• State Resolved
  • Locked Locked
  • Replies 12 replies
  • Answers 1 answer
  • Subscribers 83 subscribers
  • Views 1721 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.

TMS320F28335: Questions on thermal metrics in the datasheet

Wayne Huang
Genius 17205 points

Part Number: TMS320F28335

Dear Champs,

I am asking this for our customer.

We use TMS320F28335PGFA.

From F28335 datasheet, it says the ambient temperature is 85C and junction temperature is 125C.

In field testing, because the ambient temperature is not evenly distributed, it's hard to measure the ambient temperature precisely.

Practically, we usually measure the top surface temperature of F28335 case with a thermal couple wire.

In TI doc "Semiconductor and IC Package Thermal Metrics", SPRA953, (www.ti.com/.../spra953c.pdf),

it says what matters is junction temperature rather than ambient temperature or top surface center temperature of the case.

It also shows how to estimate the junction temperature by the top center temperature in

TJ=TC+(ψJT×Power),

where ψJT varies with both PCB construction and air flow conditions as shown in Table 3.

However, in F28335 datasheet, SPRS439N, Table 5.7.1, P39

()

it only considers air flow condition but it does not consider or show any related to PCB constructions.

Therefore, we are confused how we choose ψJT because it does not consider PCB constuctions in the datasheet?

 

Also, if we choose ψJT is 0.74 as in the table, and DSP max power consumption is 900 mW and use the formula

TJ=TC+(ψJT×Power),

125 = TC + (0.74*0.9)

TC = 124.334

In this case, the junction temperature and package top (surface) temperature become very close.

Does it make sense?

Does that imply the package top temperature can be designed very close to junction temperature 125C?

 

 

Wayne

  • 0
    TI__Mastermind 42195 points

    Wayne,

    For the "A" and "S" grade parts we only guarantee the function of the device at ambient temperatures(A=85C and S=125C ). If you exceed that ambient temperature then the device may not function correctly, and will have a reduced lifetime.

    When finding PsiJT the board will play a factor, the numbers in the datasheet will assume a realistic application. Something like a soldered device with minimal heating from other nearby devices.

    From what I know, typically TC is close to TJ... of course TJ will always be higher. So I think your numbers make sense assuming you are running maximum power through the device.

    The package temperature can, under certain conditions, be close to the junction temperature.  

    Regards,
    Cody 

  • 0 in reply to Cody Watkins
    Prodigy 150 points
    "the package temperature can, under certain conditions, be close to the junction temperature"

    Could you explain in detail what kind of conditions it is possible to make junction temperature close to case temperature?

    Is it possible for me to use PsiJT to accurately calculate the case temperature when junction temp is 125℃?
  • 0 in reply to user5308703
    TI__Mastermind 42195 points

    Dengke,

    PsiJT will be constant for a set of conditions, so the more power the device consumes the closer the junction temperature will be to the case temperature. Each package can only dissipate a certain amount of power.

    PsiJT is our standard method for estimating the Junction Temperature. Yes, there is some inaccuracy in this method, details on this are included in SPRA953 which was linked to above.

    Regards,
    Cody 

  • 0 in reply to Cody Watkins
    Prodigy 150 points
    Cody,

    I have read SPRA953 carefully. This document points out that the value of PsiJT is related to both PCB construction and air flow conditions. However, in F28335 datasheet, it only considers air flow condition but it does not consider or show any related to PCB constructions. Can you tell me how to select the appropriate PsiJT value to calculate the junction temperature accurately through the case temperature?

    Regards,
    Dengke
  • 0 in reply to user5308703
    TI__Mastermind 42195 points

    Dengke,
    As you have stated we don't provide PsiJT for different PCB constructions, there are too many factors to provide this data.

    Select the PsiJT value that aligns to the airflow your system will have. That value will provide the best estimate.

    Regards,
    Cody

  • 0 in reply to Cody Watkins
    Prodigy 150 points
    Cody,

    Can you tell me the worst effects of the PCB structure on PsiJT?

    I want to know what the maximum value of PsiJT is under the worst conditions of the PCB structure and airflow, so I can calculate the upper limit of the case temperature in the worst conditions, which is of great significance to the thermal design of my product.

    Regards,
    Dengke
  • 0 in reply to user5308703
    TI__Mastermind 42195 points

    Dengke,

    For Airflow, as you have previously noted, TI publishes different Psi-JT values.

    For PCB structure there is not any published data. Additionally I do not have a mathematical model for how much PCB structure affects the measurement of Psi JT.

    There are too many factors to predict this accurately, remember that when calculating Psi-JT you are assuming that all of the power is being dissipated through the top of the device. This of course is not the case, so I have included some best practices in improving the thermal heat dissipation below.

    PCB heat dissipation factors:

    1. Spread out hot devices
    2. Continuous GND layers in PCB
    3. Thicker conductive PCB layers
    4. More thermal vias near/under the device
    5. Isolation of other hot devices from the GND plane

    External heat dissipation:

    1. Increased Airflow/ Venting/ Fans
    2. Heat sink
    3. Screws to chassis near the part

    Hope it helps,
    Cody 

  • 0 in reply to Cody Watkins
    Prodigy 150 points
    Cody,

    Thank you for the advice about improving the DSP thermal heat dissipation. it have a great help to me in the design stage of the product.
    However, my question is how to evaluate the thermal reliability of my product when my product has completed the design stage.
    As mentioned earlier, it is easier for me to use thermocouple to measure the case temperature on the top surface of DSP. please give me some advise about how can I use the case temperature to evaluate the reliability of my thermal design.

    thank you!

    Regards,
    Dengke
  • 0 in reply to user5308703
    TI__Mastermind 42195 points

    Dengke,

    1. Measure the temperature at the top of the device
    2. Measure the power being provided to the device on the 1.8V and 3.3V supplies.
    3. Calculate TJ based on "TJ=TC+(Psi-JT*Power)" the lower your TJ is the better!

    Long term reliability goes down the higher TJ is during operation. As long as you meet all recommended operating conditions in the Datasheet then the device will meet the lifetime specifications also stated in the Datasheet. 

    If you would like to know more about calculating useful lifetimes of embedded processors read this document.

    Regards,
    Cody 

  • 0 in reply to Cody Watkins
    Prodigy 150 points

    Cody,

    As you mentioned above, I still have  difficulties during the actual operation.

    1.Measure the temperature at the top of the device.

    Using a thermocouple to measure the surface temperature, this is easy for me to achieve.

    2.Measure the power being provided to the device on the 1.8V and 3.3V supplies.

    It is difficult to measure the power of 1.8V and 3.3V  very accurately, but I can evaluate it according to the maximum loss , which can leave a margin for my thermal design.

    3.Calculate TJ based on "TJ=TC+(Psi-JT*Power)" the lower your TJ is the better!

    Because Psi-JT is related to the structure of the PCB and cannot be specifically quantified, it becomes very difficult to obtain the specific value of Psi-JT. 

    The problem has returned to its starting point.

    How can I evaluate the junction temperature of the DSP when the case temperature can be measured?

  • 0 in reply to user5308703
    TI__Mastermind 42195 points

    Dengke,

    When using Psi-JT to verify junction temperature there are some inaccuracies. The best estimation for Psi-JT can be found in the Datasheet, use that number along with the equations previously mentioned to estimate your junction temperature.

    Are you getting unreasonable numbers with this calculation? In your original calculation you used the maximum of 900mW, is that correct? I suspect you would be at the limit of the package's current dissipation.

    Regards,
    Cody 

  • 0 in reply to Cody Watkins
    Prodigy 150 points
    Cody,

    The value of Psi-JT is not required to be precise, but it needs a reasonable range. However, I can't get the range.
    Because Psi-JT is related to the actual use of PCB, I know that it is very difficult to get it.

    I noticed that 28335 datasheet required the working temperature of the DSP not to exceed 85 degrees.
    Is it possible for me to use thermocouple line to measure the ambient temperature of DSP?
    And evaluate the thermal reliability of DSP through the actual operating environment temperature of DSP.
    Please give me some Suggestions for measuring the ambient temperature of DSP with thermocouple line.

    Regards,
    Dengke