# ADC is giving fault codes when I touch thermistor attached to it

I am working on writing a control program for my pellet smoker. I am having an issue with the meat thermometer that came with the smoker.

I have a MAXIM 31865 ADC that is connected to a Raspberry Pi. Attached to the ADC I have a thermistor that is used as a meat thermometer. This thermistor (unknown type/manufacturer) is inside of a metal sheath and has 3 wires that were previously connected to the smoker. There are 2 internal wires that connect to the thermistor and then the metal cable sheath itself was also connected to the original control board on the smoker.

I only have the 2 internal wires that actually connect to the thermistor connected to my ADC. The third wire (thermistor sheath/cable sheath) is not connected to anything. If the thermometer is sitting on the grill or on my desk I am able to get an accurate measurement from the ADC. However, if I pick up the thermistor or touch it more than just lightly brushing it I instantly get an overvoltage/undervoltage error on the MAX31865 ADC that doesn't go away until I put down the probe.

I assume I need to ground that third wire by clipping it to a metal surface on the grill or back through the power supply, but I am trying to understand what is happening here. I tried measuring for a voltage drop/increase while touching the thermometer and I couldn't see anything happening when measuring with a multimeter, and I don't understand why it would be a problem anyway.

(I am not an electrical engineer, just a hobbyist.)

Image of how it is wired I put together before I started. The black triangles go to a strip of copper on the solderable breadboard. The empty triangles go to a different strip of copper, and then the grounds are merged to a single ground pin on the Raspberry Pi. The small black boxes are ferrite beads. I pretty much just followed exactly what the Max31865 docs said, but since I had a single power source I split it with a ferrite bead between the source and the analog voltage input.

I used a 350k Ohm reference resistor because the thermistor has a resistance of around 300k Ohms at 0 degrees celsius.

• The devil is in the details. How are you powering it and how is it wired? – Kartman Mar 8 at 1:40
• Depends on the impedance and with line voltage fields everywhere your finger can inject common mode noise, so you are going need some suitable cap sizes to suppress it. – Tony Stewart EE75 Mar 8 at 3:22
• I added an image of how it is wired and a description to the original post. The sheath wire on the RTD is currently not connected to anything so it is not shown. – JohnStrom Mar 8 at 4:12
• Why are you trying to use a 300K thermistor with a chip that is designed for use with a 1K maximum platinum RTD? – Spehro Pefhany Mar 8 at 4:24
• If you don't know the characteristics (resistance vs temperature) of your sensor, how are you planning on converting the voltage/resistance to a temperature? – SteveSh Mar 8 at 11:46

You need to ground the shielding (sheath). This is what keeps you from capacitively coupling noise into your high impedance inputs. Ground reference to the ADC is important. If you don't ground the ADC circuit board, at least put a 1 MOhm resistor to ground to provide some reference.

Edit: There may not be a good reason to keep your digital and analog grounds separate since the analog and digital VDD are the same. The shielding of your probe should be connected directly to the analog ground. Still, if you want the isolation, put the 1 MOhm between the analog and digital grounds.

• This is probably a stupid question, does it matter if I ground it by just clipping it to the metal smoker? Should I run a wire back into my build box and attach it to the common ground on the power supply? – JohnStrom Mar 8 at 4:11
• I don't think any question about grounding CAN be stupid. It's a tricky business. Ask some audio engineers about how important grounding is. They will all agree that it is important, but they won't all agree on how it's done properly. I added some content to the answer to try to clarify. – ScienceGeyser Mar 8 at 4:17
• Again, how are you powering it? Your power supply could be injecting half mains voltage at a few 10's of microamps - which is very common. Is your power supply grounded (ie does the mains cable have an earth pin?). Failing that, earth the 0V rail of your electronics and the sheath of the thermistor/RTD (what is it - thermistor OR RTD - there's a big difference) – Kartman Mar 8 at 9:21
• @Kartman The power supply for the Raspberry Pi is connected to the main via only the LN wires. The ground pin for the smoker is connected to the body of the smoker. The Raspberry Pi has no earth ground. I have read people connect one of the 0v rails of the Pi back out to the GND pin for AC for an earth ground. I am not currently doing that. Other places I have researched have recommended against doing that. It is a thermistor not an RTD. – JohnStrom Mar 8 at 22:30
• Measure AC volts between 0V and mains earth. What do you see? Then try connecting 0V to mains earth and seeing if the problem still exists. – Kartman Mar 8 at 22:34

This chip needs an LDO and caps across thermal sensor and from sensor to ground such that RC>>= 1 second for high impedance types.

It works by computing ratios on a bridge with an equal Ro reference but is designed for low R RTD’s.

$$\R_{RTD}(t) = R_0 [1 + At + Bt^2 + C (t – 100°C)t^3 ] \$$

$$\R_{RTD}(t) = R_0 [1 + At + Bt^2 ] \$$.
where: (1st for temp, t < 0°C) (2nd for t > 0°C)

So your sensor must match these formulae with constants A,B,C

But for now you have an EMI problem.