I am going to build temperature sensing system using PT100 sensor (and later Raspberry PI 3). I refer my work to Microchip note AN687 Precision Temperature-Sensing With RTD Circuits. On the page 4 is shown circuit I want to build: enter image description here

I am beginner. I have a question: why values of resistors R8, R9, R10 are like they are? I researched some about Sallen-Key filter but didn't find identical circuit. In other words - how values of elements were calculated?

  • 1
    \$\begingroup\$ That is a very bodgy 3rd order low pass sallen key filter that you'll struggle with to find design equations so state what frequency response and DC accuracy you require and maybe someone will help. \$\endgroup\$ – Andy aka Sep 24 '18 at 10:11
  • 2
    \$\begingroup\$ Why this filter is bodgy? \$\endgroup\$ – matoex Sep 24 '18 at 10:37
  • \$\begingroup\$ I am going to read temperature on Raspberry using MCP3008 chip (ADC converter). It's my goal, if it'll no suceedd I will use Adafruit board designed for PT100 usage - adafruit.com/product/3328 \$\endgroup\$ – matoex Sep 24 '18 at 10:41
  • \$\begingroup\$ Isn't just a capacitor enough? \$\endgroup\$ – Gregory Kornblum Sep 24 '18 at 11:38
  • \$\begingroup\$ @GregoryKornblum Obviously the filtered output follows far too slowly the temperature changes if a single capacitor filter is used and it attenuates enough at 50Hz. To reach 0,5 bit accurate settling for 12 bit ADC needs time = 9 * RC time constant. \$\endgroup\$ – user287001 Sep 24 '18 at 12:10

Your linked application note actually tells quite thoroughly the ideas behind the filter, but it's unfortunately engineer's speak to another engineer.

You can simulate the filter. It is at the same time a lowpass filter and an amplifier. The filtering is needed to attenuate all noise that the probe collects from disturbing fields. The component values are selected surely to provide enough attenuation at the most harmful frequencies 50...60Hz and simply to have values that are available. One needs accurate enough resistors, especially R10 and R11 must be ok because they affect the scaling.

3rd degree filter also needs more accurate component values than simpler filters. Bad filter, which attenuates enough at 50 Hz and up, can cause too long fluctuating output or too slow rise to the right voltage when the temperature changes rapidly (=for example the probe has been put to cold or hot liquid). It surely is a design criteria to get the filtered signal to settle fast enough without too much overshoot.

A part of the needed filtering and the calibration can be made in PIC software. To keep the calibration and programmed filtering valid the component values must stay stable.

ADD: The later discussion in comments reveal that you are sensing outdoor air temperature. A simple RC filter surely is enough, because that application can well stand for ex. 15 seconds long settling time.


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.