I am trying to assemble a circuit based on this datasheet: app note circuit The schematic I derived from the circuit is: schematic This was made into a PCB footprint pcb footprint

The readings I am getting from this circuit are much too low though. I am using GPIO2 on an esp-wroom-32ue the analog readings have 12bit resolution and therefore span 0-4096 for voltage 0-3.3V. The circuit is designed to produce 0V at 0C and 4V at 400C. But when I submerse the k thermocouple in near boiling water I only read an analog value of around 12. I am expecting to read a voltage of about 1V here this corresponds to 1V*(4096/3.3) = 1241 analog reading.

At room temperature I am reading a 0 analog value at an ambient temp of 27C. A few observations I wanted to mention:

  1. During assembly I installed IC202 (LT1025A) in the improper orientation and this may have damaged the IC. I have reoriented the IC since this error.
  2. I noticed that when I read the voltage across the 246K resistor I am getting just above 1K, I believe this to be because of the adjacent capacitor but am not sure if this is a problem.
  3. The circuit does respond to extreme changes in temperature within several seconds.

edit: It was mentioned that the impedance provided by the voltage divider circuit on the ltc1049 was providing negatively affecting the reading. I did remove R203 and 204 and am seeing somewhat better results. Does anyone know why having the 330 and 510 ohm resistors as my voltage divider was a bad choice for this circuit?

  • \$\begingroup\$ Start from the start, measure the voltage across the K-Type, see if it matches what it is supposed to. Work forward from there. \$\endgroup\$
    – Tyler
    Jul 18 at 21:07
  • 1
    \$\begingroup\$ Are R203 and R204 the correct values? \$\endgroup\$
    – James
    Jul 18 at 21:39
  • \$\begingroup\$ @james I think this may be a problem. I was just trying to step 4V down to protect my 3.3V pins. I don't expect to ever see 400C on my sensor (or any where near that) I was just doing this in case \$\endgroup\$
    – Feynman137
    Jul 18 at 21:45
  • \$\begingroup\$ Just out of curiosity (and speaking from painful personal experience) are you aware that for K thermocouples, red is the negative lead? \$\endgroup\$ Jul 18 at 23:18
  • 1
    \$\begingroup\$ Read AN28, "Thermocouple Measurement," by Jim Williams. I don't know what is wrong with your circuit. But after you read AN28 you will probably figure it out all by yourself. This app note is very famous and widely read and referenced. analog.com/media/en/technical-documentation/application-notes/… \$\endgroup\$
    – mkeith
    Jul 19 at 5:33

That particular micro-power op-amp has a typical short-circuit output current in the 400uA range with a 5V single supply, so you are overloading the output stage with the low value divider values you are using. You should not come anywhere close to that value- output swing is characterized for loads of 10K and 100K so somewhere between those two (for the divider in parallel with the feedback resistor) would be good.

Edit: Relevant curve from datasheet. Note that one should not come anywhere near to the typical values since parts vary from unit-to-unit and with temperature, and short-circuit current capability is not a good indication of what loading is acceptable under optimal conditions:

enter image description here

Based on this, I don't think drawing much more than 100-150uA is safe.

As an aside, there are many color codes for thermocouples, and red(+)/blue(-) is correct for the Chinese product you are using (as well as British standard).

There are better and cheaper zero drift op-amps that you might want to substitute, or you could change the divider. I believe that the module you are using already includes a divider so a simple series resistor R203 may suffice with R204 open.

Alternately, change the feedback resistor ratio around the op-amp and avoid the external divider entirely.

Note that IC202 is supposed to be measuring the temperature at the two junctions formed where the Chromel and Alumel wires connect to copper, and any temperature differences between those will result in (likely varying) errors in your reading.

Edit: You want the three points shown with arrows below to be at the same temperature as each other (the T/C points would be the two points where the copper terminal strip and lug contacts the thermocouple wire). Everything in the green circle should be thermally isolated from other heat sources on your board and otherwise. Any errors in the temperature measurement are directly reflected in the reading.

enter image description here

The LT1025A produces an analog output that is a rough linear approximation of the specified thermocouple output near room temperature, and to do so it needs to be at the relevant temperature.

  • \$\begingroup\$ Is IC202 being used incorrectly for cold-junction compensation or are you just highlighting the importance of this? Wanted to ask if you could explain why a low resistance divider would overload the op amp versus a higher resistance amp. To someone with little experience (myself) it seems like it would be the exact opposite. \$\endgroup\$
    – Feynman137
    Jul 19 at 13:47
  • 1
    \$\begingroup\$ I am highlighting the importance because I don't see it thermally isolated (from other parts on the board) and thermally coupled to the terminal block or whatever you are using to connect to the thermocouple materials. As to the second point, a low resistance divider draws more current from the op-amp. For example, with maximum output voltage of 4V your 510+330 = 840 ohm divider (837 ohms taking the 246K feedback resistor into account) would draw 4/837 = 4.8mA or more than 10x what is it typically capable of under short-circuit conditions. \$\endgroup\$ Jul 19 at 13:52
  • 1
    \$\begingroup\$ @Feynman137 - You want the LT1025 to be as close as possible to the points where the TC wires are connected to the PCB. It provides a voltage source for the TC which is determined by its temperature - and assumes that the connection point will be at the same temperature. You don't show the terminal strip you probably use with that TC (given that it is provided with spade lugs). If it is at any remove from the LT1025 you will introduce errors. \$\endgroup\$ Jul 19 at 16:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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