20°C deviation of CPU temperature between different SoMs

[Kimych]

Dear Support Team,

We are producing a series of measurement devices based on Colibri VF50 IT using our custom baseboard. Suddenly we found that some SoMs have significantly (like +20 °C) higher CPU temperatures than others. Fortunately, there are just some of them (I have 3 modules right on my table).
Our assumption that we can have a HW problem on our custom carrier boards or… it may be related to Colibri VF50 SoMs themself (but I do not think so).
Maybe you can give me some advice on how I can find a reason for higher CPU temperatures?
Here a list of SoM S/Ns a and according CPU temperatures, measurements are done using exactly the same carrier board and exactly the same FW and configuration:
07283226 45°C *
07283223 45°C *
07283224 44°C *
07282922 41°C *
07282923 45°C *
07282924 43°C *
07282929 44°C *
07282927 44°C *
07283227 44°C *
07283230 43°C *
dad one:
06923351 59 °C
06923417 62 °C
06922473 61 °C
Here some pictures:

Thanks in advance

Colibri VF50 128MB IT V1.2B
Colibri Evaluation Board V3.2A
Custom Linux and U-Boot image based on console image Linux BSP 2.8b6

[alex.tx]

The observed temperature variation is due to differences in IC structure that occur during chip manufacturing. According to the VF50 datasheet, the maximum SoC power consumption can be up to two times greater than the typical value. Since the temperature of the ICs you’ve observed is within the operational limit in all cases, there’s no cause for concern

[Kimych]

Thank you @alex.tx for your answer. But still such a big difference looks a bit “strange” to me. We will proceed with additional tests and if I find something then I’ll come to you again.