0
\$\begingroup\$

While trying to get temperature values from my thermocouple amplifier, I am able to read the cold junction temperature, yet the hot junction temperature isn't even measured by the converter. Reading the status register and measuring the OC output pin unambiguously indicate, that my K-type thermocouple isn't connected (STATUS bit 4), no matter if it is or not, and the measurement is constantly being indicated as "not ready" (STATUS bit 6). As the datasheet suggests, I measured the input voltages. When the TC is disconnected, the voltage should be 0.66*Vdd on the positive input Vin+ and 0V on the negative input Vin-. When it's connected, both of them should have a common potential of 0.5*Vdd. What's surprising, in my case it is constantly 0.5*Vdd, which suggests that the TC should be detected as constantly connected. I've measured resistance dividers values (correct) and input capacitance between Vin+ – Vin-, Vin+ – Vdd and Vin- – GND. All of them seem to be around 20-50 uF, but even taking into account the pretty big RC time constant, after disconnecting the TC, the voltages don't seem to drift toward mentioned values after a minute or so. What I found in the datasheet is that the circuit's system ground and regular ground don't have to be physically shorted, whereas in my circuit they are.

The GND pin is the system ground pin. Pins 1, 3, 5, 13 and 17 are system ground pins and they are at the same potential. However, pins 6, 7, 9, 10 and 18 must be connected to ground for normal operation.

Could that possibly be the reason for such behaviour? edit I don't think so, because the suggested layout shows them shorted as well. Reading raw ADC data (despite the result being indicated not ready), clearly shows that blowing with hot air on TC causes the ADC register to return consequently higher values. It would mean, that both Vin+ and Vin- pins are soldered as they should be. Link to the datasheet I use.

Described schematics: described circuit

\$\endgroup\$
1
  • \$\begingroup\$ I also suffer with this IC. MCP9600 was my primary goal, but you can't get it anywhere. This parts shortage is a disaster. Someone please correct me if I'm wrong about the following: ADC is the differential. It has a 100nF on it. But it is still differential. And Vsense is not. That is, whatever noise there is in the system, Vsense will measure it with respect to the ground. I put an Isolated Scope on the Vsense connector. The mains 50Hz sine wave looks beautiful. 1-5V peak, God knows how, somehow it is capacitively coupled there. The situation is even worse when I touch the thermo sensor. The \$\endgroup\$ Commented Nov 28, 2022 at 19:56

3 Answers 3

2
\$\begingroup\$

Alright, I found the source of the problem. In the errata to the linked datasheet, I found a similar issue with OC detection while TC is disconnected. I dug deeper, and I found out that it doesn't concern versions Rev. 0x4111 and upper. I checked my device ID saved in its register and it was exactly 0x4111. But OC Alert unwanted changes when TC is disconnected weren't my problem - contrarily, my OC Alert never changed. What else could be wrong? I checked the suggested circuit - resistors values were changed and the OC detection threshold was modified as well. Even though my common voltage was then about 2,3V (47% of used Vdd), it was way above the threshold specific to my device's version which was 19% of Vdd. After the replacement of resistors (actually, after building ridiculous blocks and trains out of them because of a lack of required values), everything started to work properly. Thanks to everybody who tried to think about the solution.

\$\endgroup\$
1
  • \$\begingroup\$ Yes, the 'F' datasheet you linked has the updated resistor values. Errata is here \$\endgroup\$ Commented Nov 13, 2022 at 18:08
0
\$\begingroup\$

I don't think your assumptions on input voltages are nearly correct. The inputs have a differential input impedance of ~300kΩ and there seems to about 1uA of bias current per input near ground (the datasheet is silent on bias current as far as I can see). So your (-) T/C input would be at more like 2V when the T/C is open. Common-mode input impedance is given as 25MΩ typical.

The MQFN package is not easy to solder or inspect with 100% confidence, I suggest that your (+) input pin (chip pin 2) may not be soldered properly.

The paragraph you copied from the datasheet is a bit confusing and appears to apply only to the MCP9600.

\$\endgroup\$
2
  • \$\begingroup\$ These are not my assumptions, but rather what is written in point 6.3.3.1. However, you're right - the voltage on Vin- pin, while TC is disconnected, is 2.03V right after disconnecting, and the voltage on Vin+ and Vin+, while it's connected at room temperature, is 2.34V. Of course, I also tested voltage at disconnected TC and it seems to react correctly to temperature change, so it's in fact not open. Looking under the microscope, each pin seems to be properly soldered. What's more, I checked if Vin+ is shorted to any other pin - it's not. \$\endgroup\$
    – drydre
    Commented Nov 13, 2022 at 9:31
  • \$\begingroup\$ In the above comment, I meant "voltage on Vin+ and Vin-, that's a typo. What's even more peculiar, reading raw ADC data (despite the result being indicated not ready), clearly shows that blowing with hot air on TC causes the ADC register to return consequently higher values. It would mean, that both Vin+ and Vin- pins are soldered as they should be. \$\endgroup\$
    – drydre
    Commented Nov 13, 2022 at 10:20
0
\$\begingroup\$

I managed to get this information from Microchip. For some reason the datasheet version I had did not have this updated information.

Bias resistors for MCP9601/L01/RL01

\$\endgroup\$

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.