Is it possible to write to a remote thermal sensor (Over temperature Value) from OS

Discussion in 'Embedded Systems and Microcontrollers' started by ekhoenix, Apr 7, 2016.

  1. ekhoenix

    Thread Starter New Member

    Apr 7, 2016
    12
    0
    So my inspiron 14r remote temperature sensor present on the motherboard is defective causing the laptop to thermal shutdown.
    The remote sensor chip ADM1032 which is connected to the Embedded controller IT8502E using SMBUS . The embedded controller can be communicated via LPCIO using 66/62 ports .Using RWeverything application i can see the EC RAM and write to the registers. My question is ..

    Is it possible to write the over temperature limit of ADM1032 from the OS (OS->EC-->Sensor)?
     
  2. viralvasa

    New Member

    Feb 24, 2016
    6
    0
    Probably not...! and there are also few reason for that.

    1. Inspiron Hardware when used with 66/62 port it has certain protocols for communication and to read write over that protocol requires authentication.
    2. Also for Hardware and Software security purpose external alternation might be blocked or secured one.

    My question is:
    How you came to know tha ADM sensor is defective ?
     
  3. ekhoenix

    Thread Starter New Member

    Apr 7, 2016
    12
    0
    The GPU temperature shown by the EC at start should not raise to 88 straight after cold boot right?. . Because this piece of shit ADM or the transistor inbetween the gpu sensor diode makes the temperature shoot upto critical temp(105C) thereby making the system do a thermal shutdown. Now i can only boot the machine by shorting out pins sdata and alert of the ADM--> Thats the only way i can make the fan spin at full speed at boot

    The GPU temperature read directly by chipset is normal.
     
  4. ekhoenix

    Thread Starter New Member

    Apr 7, 2016
    12
    0
    You mean a protocol is defined between EC and ADM?
     
  5. hp1729

    Well-Known Member

    Nov 23, 2015
    1,950
    219
    Is it the remote sensor that is bad or the ADM1032 itself?
    Is the ADM1032 accessible and replaceable? 8-pin? SMD?
    Dirty remote sensor?
    Bad cap in the remote sensor circuit? There's usually a cap across the sensing transistor.
    (Just considering all the possibilities.)
     
    Last edited: Apr 11, 2016
  6. ekhoenix

    Thread Starter New Member

    Apr 7, 2016
    12
    0
    Is it the remote sensor that is bad or the ADM1032 itself? i am not sure ... which one is .Do u have any suggestions to check
    Is the ADM1032 accessible and replaceable? 8-pin? SMD? yes its 8 pin and i dont plan on soldering since im an amateur at that.
    Dirty remote sensor? does that actually cause it to raise temperature?
    Bad cap in the remote sensor circuit? There's usually a cap across the sensing transistor. --possible! - what are the symptoms of a bad cap?
     
  7. viralvasa

    New Member

    Feb 24, 2016
    6
    0
    There can be one possible solution :
    1. You can isolate the ADM and let it be the open loop system. It means either you cut the track of pcb going towards ADM (not ideally recommended) or you desolder it. By this way there will be no values to write on EC RAM.

    Yes there is always a protocol between EC & ADM and EC & OS and to identify that its really complex task. I can't help you regarding this protocol. but you can try above solution.
     
  8. ekhoenix

    Thread Starter New Member

    Apr 7, 2016
    12
    0
    You mean cut off the vcc pin going to the ADM right? Well ADM has a thermal unmaskable interrupt which goes high during normal operation and is connected to the system power chip directly... So if it goes low the system shuts down.. Its not the ec that does the thermal shutdown
     
  9. hp1729

    Well-Known Member

    Nov 23, 2015
    1,950
    219
    Without soldering skills you are pretty much limited to just cleaning. How dusty is it inside the computer? I don't know about the programming option. I've never considered that before. Hardware problems meant hardware solutions to me. It sounds like you know what you are doing system and software wise. Try it and let us know what you find, please. You can't be the first or last to see this problem.
     
    Last edited: Apr 11, 2016
  10. hp1729

    Well-Known Member

    Nov 23, 2015
    1,950
    219
    Cutting the trace on the board sounds like a bad idea. Cut the lead of the ADM1032 where power goes, pin 1, or the Alert line, pin 6, if you are going to do this at all. Regarding the Alert line, it is active low, as you say, but if it is not pulled low by the chip there is probably a resistor on the board to pull the line high. Standard SMBus design. I suspect the line would default to a high.
     
    Last edited: Apr 11, 2016
  11. ekhoenix

    Thread Starter New Member

    Apr 7, 2016
    12
    0
    There is another interrupt THERM apart from SMBUS alert that does the thermal trip directly .From the datasheet it is active high during normal operation..Thats what stopping me from doing the cutting.

    please have a look at the images attached

    So if cut the pin1 im afraid the system wont even boot up
     
  12. ekhoenix

    Thread Starter New Member

    Apr 7, 2016
    12
    0

    the back side of the vga gba has a mechanical interface to support the heatsink.. it is the only place i see little amount of dust. Does the sensor lie on the back side of the vga ?
     
  13. hp1729

    Well-Known Member

    Nov 23, 2015
    1,950
    219
    You make a good call then. Take it to somebody who has the part and can solder. You still can't tell if it is the remote sensor or the ADM chip itself?.
     
  14. ekhoenix

    Thread Starter New Member

    Apr 7, 2016
    12
    0
    nope i dont know which one is .. i dont know a way to identify. i have tried shorting d+ and d- and sdata and alert . the first case causes the adm to show a temperature reading of -128(0x80) . the second case makes my EC to spin fan at full speed. this is how i can run the system right now.So after the system boots i set the BIOS fan control of the EC to manual through an application . If i play a game for some time the ADM chip does the thermal stop directly even when the fan is at full speed and GPU direct temperature is only at 55C.

    Im just goin to try the first case -keep shorting the d+ and d- while the game runs to see if it makes any difference

    If it doesnt do thermal shutdown while im shorting d+ and d- then the remote sensor is the problem right? if thats the case i ll just use some wire or something to connect those 2 pins for quick fix
     
  15. ekhoenix

    Thread Starter New Member

    Apr 7, 2016
    12
    0
    So apparently shorting d+ and d- during system run blocks the thermal shutdown by the retarded adm . so which one is defective remote sensor /adm or the cap/resistor in between them
     
  16. ekhoenix

    Thread Starter New Member

    Apr 7, 2016
    12
    0
    ignore the above msg.. it still shuts down .. so its adm that is defective
     
  17. hp1729

    Well-Known Member

    Nov 23, 2015
    1,950
    219
    Time to take it to a tech?
     
  18. ekhoenix

    Thread Starter New Member

    Apr 7, 2016
    12
    0
    Nope got it fixed by shorting d+ and d- this time tightly :D. so its bad remote sensor. Tech here are very costly .. i would get a new board for that cost
     
  19. hp1729

    Well-Known Member

    Nov 23, 2015
    1,950
    219
    Super!
     
  20. viralvasa

    New Member

    Feb 24, 2016
    6
    0
    that's wonderfull ekhoenix... this is the beauty of this forum.. Good luck ahead..!
     
Loading...