Post Go back to editing

LTC2984 Demo Board Strange Behaviors

Category: Hardware
Product Number: LTC2984

Hi, I have LTC2984 demo board and DC2212 thermocouple daughter board and I'm working to use it with FPGA. I cannot obtain correct temperature values from cold junction compensated channel inputs. I tried both with precision voltage source and thermocouple sensor. However I have got wrong values. The sensor result was -0.5 or similar celsius degrees on room temperature. With precision voltage source, results were 25-30 degree higher than the real value.

When I use the calibrator input with a precision voltage source, the LTC2984 returns correct temperature values, but when I use the thermocouple sensor, it turns -0.5 degrees again. I configured this input without cold junction compensation.

What am I doing wrong?

  • Hi  ,

    The DC2212 does feature a cold junction compensation through a 2n3904 transistor. However, you would still have to configure the transistor as an Off-Chip Diode input. You would also have to configure the thermocouple channel reading to point to that specific transistor channel as a cold junction compensation.

    If you're using the LTC2984 TestBench Software (just to check your configurations, without the need for linduino), the configuration should look like the image below if you're using the J2 to connect your thermocouple to the board.

    If this doesn't work, would you please share your configuration with me? Thanks.


  • Thanks for reply,

    My configuration is same as yours. I already configured off-chip diode for 2n3904 and ideality factor around 1.003.

    I would also like to repeat my question in the case where the calibrator input is used. Why does the error occur when the calibrator input is used with real sensor?

  • Hi  ,

    Would you please enlighten me on your setup? What is this calibrator input?

    Is the calibrator input, just a channel configured as a thermocouple, without the cold junction compensation(CJC)? If it is, it would be possible that using a precision voltage source returns the correct temperature value due to the fact that it uses copper for its wires. Therefore the connections in the path from source to the LTC2984 is only copper-copper, removing the need for CJC. 

    When using a real sensor(Type J for example) however, the connections from the Iron to copper(DC2212A board), and constantan to copper(DC2212A board) produces unwanted voltages that introduces error to the measurement. This is why there is a need for cold junction compensation, and possibly be the reason why an error occurs when you're using a real sensor. 

    As for troubleshooting, have you tried measuring the voltage/temperature output of the 2n2904 transistor? It woud be possible that the thermocouple measurement is incorrect due to the fact that the CJC is not working properly.


  • Calibrator input is banana jacks of DC2212 board and I will also check 2n3904's measured values. 

    I want to learn how is working CJC in LTC2984. Does LTC2984 assumes thermocouple material always conducts with copper? I'm planning to use a connector that has BeCuAu contacts for sensor connection in my custom board, can I still use CJC? In addition, can soldering have a negative effect on CJC?

  • Hi,

    Regardless of the metal traces being used on your custom board, as long as both ends of the thermocouple are connected to the same kind of metal, and both junctions are of the same temperature, you could use the internal cold junction compensation calculation of the part. 

    Regarding soldering, it uses the same principle as your previous question. As long as both sides of the solder is of the same temperature, the solder should not introduce an error to your readings. However, this limits the temperature of your cold junction to the temperature your solder starts to flow or melt.