be used? \$\endgroup\$ – hadez Mar 2 '16 at 15:33
  • 1
    \$\begingroup\$ Yes, that, absolutely, and the fact that a 0.1% resistor means that no calibration is required for most purposes. Calibration costs money and can be lost (screwdriver-based circuit drift or EEPROM corruption, for example) and then must be restored. A 0.1% 25ppm resistor is very, very cheap in comparison (< 10 cents in quantity). \$\endgroup\$ – Spehro Pefhany Mar 2 '16 at 17:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.