• State
  • Locked Locked
  • Replies 1 reply
  • Subscribers 31 subscribers
  • Views 576 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.

AM2634-Q1: heat dissipation and internal temperature of microcontroller

Ugo Cesari
Prodigy 10 points
Part Number: AM2634-Q1
Other Parts Discussed in Thread: TMDSCNCD263, AM2634

We find a difference of about 10°C under the same environmental and operating conditions (same FW loaded) between the internal temperatures as read by SW via the internal temperature controller  between the components mounted on our PCB (about 52°C on all 4 internal sensors) and the component mounted on its EVB TMDSCNCD263 (42°C, Tamb=20°C). our PCB and EVB have similar areas but different form factor (our PCB is rectangular with short side around 5 cm, the uC is approximately in the center of the PCB).

Checking stackup and layout regarding size and connection towards component power planes and after proper debugging of our board to exclude bus contention eventualities and other amenities that could have overloaded the Sitara, none of the above reasons seems to give an explanation.

I therefore try to ask if there could be an explanation related to internal silicon review and/or activated functionalities, because these components (being  also prototipal part numbers) are from different production batches.

Specifically, looking at the marking reported above the components, they are as follows:

  • EVB: XAM2634B, OLFHMZCZQ, Lot 21ACLSW G1, Prod Code 548 ZCZ
  • our PCB: XAM2634B, OLFHMZCZ, Lot 23C9YQW G1, Prod Code 548 ZCZ

Looking at the component revision and feature codes the two components would appear to be the same, however as mentioned, we find the aforementioned internal temperature difference; a curious thing, is also that the 4 sensors return a very similar temperature ( 1-3 degrees difference), at least at steady state, despite 2 cores being engaged in processing and, at present, 2 being almost idle; but this could of course be due to a particularly thermally conductive intermediate layer that contributes to equalizing temperatures across the silicon...

  • 0
    TI__Mastermind 22685 points

    Hi Ugo,

    Heat dissipation for a performance MCU (like the AM2634) comes down to several factors in the PCB.  The first is how efficiently heat can be transferred away from the MCU to the rest of the board, and the second is how well the rest of the board acts as a radiator for dissipating that heat to the surrounding environment.

    Heat is transferred away from the MCU using the thermal vias, which for the AM2634 double as the ground vias underneath the center of the package.  The more GND vias you have placed in the central ground area and the larger diameter those vias are, the more copper there is to help transfer heat down into the PCB from the package.

    Once the heat has transferred down to the PCB, it transfers out to the ground planes, which act as radiators, dissipating heat out of the PCB into the environment.  The more solidly filled the ground plane(s) are near the MCU, the better the heat will spread out over the ground plane.  And the larger the ground planes and the more planes there are (to an extent), the better the board will act as a radiator to dissipate the heat.

    The TMDSCNCD263 uses a single 18mil OD, 8mil ID via per BGA pin, which makes a 1:1 ratio of vias to ground pins, which is very efficient in transferring heat away from the MCU to the ground planes.  The TMDSCNCD263 also has a 10 layer PCB, with 4 ground planes (layers 2, 4, 7, and 9), which are minimally chopped up near the MCU, maximizing the radiating performance of the TMDSCNCD263.

    Can you compare your PCB design to the TMDSCNCD263 with regards to vias to ground pins ratio underneath the MCU, number of PCB layers and ground layers, and how chopped up the ground layers are near the MCU?  If there is a reduction in any of these, I would expect poorer thermal dissipation on your custom PCB, leading to the higher temperatures you are seeing.

    Thanks,
    Mike