# Why am I getting inconsistent voltages at op-amp's inputs in a simple DC amplifier?

I'm new to circuit design and am unable to figure out why I'm getting inconsistent voltage measurements under certain circumstances at the inputs of an op amp in a very simple non-inverting DC amplifier, when the input voltage is from a TMP36 temperature sensor.

First, a circuit that uses a voltage divider as the input and that gives the expected results:

In particular, the voltages at both inputs are very close to equal (sometimes they differ by a millivolt).

Now I replace the voltage divider with a TMP36 sensor, which outputs a voltage proportional to temperature. According to the the datasheet, it's able to source up to 50 uA from the output.

Now I see an 8 mV discrepancy between the inputs when they're measured relative to ground, but if I measure directly across the inputs, I measure a 0 mV difference.

Why the inconsistency? My multimeter has 10M impedance and so shouldn't be drawing more than 0.07 uA, which is far, far below the published 50 uA limit. The TMP36 datasheet doesn't give any hint that the output voltage should be influenced at all by the output current within the specified range. A 1% reduction in Vout seems too large, but maybe my expectations are too high.

• In the second circuit, do you get the correct or expected output? Also, if you have two meters, to debug the second circuit, put one meter on the op-amp output and watch it while you probe the op-amp inverting input (or really, try both op-amp inputs). See if the output jumps when you connect the probe. Even though it is a little surprising, it does sound like maybe your meter is loading the circuit a bit. Feb 23, 2022 at 19:02
• Oh, and just to be more explicit, I do not conclude that you have an 8mV offset. Because you are not measuring 694 and 702 at the same time. My conclusion is that connecting the probe is causing the voltage to change. Feb 23, 2022 at 19:04
• Understood, thanks. So a second meter might show pin 2 dropping when I measure pin 3. Feb 23, 2022 at 19:13
• Or it might show pin 1 dropping when you measure either 2 or 3. Feb 23, 2022 at 19:23
• There could be an interaction between your TMP36 output and MCP6004 input. Here is a simple test. Put a 4.7k resistor between the two along with a 0.5uF cap to ground on the +input of the MCP6004. Such filter may make the difference. Feb 24, 2022 at 1:23