# Thermal resistance of Orin NX

Continuing the discussion from Thermal resistance of Orin NX:

I also have a difference in the thermal resistance value and Tc value of Orin NX 8GB.
The Tc measured with a thermocouple is about 80°C.
From these results, the calculated thermal resistance is approximately 1.33°C/W.
Why is it different from the TDG value?
power_GUI.pptx (638.3 KB)

Hi, as you can see in TDG, "Thermal model for Orin package is based on “Uniform heat loading of die.” In practice, the die will be non-uniformly loaded depending on the type of workload running on the die. Designers must account for adequate margin when designing a thermal solution. And thermal specifications are preliminary estimates…”. As per your test result, it looks like the test time is not enough since 101°C is much higher than 80°C. Is there any thermal solution attached? Maybe you can run test for more time to check if the final Tc is still 80°C.

Since the measurement results were taken after 3 hours or more, the measurements were taken in a state of thermal equilibrium.
Is the temperature sensor reading okay?
Is there anything I can check for myself?

Is it possible for the thermal resistance to be several times higher when not in equilibrium with the same Jetson AGX Orin module?

Jetson_AGX_Orin_Series_TDG-10943-001_v1.2.pdf (nvidia.com)

The Maximum Orin SoC operating temperature is 99°C, I don’t think your test with 101°C is reasonable. Can you share your test setup photo?

Result_20230809.pdf (1.5 MB)

In order to abide by Orin SoC Hardware Thermal Throttling, it was put into a constant temperature bath and re-tested.

T_C(T1) measured with a thermocouple is 68.2°C.
Orin VDD_IN is 23W.
From the above, the thermal resistance R is as follows.
R=(82.5℃-68.2℃)/23W=0.622℃/W

There is a deviation of about 4 times from the TDG value.

Hi, this time the test setup looks more reasaonable. For more accuracy, you may need to find out the hotest point on case and list the room temperature and air flow speed. But anyway as the early comment, the real Tjc could be different to the value in TDG, as the thermal model for Orin package is based on "Uniform heat loading of die.” In practice, the die will be non-uniformly loaded depending on the type of workload running on the die. Designers must account for adequate margin when designing a thermal solution. So for custom thermal solution design, customer should run their real use case and then get the thermal resistance of solution accordingly.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.