HI Board designers, 

Refer below information on the VTM reset.

6.3.5.4 VTM Thermal Alert Reset
Reset Overview
The VTM module outputs an alert signal (THERM_MAXTEMP_OUTRANGE_ALERT), when the device
temperature goes beyond a maximum threshold.
This reset signal creates a MCU domain warm reset.
VTM alert will be active as long as the error condition (Device Temperature > Maximum Temperature Threshold)
is still TRUE.
When the error goes away, VTM alert will be de-asserted and device will come out of warm reset.
Note
This signal, when asserted HIGH, will cause a device warm reset similar to MCU_RESETz.

(+) AM625: Linux Thermal shut down - Processors forum - Processors - TI E2E support forums

(+) [FAQ] SK-AM62: How do I measure power and temperature on the AM62A and the AM62X? - Processors forum - Processors - TI E2E support forums

Please refer to above E2E FAQ for configuring thermal shutdown based on configuration of on die temp sensors. When you hit the temperature threshold, it will cause a thermal shutdown. 

Temperature test reading

https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1256650/am625-am625-temperature-test-reading-help-to-get-accurate-reading/4757549?tisearch=e2e-sitesearch&keymatch=am625%252520thermal#4757549

We have AM62xx-EVM board. We are doing some basic test on SK-EVM.
Observed some temperature on that. If read temp from command it is showing like below
cat /sys/class/thermal/thermal_zone0/temp
49042
if we convert that it is in 49.0C^ if I am right.
This 49.0 C is abnormal which is we think. It should come at lease between 20 C to 30 C in normal temperature.
Can we know what should output can come in normal if we test.
Anything we need to change from themal side ? we are expecting below 30 C degree.
Let me know if missed anything or need to check other way?

I think that temperature is accurate and well within range. As a reminder, that zone reads SOC temperatures. Even if it is idling, the operating temperature of the device should be close to 50C. If you wish to decrease the idle operating temperature, then look into adding a heatsink.
To put this into perspective, most of the embedded boards I've used have tended to idle around 45-50C with a room temperature of 23C. The Raspberry Pi 3 for example idles around 48C, so this is not an abnormality.
For more information about device thermal zones please see Section 6.2.5.5 Voltage and Thermal Manager (VTM) in the Technical Reference Manual for AM62x

Regards,

Sreenivasa

HI Board designers, 

 AM62A7 VTM thermal shutdown based on which temperature zone temperature？

The VTM also has a hardware alert signal that indicates a fault when any of the below sensors detect the device is nearing the maximum temperature allowed.

In AM62A, the numbering is:
Temp Sensor 0: c7x
Temp Sensor 1: A53
Temp Sensor 2: DDR controller

This alert can be used to reset the device or part of the device to reduce the temperature. This alert is not a standard interrupt and so it does not include the usual set, clear, and enable features. It does include an overall enable MMR bit

Table 14-7450. MMR_VBUSP_CFG2, MMR_VBUSP_CFG2 Registers, Base

VTM_MISC_CTRL 

along with enables per sensor

14.2.2.2 VTM Registers
Table 14-7449. MMR_VBUSP_CFG1, MMR_VBUSP_CFG1 Registers, Base

The VTM module outputs an alert signal (THERM_MAXTEMP_OUTRANGE_ALERT), when the device
temperature goes beyond a maximum threshold.
This reset signal creates a MCU domain warm reset.
VTM alert will be active as long as the error condition (Device Temperature > Maximum Temperature Threshold)
is still TRUE.
When the error goes away, VTM alert will be de-asserted and device will come out of warm reset.
Note
This signal, when asserted HIGH, will cause a device warm reset similar to MCU_RESETz.

FAQ reference 

(+) [FAQ] AM62A7: AM62A7 VTM thermal shutdown based on which temperature zone temperature？ - Processors forum - Processors - TI E2E support forums

(33) AM62A7: main*_thermal critical temperature threshold is too high? - Processors forum - Processors - TI E2E support forums

Regards,

Sreenivasa

HI Board designers, 

Refer below FAQ

[FAQ] SK-AM62: How do I measure power and temperature on the AM62A and the AM62X?

https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1273364/faq-sk-am62-how-do-i-measure-power-and-temperature-on-the-am62a-and-the-am62x

AM62A7: Linux VTM thermal shutdown status

https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1386179/am62a7-linux-vtm-thermal-shutdown-status/5303389?tisearch=e2e-sitesearch&keymatch=VTM#5303389

Please refer to above E2E FAQ for configuring thermal shutdown based on configuration of on die temp sensors. When you hit the temperature threshold, it will cause a thermal shutdown. You will need configure the PMIC to restart the device as a system level solution as there is no restart mechanism associated with thermal shutdown that can be initiated after the thermal shutdown takes place.

main*_thermal critical temperature threshold 

https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1480999/am62a7-main-_thermal-critical-temperature-threshold-is-too-high/5720718#5720718

We set the SDK to primary market for each SoC.. Am62x is industrial. AM62a and am62p were launched for auto only and then industrial parts added.. We havent done any formal kick off or support for industrial package.
Expecting customers to modify the values offline

We now are specifying an accuracy of +/- 5C

There are no additional margins included in the max operating temp to account for the sensor error.  The shutdown point needs to be set 5C degrees lower than the max operating temp to ensure the limit is not exceeded.

When trip point is set to 120C there is a likely chance that the SOC trips at 115C which is 10C less that the specified.

Customer should be considering this variation in their design. 

Regards,

Sreenivasa

HI Board designers, 

Refer below

In order to measure the SOC temperature, you can follow the steps mentioned below:

1. From M4 core, you can check the value at the register 0x80B00308. Because of RAT configuration, the address 0xB00308 gets mapped to 0x80B00308 in M4.
2. Convert the value read at the register to binary format and take the 10 least significant bits Calculate the decimal value of the number obtained.
3. You can substitute the number calculated in step 2 and substitute it in the equation mentioned in the TRM and calculate the SOC temperature.

(Adding an example below for better understanding)

Example

1.

2. 8095F = 00000000000010000000100101100001 -> 0101100001 = 353 (Taking 10 least significant bits)

3. a.

         a. Substitute x=353 in the equation above and calc. the SOC temperature.

         b.(-9.2627E-12)x(353)⁴ +(6.0373E-08)x(353)³ +(-1.7058E-04)x(353)² +(3.2512E-01)x(353) +(-4.9002E+01) = 47.02384904775517°C

Regards,

Sreenivasa