• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 30 subscribers
  • Views 842 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

LM86 Temperature Sensor Accuracy

Scott Robinson1
Prodigy 30 points
Other Parts Discussed in Thread: LM86, LM95235

Could you please provide whatever compatibility information (including offset register values) regarding the use of the LM86 Temperature Sensor with the Intel Celeron M (ULV 423, 1.06 GHz) processor thermal diode that is available? 

We’re trying to understand what the expected accuracy limits are for this system just considering the ideality / non-ideality factor for the thermal diode shown on page 66 in Table 19 of the attached.

(This was originally sent to the National Semiconductor group, hardware.monitor.team@nsc.com)

Thank you in advance,

Scott

Celeron M ULV 423 Datasheet 31272604.pdf
  • 0
    TI__Mastermind 37160 points

    Hi Scott,

    I am investigating on this, and I will get back to you sometimes tomorrow.

    Aaron Heng

  • 0 in reply to Aaron Heng
    TI__Mastermind 37160 points

    Hi Scott,

    In the future, if you have any questions regarding temperature sensor products, please post your thread in the Temperature Sensor Forum so that we can monitor your post in a timely manner.

    We have never characterized the Intel Celeron M 65nm processor for the LM86. I will get back to you if you can use the LM86 with Intel Celeron M 65nm process.

    Another route is to use the LM95235. The LM95235 is calibrated to support for Intel Celeron M 65nm process.

    Aaron Heng

  • 0 in reply to Aaron Heng
    Prodigy 30 points

    Hi, Aaron and thanks for looking into this.

    Unfortunately, this is an off-the-shelf board that we've bought from another manufacturer. It's an older design and isn't easily modified to use a different temperature sensor. I don't know how they came to choose the LM86 over the LM95235, my only guess would be that it allowed them to keep the same driver software from previous products.

    It would be very helpful if you could provide any detailed information on why the LM86 is "not usable" with the Celeron M. For example, the range of values for the ideality factor is too great and the theoretical range of accuracy is -2 to +25 degrees C. Correspondingly, any guidance on register values which could help to improve the accuracy would also be appreciated.

    Scott

  • 0 in reply to Scott Robinson1
    TI__Mastermind 37160 points

    Hi Scott,

    I just want to clarify that the Intel Celeron M is possible to work with LM86, but we have never characterized the data. I will look into this and get back to you.

    Aaron Heng

  • 0 in reply to Aaron Heng
    TI__Mastermind 37160 points

    Hi Scott,

    Based on the typical diode ideality factor value and series resistance of the Intel Celeron M 65nm processor, I am able to calculate the offset temperature to be 2.4°C for the LM86. However, you need to read the temperature offset value from a processor model specific register (MSR). This value will be used to subtract 2.4°C from the MSR value to avoid the ideality spread wider before you can program into LM86 offset register to compensate narrow spread and accuracy. If the driver is not already using the values from the MSR register you will need to change your driver to do so, or as you calculated your errors will be esteemed.

    Do you use the status and configuration register of the LM86? I went through the LM86 and LM95235 datasheet side-by-side; the local and remote temperature registers are similar. The status and configuration registers are a bit different. If you are not using these registers, then you can replace the LM86 with LM95235 without worrying about the ideality factor as the LM95235 is tuned for the Celeron M.  The LM95235 should take care of that for you. Also your driver will not need changing.

    You mentioned that you got the board off the shelf, which means that your current driver should be working; unless, you are replacing the old processor with the Intel Celeron M 65nm process. What kind of processor did you use before? Can you provide me the schematic? I hope you can provide me more information so that I can understand your setup.

    Aaron Heng

  • 0 in reply to Aaron Heng
    Prodigy 30 points

    Good morning, Aaron,

    Thanks for the information. I'll have to check with the software group to see how the registers are being used. I don't believe that the configuration register is being changed from its default value other than to control the external alarm bit (threshold?). The offset you mention makes sense but I'll have to go back to the datasheet to understand the right way to use / apply it. It sounds like having this offset configured properly will help things but will it limit the "worst case" effect of an improper ideality factor?

    Using the LM95235 is not an option for us - soldering is not allowed, it voids the warranty...

    Previously (about ten years ago?), we had also used a board from a third-party and the processor used in that case was a Pentium III so there was little if any need to adjust the LM86 registers. Since it was a third-party off-the-shelf board, we don't have the schematics (and even if we did, sharing them couldn't happen without NDAs among all of us, I would think).

    Scott