# Voltage divider : using board voltage vs Power supply

I am using an ncd relay power switch that is capable of reading 8 sensors.

When I use the 5V and GND provided by the board I get a Vo=3.46V. If i switch R1 to 100k I actually get Vo= 2.7.

When i use an external 5V and GND provided by a power supply I get Vo=2.6, still using R1=10k here.

Question: Why am I getting different voltage values from the two different power sources. Why am I having to switch R1 to 100k to get the proper Vo when using the relay board? Is there some voltage reference I need to take into account?

According to curstomer support "the ADC inputs on that board float(they are not pulled either direction". simulate this circuit – Schematic created using CircuitLab

• Are you sure your thermistor stays at the same temperature between measurements? Because if you do any soldering to replace that R1 you most likely affect thermistor too. – Maple Jun 28 '18 at 19:11
• Replace the thermistor with a 10K resistor and test again. – Spehro Pefhany Jun 28 '18 at 21:20
• I did that and my Vout is 3.34. If i dont connect to the AD input i get the proper voltage of 2.5v. – AP Shwarts Jul 2 '18 at 15:44

The input impedance of an ADC with an analog mux can be pretty low. Typical microcontrollers recommend a source impedance no more than tens of kilohms. Changing R1 to 100k$\Omega$ may lead to a situation where the input impedance of the ADC changes the measured voltage. Either stick with low value resistors or add a unity-gain buffer between the thermistor and the ADC.
The "5V" provided by a typical board can easily vary by $\pm$5%. If you are relying on voltage dividers for precise measurements you need to give them a precise voltage to begin with. Consider using a precision voltage regulator rather than either of the 5V sources you've mentioned.