We have designed custom carrier boards for the Jetson AGX Xavier with a 72W Switching Regulator. We now face the issue that when the Module is powered on by VIN_PWR_ON, there is a dropdown of the voltage due to the high current necessary for the module and then the Xavier is not powering up. We use the pushbutton IC circuit from the design guide.
Can you number the inrush current for SYS_VIN_HV and MV? How many capacitors do you suggest to be placed in each power rail to the module? Is there any thing else to consider?
Thank you.
Hi, as you can see in module datasheet, the IDDmax of HV is 5.4A@20V, and 6A@5V for MV. Are you powering on with external device attached which might draw too much current? If so, you can calculate the capacitance accordingly or just try different caps until power on normally.
The voltage drop occurs when the Jetson Module powers on, on the carrier board, nothing is then yet turned on. As you provide a 65 W power supply with the Devkit, how should this work it the Jetson AGX Xavier would draw 5.4A@20V and 6A@5V?
Does the module itself has a voltage supervisor? At which voltage it shuts itself down?
5.4A@20V and 6A@5V are software limited maximum power consumption by power monitor in module. The threshold of input voltage loss is ~8V which is monitored by VIN Loss Detection design on carrier board. Did you measure the voltage and current when power down happen?
We measured the current and we see peaks up to 8 A (HV and MV together). The Module is supplied with 12 V and 5 V. The 5V supply does not show voltage drops, but the 12V shows these drops down to 4V. But even with these voltage drops, we see some systems that start and some that after some retries do not start at all. Attached a picture, where the blue curve is the 12V always-on, the red curve is the power good signal of the switching regulator and the yellow line is the 12V supply to the Module. When this voltage goes up, the switching regulator output voltage goes down. The yellow curve then follows the blue one.
Thanks for the sharing. It would be better if have current value of HV. Also we are checking this kind of 12V HV problem internally, will update if any finding.
Did you try any big capacitors? Is it better or not?
We added 100uF capacitors to HV and MV close to the Xavier Module, but still the same voltage drops. Weirdly, the jetson AGX xavier is not reacting to these, as we see the power button press is generated by the pushbutton IC circuit only once before and the system boots up anyway.
In general 100uF caps are big enough, so if no external device attached when power on, the 4V is too low which might indicate your board design has some problems. Do you have dev kit to compare the voltage drop? Any clue on the power rail circuit and power supply cable?
