About THERM_ALERT_N signal of AGX Orin

Hi,

What are the conditions for assertion of the THERM_ALERT_N signal in AGX Orin?
Which parameter is asserted at what degree?
Is this related to the BPMP Thermal Framework below?

BPMP Thermal Framework
https://docs.nvidia.com/jetson/archives/r35.3.1/DeveloperGuide/text/SD/PlatformPowerAndPerformance/JetsonOrinNanoSeriesJetsonOrinNxSeriesAndJetsonAgxOrinSeries.html?highlight=throttle#bpmp-thermal-framework

Best Regards,
UNA

Hi UNA_H

Could you help to provide the full serial console log when you hit this assertion issue?

Hi Kevin,

Thank you for your reply.

I am not currently facing any problems regarding THERM_ALERT_N signal.
I just want to know the specific conditions under which the THERM_ALERT_N signal is asserted.

Best Regards,
UNA

THERM_ALERT_N(PG.01) seems be used by nct1008_nct72 driver.

Could you share the full dmesg for further check?

As I commented in the previous section, THERM_ALERT_N is not asserted in my environment (JAO Devkit).

I am attaching the full dmeg log, but is it possible to determine the assertion condition of THERM_ALERT_N from this information?

dmesg_full.log (73.4 KB)

There’s a tmp451 thermal sensor on AGX Orin, THERM_ALERT_N is a input pin to detect the high temperature.
There’s no interrupt handling for this pin currently. (i.e. it would do nothing even when this pin tiggered)

But TEMP_THERM would be asserted when the temperature is too high (107C) and causing shutdown.
You could refer to the the following logs from your dmesg.

[   13.456401] nct1008_nct72 4-004c: starting parse dt
[   13.456450] nct1008_nct72 4-004c: !!!Found deprecated property!!!
[   13.456452] nct1008_nct72 4-004c: success parsing dt
[   13.456582] nct1008_nct72 4-004c: success in enabling tmp451 VDD rail
[   13.486161] tegra-hda 3510000.hda: Adding to iommu group 58
[   13.493406] nct1008_nct72 4-004c: LOC shutdown limit 107
[   13.496524] nct1008_nct72 4-004c: EXT shutdown limit 107
[   13.533431] nct1008_nct72 4-004c: initial LOC temp: 28000 
[   13.534153] nct1008_nct72 4-004c: initial EXT temp: 35000 
[   13.535128] nct1008_nct72 4-004c: hi_b:-4, lo_b:0
[   13.535860] nct1008_nct72 4-004c: nct1008_probe: initialized

Actually, there should be other shutdown temperature in the system.

  1. SW shutdown (104.5C)
  2. soctherm HW shutdown (105C)
  3. tmp451 HW shutdown (107C)

THERM_ALERT_N is not related to this BPMP doc.

According to the AGX Orin Design Guide, as shown below, THERM_ALERT_N is Output, does this mean the documentation is wrong?

If THERM_ALERT_N is meant to be an input pin, is this pin intended to be used to receive alerts of rising temperatures from a system temperature sensor external to the Jetson module or a host MCU monitoring the total system temperature?

Is the TEMP_THERM signal used inside the Jetson module and not exposed externally like the THERM_ALERT_N pin?

Best Regards,
UNA

THERM_ALERT_N should be an input pin, it seems wrong in design guide.

No, it is directly output from tmp451 thermal sensor.

Yes, it is also detected through tmp451 and it would be asserted if the temperature higher than 107C.

Are you saying that the AGX Orin module is configured as shown in the figure below?
If so, what is the THERM_ALERT_N pin provided as an external input pin for?

The direction in the figure should be like the following.

THERM_ALERT_N is output from TMP451 indicating the thermal alerts.
It is an input pin for AGX Orin SoC, but it is an output pin for the module so that the design guide is right.

You could also refer to thermal design guide for more thermal management details.

Could I summarize the discussion so far.

  1. THERM_ALERT_N (L56) pin is output.
  2. TMP451 assert signal is output from the THERM_ALERT_N (L56) pin.
  3. TMP451 is asserted above 107°C.
  4. That is, the THERM_ALERT_N (L56) pin can only be used as a signal to monitor whether the TMP451 has reached 107°C.

Is my understanding as mentioned above correct?
Also, is it possible to change 107°C, the HW shutdown temperature of TMP451, to the user’s desired temperature?

Best Regards,
UNA

Yes, your understanding is correct.

It could be configured through the shutdown-limit in device tree as following.

TMP451_soc: temp-sensor@4c {
	status = "okay";
	#thermal-sensor-cells = <1>;
	compatible = "ti,tmp451";
	reg = <TMP451_SLAVE_ADDR>;
	sensor-name = "tmp451-ext";
	supported-hwrev = <1>;
	offset = <REMOTE_SENS_OFFSET>; /* Actual value will be programmed as REMOTE_SENS_OFFSET/4 */
	conv-rate = <0x06>;
	extended-rage = <1>;

	interrupt-parent = <&tegra_main_gpio>;
	interrupts = <TEMP_ALERT IRQ_TYPE_LEVEL_LOW>;
	temp-alert-gpio = <&tegra_main_gpio TEMP_ALERT 0>;

	/* vdd-supply configured in power tree */

	loc {
		/* set to maximum for not having local  shutdown */
		shutdown-limit = <107>;
	};
	ext {
		shutdown-limit = <REMOTE_SENS_SHUTDOWN_LIMIT>;
	};
};

and you could check the current configuration from the dmesg as I shared before.

Hi KevinFFF,

Thnak you for your information.

Please let me clarify one last point.
Am I correct in understanding that the HW shutdown by TEMP_THERM is triggered as soon as the assertion of the THERM_ALERT_N signal is observed?
In other words, it cannot be used to alert the outside before Jetson is forced to HW shutdown?

Best Regards,
UNA

It seems could not alert the outside before it get asserted.
It is asserted and shutdown at almost the same time.

Thank you for your answer. I got it.

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