I'm designing an automotive oil temperature sensor based around a 3-wire RTD. This is my analog circuitry:

3 Wire RTD Temperature Sensor

U1 and U2 provide a 1mA bias on the RTD which is measured by U3 and gained up by U4. R5, R10, D1 and D2 are for input protection on the op amps. Op amps are a quad package MCP604 single supply and are properly bypassed with a 0.1u cap.

I have the whole thing prototyped out and it works perfectly if Vin to the 7805 circuit is powered by my 12V bench supply, but as soon as I connect it to 12V battery power in my car, I get some sort of ripple voltage or oscillation in the RTD that throws the ADC readings way off.

I don't have access to a scope at home to really see what's going on, but there's a small ~10mV ripple voltage on the RTD which is being amplified by the gain stage. I added the 1uF caps in the feedback path to try and remove it, which helped, but the ripple is still bad enough to render the measurements useless.

If I disconnect Vin from the vehicle 12V and power it with my bench supply (with GND remaining through the vehicle chassis) the ripple disappears and the temperature reading is spot on.

Powering it with a 9V battery also works just fine.

This is with the vehicle ignition off, so there shouldn't be any crazy noise present.

Also, if I leave it powered by vehicle 12V and replace the RTD with a 100 ohm resistor, it works fine. So it looks like some sort of interaction with either the platinum resistance element or the sensor leads.

Is there some sort of interaction I'm not accounting for with vehicle power vs. my bench supply or 9V battery? Or is there a potential stability issue with my circuit that is being pushed over the edge when connected to vehicle power?

EDIT: I'm fairly certain this is a ground loop issue. Rewiring the circuit into a star-ground configuration improved the situation, but there are still measurable stray ground currents through the RTD causing unpredictable results.

  • \$\begingroup\$ U3 seems to be a unity gain follower since there's no voltage across R8. I would make U3 a differential amp connected right across RTD1 to try to reject any common-mode noise. I'd also give it some gain, since the earlier you can get gain the better from an overall noise perspective. \$\endgroup\$
    – John D
    Aug 17, 2014 at 1:07
  • \$\begingroup\$ I should cite microchip AN687. U3 is effectively giving the RTD voltage minus the voltage drop on the wire. Using a differential input would require a 5th opamp for wire resistance compensation as in TI AN1559, which I was hoping to avoid. \$\endgroup\$
    – slintux
    Aug 17, 2014 at 2:13
  • \$\begingroup\$ In an unrelated note, perhaps you should move D3 catode's connection to VIN in order to effectively deflect excesive voltage from the 5V line. It wouldn't be effective against voltages below VIN, though. \$\endgroup\$ Aug 19, 2014 at 2:18
  • \$\begingroup\$ The intent was for D4 to handle that, but now that you mention it, VIN is going to be lower than the breakdown of D4. \$\endgroup\$
    – slintux
    Aug 20, 2014 at 16:54

1 Answer 1


Increase R5 and R10 to 50K and add a 1uF ceramic from U3 +input to ground and from U2 +input to a solid analog ground (keep the analog grounds together). Split R8 into two series 50K and add a 1uF ceramic capacitor to ground from the mid point. Lose the two Schottky diodes (too leaky, especially with 50K) and R11 (useless).

Make sure your return wire from the RTD (you must have three wires running back to your circuit from the RTD, which must not be grounded at the chassis) is connected to the low side of R14 and thence to circuit ground as I said above about keeping them together.

If you don't have access to a 'scope you'll be fairly limited, but if you're still seeing 'noise' try measuring the voltage at each op-amp output (including the current source) using a multimeter on the minimum AC volts range. Many will block the DC and show you some indication of what is going on in the AC world.

  • 1
    \$\begingroup\$ This did the trick. The real key was grounding the RTD on the low side of R14, rather than at the star-ground point. \$\endgroup\$
    – slintux
    Aug 18, 2014 at 1:34
  • \$\begingroup\$ @slintux Glad to hear it, the other suggestions may help when there is more electrical noise (eg. when the engine is running). \$\endgroup\$ Aug 18, 2014 at 1:55
  • \$\begingroup\$ R11 may be useless for normal circuit functionality, but it helps with troubleshooting when you want to figure out whether the signal source is dead, or the signal sink: just open the resistor and see what changes. I've had a product where a lot of time got wasted when those resistors were absent. Dozen per each board - one op amp died and good luck figuring out which one was loading down the common reference voltage. They were a mea culpa case of penny wise, pound foolish. Next rev had that fixed with vengeance. \$\endgroup\$ Apr 19, 2022 at 6:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.