Forum Rule: Always post complete source code & details to reproduce any issue!
Results 1 to 25 of 25

Thread: Teensy 4.1: CPU Temperature 167 degrees

  1. #1
    Senior Member
    Join Date
    Jun 2017
    Posts
    140

    Teensy 4.1: CPU Temperature 167 degrees

    I have a Teensy 4.1 that is placed in a metallic enclosure outdoors.

    As expected, during daytime, the Teensy's 4.1 temperature (as measured by the temperature library) is quite high but I did not expect it to be 167 degrees on a hot day. The rest of my circuit's components can withstand up to 175 degrees (ambient).

    The Teensy 4.1 has been working fine and no instability was present.

    I realize that I can reduce its clock speed (< 600) but I need maximum performance for my application. I also have built into my code to de-throttle the processor but I set the temperature at over 175 degrees.

    Would this operational scenario eventually destroy the Teensy?

    Thanks in advance for the community's input.

  2. #2
    Senior Member+ defragster's Avatar
    Join Date
    Feb 2015
    Posts
    14,836
    NXP publishes a doc on temp effect on expected lifetime - there was a post some months back with discussion on it.

    Is the T_4.1 the primary heat source in the box or are there other heat makers in there too?

    If 528 MHz would be enough that drops core voltage measurably.

  3. #3
    Senior Member
    Join Date
    Jun 2017
    Posts
    140
    The T_4.1 is the primary hear source. There are a couple of other sensors but their impact should be very small.

    I can try to lower the frequency to 528 and evaluate its impact and will read into how this can be done.

    Thank you.

  4. #4
    Senior Member
    Join Date
    Jul 2014
    Posts
    3,316
    Quote Originally Posted by jimmie View Post
    I have a Teensy 4.1 that is placed in a metallic enclosure outdoors.
    I would avoid to have anything in metallic enclosure with direct sunshine (OK it will debug your system)

  5. #5
    Senior Member
    Join Date
    Jun 2017
    Posts
    140
    I agree, a plastic enclosure is better but it is what was called for so I have no control.

    I will change the frequency and maybe also add a heat sink. Any recommendations for a specific heat sink? How does one connect the heat sink to the Teensy as I am concerned about double-sided tape becoming loose especially that the Teensy is placed vertically.

    I also found this link:

    http://forum.pjrc.com/threads/54711-...a-Test/page173

    [Alphacool 17426 GPU RAM Copper heatsink] glued to the T4 processor with thermal glue [Silverbead Thermal Glue Adhesive SG100X].

  6. #6
    Senior Member
    Join Date
    Jul 2014
    Posts
    3,316
    IMHO, it is not the enclosure per se, but external heat.
    If you cannot cool the box, a heatsink on processor is useless, as the box inside will heat up.
    plastic may even be worse, as it is bad heat conductor.
    not knowing the application, invent a system that avoids external heat (shield) or removes heat from box (water, air, external heatsink)

  7. #7
    Senior Member
    Join Date
    Jun 2017
    Posts
    140
    Yes, you make good points. It may have to be a combination of an enclosure fan, and a processor heatsink.

  8. #8
    Senior Member
    Join Date
    Jul 2020
    Posts
    1,178
    That gave me a shock till I realized you mean Farenheit. Electronics uses degrees C, read any datasheet.
    75 degrees C doesn't sound too bad. I mean you'd like it to run cooler if possible, but most chips are rated
    -40 to +85C or better.

    I reckon the most likely failure model is through thermal cycling stressing the ball-grid array solder balls, eventually
    fatiguing them. This is a known issue with BGAs compared to leaded chips which have some flex in the leads.

    So the number and severity of thermal cycles will likely affect the lifetime, but what that means in years or decades
    I don't know.

  9. #9
    Senior Member
    Join Date
    Jun 2017
    Posts
    140
    Thank you MarkT.

    Yes, I meant Fahrenheit and should have appended the measurement unit to the title.

    I am less concerned about thermal cycling because the application runs 24/7 and the temperature range between cold and hot weather operation is only about 40 degrees Fahrenheit.

  10. #10
    Senior Member
    Join Date
    Jul 2020
    Posts
    1,178
    Quote Originally Posted by jimmie View Post
    I am less concerned about thermal cycling because the application runs 24/7 and the temperature range between cold and hot weather operation is only about 40 degrees Fahrenheit.
    Well the diurnal variation can be 20C in temperate areas and 60C in a desert,
    and a summer thunderstorm is a large and rapid cooling event in a metal enclosure.

  11. #11
    Senior Member
    Join Date
    Jun 2017
    Posts
    140
    True, there can be large variations due to day/night and thunderstorms.

    The application is in an urban environment. Summer thunderstorms do lead to a rapid reduction in temperature of about 15C but I think fatigue caused by thermal cycling requires a higher temperature difference. I do not know the number but it is higher than 20 degrees .....

    Eventually, of course the board will fail, I am just hoping that it lasts > 5 years.

  12. #12
    Senior Member
    Join Date
    May 2015
    Location
    USA
    Posts
    1,048
    A heatsink stuck to the T4 processor will make it run cooler, even inside of a sealed box.

  13. #13
    Senior Member+ mjs513's Avatar
    Join Date
    Jul 2014
    Location
    New York
    Posts
    7,409
    Guess think of it like the inside of a pc - CPU has a heat sink plus fan then you also have the case fan to get the heat of the case.

  14. #14
    Senior Member
    Join Date
    Jul 2014
    Posts
    3,316
    Quote Originally Posted by jonr View Post
    A heatsink stuck to the T4 processor will make it run cooler, even inside of a sealed box.
    depending on the thermal properties of the box, if no heat is removed from the box, a sealed box will keep all heat inside, and at certain point processor will overheat, right?

  15. #15
    Senior Member
    Join Date
    Apr 2014
    Location
    Germany
    Posts
    1,585
    if no heat is removed from the box, a sealed box will keep all heat inside, and at certain point processor will overheat, right?
    That is of course correct in principle. However, due to the large surface of such a box, I would say it won't be easy to thermally isolate it good enough that you'll get a significant increase of temperature inside inside the box. If I remember my Thermodynamics classes correctly (which might not be the case :-) ) one can do the following rough estimation (neglecting radiation and convection effects)

    Assume e.g. a 100mm cubic aluminum box with a wall thickness of 3mm

    Box surface: A= 6*100mm*100mm = 0.06 m^2
    Thermal conductivity Al: k=200 W/(m * K)
    Thermal Resistance R_th = l/(k*A) = 0.003m / ( 200 W/(m*K) * 0.06 m^2) = 2.5E-4 K/W

    If the processor produces a power of say P = 500mW one gets a temperature difference of delta T = R_th * P = 2.5E-4 * 0.5 = 125K between inside and outside the box.
    If you do the same calculation for a thermally isolating material like plastic (k ~0.2 W(m*K) you'd get delta T = 0.125K which is still irrelevant.

    Thus I'd say that the thermal properties of the processor heatsink will largely dominate the properties of the box.

  16. #16
    Senior Member
    Join Date
    Jul 2014
    Posts
    3,316
    @luni,
    now put all in plain sunlight
    I would be curious to see if and when the 85 decC will be exceeded.
    (OK, this may not be the case of OP, but for the sake of argument)
    Also, I guess you assumed perfect external heatsink (otherwise you would generate an isolating air layer around the box)

  17. #17
    Senior Member
    Join Date
    Apr 2014
    Location
    Germany
    Posts
    1,585
    now put all in plain sunlight
    I would be curious to see if and when the 85 decC will be exceeded.
    That's of course a different story. In this case the thermal equilibrium would be defined by the radiating properties of the box I'd assume. Might be fun to do some radiation estimations. On the other hand general life experience shows that cars do heat up considerably in the sun :-). So the gain in knowledge might be small :-)

  18. #18
    Senior Member
    Join Date
    May 2015
    Location
    USA
    Posts
    1,048
    > depending on the thermal properties of the box, if no heat is removed from the box, a sealed box will keep all heat inside

    But a perfectly insulated box doesn't exist. And since the box is relatively large, there is significant heat flow through it.

    On the other hand, always consider the surface air films when calculating heat flow.

    Ideally, the processor heat sink would be connected directly to an aluminum case. But this is not necessary in this case.

    If the box will sit in the sun, you want it to be painted white (high reflectance and high emissivity).

  19. #19
    Senior Member xxxajk's Avatar
    Join Date
    Nov 2013
    Location
    Buffalo, NY USA
    Posts
    570
    Actually, paint white on the top, black on the bottom, yes, that's really a thing.

  20. #20
    Senior Member
    Join Date
    May 2015
    Location
    USA
    Posts
    1,048
    Agreed, black in areas not in the sun would radiate about 5% better than white. Shiny aluminum has poor radiant performance.

  21. #21
    Senior Member
    Join Date
    Apr 2014
    Location
    Germany
    Posts
    1,585
    Quote Originally Posted by jonr View Post
    Agreed, black in areas not in the sun would radiate about 5% better than white. Shiny aluminum has poor radiant performance.
    That is certainly true, but how much power would radiate from the box at about 25C? Radiated energy is proportional to T^4. I may be wrong but I'd say at those low temperatures the dominating mechanism to transfer heat out of the box is conduction not radiation. So I'd go for the common sense solution and paint it white or wrap the housing in some reflective aluminum foil to reduce the amount of energy radiated into the box and rely on conduction to transfer heat out of the box.

    Another quick estimation for the fun of it:

    The maximum solar power density at noon in summer in Athens is about 900 W/m^2. (https://www.researchgate.net/figure/...fig2_319864978)
    Using the numbers from #15 the sun radiates 900W/m^2 * 0.01m^2 = 9W onto one side of the box. At a reflectivity of about 80% only 1.8W would enter the box. Given the low thermal resistance of the aluminum box (see #15) those 1.8W shouldn't heat it up significantly even if you would/could prevent out radiation completely. But again, might be completely wrong, just playing with numbers...

  22. #22
    Senior Member
    Join Date
    May 2015
    Location
    USA
    Posts
    1,048
    > the low thermal resistance of the aluminum box

    Agreed, but the conductivity of the air films is critical to the correct calculation of conduction from a free standing box. Maybe k=4, about 1/50th of what was calculated in #15?
    Last edited by jonr; 07-24-2021 at 04:40 PM.

  23. #23
    Senior Member
    Join Date
    Apr 2014
    Location
    Germany
    Posts
    1,585
    At k=4 a heat source of 1.8W inside the box would generate a temperature rise of about 20mK if my estimations are correct. Spreadsheet attached.

    (Assuming k=4 for the thin air film it would be the dominant thermal resistance. The small resistance of the aluminum can then be neglected of course). A more correct calculation would take convection into account. But since that would lower the resistance even further I'd say it is not worth the hassle.

    Edit: I assumed 3mm for the thickness of the air film which sounds way to much. What would you expect?

    Might place a box with a thermocouple into the sun tomorrow to check :-))
    Attached Files Attached Files

  24. #24
    Senior Member
    Join Date
    Jul 2014
    Posts
    3,316
    Quote Originally Posted by luni View Post
    Might place a box with a thermocouple into the sun tomorrow to check :-))
    Hopefully we get sun tomorrow (PLZ 93xxx)

  25. #25
    Senior Member
    Join Date
    May 2015
    Location
    USA
    Posts
    1,048
    It's been awhile, so you are providing some good practice.

    Looks like air films are normally specified as conductance "C" in W/m^2 K, not conductivity "k" (as I wrote above). So it doesn't need an adjustment for thickness.

    1 / ( 4 W/m^2 K * 0.06 m^2) = 4 K/W or 4C warmer with 1W of heat. Noticeable, but generally not significant.

    Small holes in an outdoor case are important. Not because of the small increase in convection cooling but because of the danger of moisture accumulation. Too big and bugs become a problem.
    Last edited by jonr; 07-24-2021 at 06:08 PM.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •