Voltage divider - resistance values not stable

Hi. I hope someone can help me with a project I am working on. It's basically a temperature meter and a voltmeter in one unit.

I am using an Arduino nano and 1. a non linear temperature sensor, which I have then interpreted with a steinhart-hart equation and the numbers are surprisingly accurate. I first read the resistance of the sensor using a voltage divider.

The values of the sensor are 20°C 2000ohm, 80°C 200ohm, 100°C 110ohm, 150°C 30ohm. Currently I have the known resistor of 10kohm, which I will replace with a 1kohm or 220ohm, since I am mostly interested in the values from 50°C to 150°C and thus only need to accurately read from 570ohm to 30ohm. I am not sure if this will solve my problem.

The problem is that the resistance values are jumping in the 10% region. I would like them to be smooth and in term not jump the calculation numbers. I would like to address the issiue on the hardware side. I have tested the thermistor and it is stable. If i replace the thermistor with another ordinary resistor the values are also jumping. Would it be possible to solve this with capacitors? I am not sure where to place them as I have not seen an example of their usage on a voltage divider.

The same problem occurs when I am measuring voltage on another analog pin. There I measure 0-15V and the values are also not stable. How can I use capacitors (and which ones) in this case?

I would really appreciate your help and would prefer an answer for dummies since I am only a hobbyist and don't know that much about circuitry :slight_smile:

Sounds like one or more bad connections. If you are using a breadboard, use different slots for the part leads, or solder the connections.

Is the power source stable? Are there other things connected to the Arduino? Can you post a circuit diagram and your code?

zigahim:
I first read the resistance of the sensor using a voltage divider.

Show a diagram of your circuit. And show what you are measuring, such as a voltage.

Also show some details or data about this non-linear temperature sensor you have - such as what values of resistance is it expected to have at particular temperatures.

Wow, I am very thankfull to get so many answers so quickly.

I will provide a diagram but i have to draw it in a digital form first :slight_smile:

The circuit is not on a breadboard anymore but soldered on a protoboard.

I have managed to make the results more stable by using a 1k resistor instead of the 10k one. However in the range that I am most interested in there is still some shifting. Reading 1k now gives a stable result.

The 5v supply is from the arduino nano 5v pin. The question is can i solder a 100ohm resistor in there or will it draw to much current? Should i go as low as 220ohm at the most?

The Nano gets its USB power from where?

As mentioned previously we really need to see a circuit diagram.

Also, please elaborate on this statement "I first read the resistance of the sensor using a voltage divider".

It sounds like it may be a case of circuit 'loading' wherein the device used to make the measurement is altering the circuit. It is accurately measuring the value that is there while the measuring device is connected but this is not the value that was there before the measuring device was connected.

Don

zigahim:
The values of the sensor are 20°C 2000ohm, 80°C 200ohm, 100°C 110ohm, 150°C 30ohm. Currently I have the known resistor of 10kohm, which I will replace with a 1kohm or 220ohm, since I am mostly interested in the values from 50°C to 150°C and thus only need to accurately read from 570ohm to 30ohm. I am not sure if this will solve my problem.

For starters I'd suggest you to find an NTC that has a much higher resistance. For those temperatures a nominal resistance of 100k at 25°C may be a good start, you would typically read it with a 10k pull-up for those temperatures.

The same problem occurs when I am measuring voltage on another analog pin. There I measure 0-15V and the values are also not stable. How can I use capacitors (and which ones) in this case?

15V on your analog pin? That can't be right.

If your multimeter is fast enough to measure instability then it's not electric noise which you could filter out with a small capacitror, indeed it's more likely a poor connection. Is everything soldered, or are you just wrapping wires around one another or (even worse) a solderless breadboard?