Industrial AGX ORIN not booting with custom carrier

Hi NVidia Team

We have a custom carrier board with which the AGX Orin Module boots always without any issue seen. However if we use the AGX Orin industrial module, the system does not start when applying power for the first time. If we take off power and quickly put it back, the system boots. We checked the signals and CARRIER_POWER_ON and VIN_POWER_ON are never asserted in the case the system does not start. On our custom carrier, we have the identical Power-On Button Circuit as on the DevKit and L53 is connected also to 3V3_AO. Is there anything else we have to consider for the industrial module?
Thank you.

Kind regards

Hi, did you check the timing of 3V3_AO? As said in DG: The SYS_VIN_SV (JAOi only), SYS_VIN_HV, and SYS_VIN_MV are derived from this power source. SYS_VIN_SV (JAOi only) must be powered up first or may have remained powered from previous power cycle.

Hi Trumany

Thank you for your answer. As the 3V3_AO is supplying the Power-On MCU and SYS_VIN_HV and SYS_VIN_MV are controlled by the MCU over VIN_POWER_ON, it is for certain powered first. Anything else we can check?

No clue. Suggest to compare your design to reference to find out why even VIN_POWER_ON is not asserted in first plug in. Was the ACOK signal correctly asserted in this case?

Hi Trumany

The ACOK signal gets asserted:

1 Yellow: ACOK
3 Pink: SYS_VIN_MV
4 Green: SYS_VIN_HV

It seems like the CARRIER_POWER_ON never gets active even if MODULE_POWER_ON is asserted.
Thank you for your help.

So the VIN_POWER_ON is asserted correctly? But SYS_VIN_MV will drop after a while? That’s not same as your last comment.

Hi Trumany

You are right, in the first comment VIN_POWER_ON is wrongly mentioned, it is only CARRIER_POWER_ON that does not get active. SYS_VIN_MV and SYS_VIN_HV will drop after a while as VIN_POWER_ON goes low again.

We pulled-up the ACOK signal on Pin P0.6 of the Powerbutton MCU to VCC3A, this does not change the behavior. Anything else we can try?

Hi, please measure MODULE_POWER_ON to check if any glitch during and after SV is up. MODULE_POWER_ON must be low stably during SV ramp up. Please increase the sampling rate of the scope and check, we ever encounter such issue JAO can boot while JAOi cannot boot, it turns out to be such glitch.

Hi Trumany

Thank you for your answer. Here our measurements: (609.9 KB)

MODULE_POWER_ON is CH1 (yellow) and SYS_VIN_SV is CH2 (blue)

The first glitch when 3V3_AO is rising is explainable, but how can there be a second glitch? The MODULE_POWER_ON is only controlled by the JAOi and the Button MCU. There is also a Level Shifter exactly as in the DevKit Design (SN74LVC1T45DBVR) and a 100 kOhm pull-down on the side to the Orin Module.

When the system boots, the second glitch in the MODULE_POWER_ON signal is not there. The power supplies to the level shifter (3V3_AO and 5V_AO) are stable.

Are you using the same button MCU and same module_power_on design as P3737_A04? Please share this part design for check.

Did you add the 10k pull-down as below from Design Guide?

Hi Trumany

Here the schematic of the button MCU circuit, it should be identical to the P3737_A04:
buttonMCU.pdf (150.3 KB)

We did not add the 10k pull-down as there is also no pull-down in the P3737_A04 design.
Thank you for checking.

Hi Trumany

We added a pull-down to the level-shifter on the button MCU side and the glitch disappears, which leads to a working power-on with the indutrial AGX Orin.
As on the DevKit, this pull-down is also missing, we wonder why it still boots with the industrial Orin while our custom carrier board does not? Could there be done something in the MCU firmware?
Thank you.

There is rework on P3737_A04 board. You can find the pull-down somewhere on board.

Hi Trumany

Shouldn’t this then be mentioned somewhere in an Errata of the schematic or likewise? In the design guide it is written to copy the circuit from the DevKit.

It has been fixed in the latest schematic that will be released soon. Thanks.

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