This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LM92: Massive Temp Sensor errors at -40degC

Part Number: LM92
Other Parts Discussed in Thread: LM76, LM77

We are experiencing massive Temp Sensor errors at -40degC using LM92. So far it is happening on two boards out of two and we are at a very urgent stage of our project.

The temp data is polled often and when it goes wrong around -40degC it looks like the following:

-36.01,  -36.01,  -36.01,  -36.01,  -12.01,  -76.01,  -76.01,  -76.01,  -226.89,  -38.12

Does this look like an I2C communications problem?  How far can I2C be used over a ribbon cable, ours is 17 inches? should pull up or pull down resistors be used? Are there any things we can do to improve digital error margin?

Are there pin compatible devices which might work better at -40degC? or have better I2C operation? Can we change the I2C data rate? (I am not a digital designer)

Why does the datasheet only give data down to -25degC? Are we exceeding the range?

Thanks,

Chris

  • Hi Chris,

    LM92 is a long-time released device from National's folks that were no longer with us. The operating range for the temperature is from -40°C to 150°C; however, we only specified the temperature accuracy is from -25°C to 150°C. Anything out beyond this specs, we will not gauranteed.

    I reviewed the characterization data. I don't see this device has a big gap of temperature error across temperature and voltage. It looks like that you have a lot of noise coupling to the I2C signal. Could you please capture the scope photo when you see the bad reading temperature goes beyond -36°C?

    Aaron

  • Hi Aaron,

    I will try to do so but it may not be possible due to integrated board and system design.

    Could you please advise if there is maximum cable length for I2C?

    Thanks

    Chris

    PS could you please change my name shown on these posts to make it anonymous?

  • Could you please change my name to anonymous.

    Also when you state the error is 1% what is 1% of 0degC?

  • Hi Chris,

    Unfortunately, I do not have any IT privilege to change user's account setting. I think you might be able to change the setting in your user profile.

    We don't normally specify the temperature accuracy in percentage. All temperature sensors accuracy are specified in °C based on the characterization data.

    I don't have the maximum cable length for this particular device, but most of our temperature sensors in TMPxxx tested less than 1m. With the length you currently have on your setup. I don't see it is an issue.

    Hope this helps. Let me know if you have further questions.

    Aaron

  • Sadly no way of switching to and remaining anonymous at this end.

    Suspecting cross talk we were able to achieve some improvement by splaying the wires of the ribbon cable.
    Would you be able to recommend a new pin compatible version of this device which might have more robust I2C signals and have better accuracy at -40degC?
  • Hi Chris,

    If you want to change your display name, here are the steps.

    1. On the top right corner, there is a human icon profile. When you hoover the mouse to the icon, you should be able to see your name.
    2. Then, you click on your name. Afterward, it will bring you to the profile page then simply click on the hyperlink "About(edit)".
    3. This enters to the E2E preference setting. You should be able to change your nickname under the "settings" section.

    Regarding to the temperature sensors P2P compatible, the closest device are LM76 and LM77. Again, these are not specified at the -40°C temperature. The only way to track where the extra bits are coming from is to monitor the I2C communication of SCL and SDA pins. Other test you can do is by shielding the I2C signals.

    Please let me know if you can perform test above.

    Aaron

  • Thanks Aaron, The splayed cable seems to be good enough. I think crosstalk was causing signal degradation.