TX2 not booting in cold conditions

Hi there,

I am having issues with booting up the TX2 in cold conditions with our carrier board. When it warms up the TX2 is able to boot. We are wondering what signals are required to boot the TX2. We have a FPGA on the carrier board that does power sequencing and controls the RESET_OUT# and make sure all the power is good. We measured the rails of all the power on our carrier board and it is good but just when the TX2 is in a cold chamber it does not boot with our carrier board. The carrier board is at room temperature. Any help is much appreciated.

Hi avew, the temperature range of module is -25C~80C, what’s the chamber temperature when this happen?

The chamber is 0°C. We tried it on the actual TX2 carrier board and it does boot up. I have a feeling one of the signals for boot up is weak on our carrier board and may need a pull up. We do receive the CARRIER_POWER_ON signal on pin A48 from the TX2 upon start up. After receiving that signal we wait 2 seconds before we pull RESET_OUT# high. Do you know which signals then boot up the TX2? We would like to probe them to see the voltage levels.

In normal condition, if RESET_OUT# is released, the system will boot up, no other signal needed. The typical duration from CARRIER_PWR_ON to RESET_OUT# released is 77.4ms as you can see in Table 7 in OEM DG. Your 2s setting seems too long, did you try shorter duration?

You mean this issue happen when only module is in chamber and carrier board in room temperature? So what’s the connection between module and carrier board? A FPC? If so, that could meet some unstable signals problem.

No we remove the module from the chamber and connect to the carrier board which is at room temperature. I will try to shorten the time between signals. We noticed something unusual with the Vin_PWR_BAD#. Since it is controlled by the FPGA, at start up when the FPGA is in the mist of turning on all its I/O pins are spike for high momentarily, thus Vin_PWR_BAD# is high for very short amount of time, will this affect the boot up of TX2 at cold temperatures? So our Vin_PWR_BAD# is pulled high then goes low and follows the power sequence on the OEM DG once the FPGA is booted up.

That momentarily start when Vin_PWR_BAD# is high at the beginning does not seem to affect the TX2 boot up when the TX2 module is at room temperature I am not sure if this is the root of the problem when the TX2 module is cold. Any help is much appreciated. Thanks for your help Trumany.

We managed to remove that spike with a large capacitor but the TX2 still does not boot when cold.

I dont understand why the TX2 fails to boot in 0°C our FPGA logic remains the same for power sequencing and is within the design guide specs. It works fine in room temperature. I know the colder TX2 module does boot up if we use Nvidia TX2 Carrier board and CTI carrier board so this puzzles me even further. Do you have any ideas why it fails to boot? We tried shortening and lengthening the RESET_OUT# signal. We played with the pwr btn signal as well. None of our changes seem to make the cold TX2 module work with our carrier board. Do you have any suggestions on what could be the reason?

We power the cold TX2 module with 12V and 1.5A same as when it in room temperature. We see that the module doesn’t draw much current and gets stuck at 0.08A when cold. When it is at normal temperature the current jumps around as the module starts to boot. The biggest difference I find with our board is that 1V8 for the TX2 Jetson and carrier board powers on at the start even before the power sequence. Even with this difference TX2 does boot up successfully when at room temp but I don’t understand why this is affected at colder temps. Any ideas would be much appreciated.

Is your power source also in the cold environment? I know some people have added their own power supplies which fail to boot if they do not either add a large capacitor at the power input connector or use larger gauge wire…I’m thinking the leading edge power up spike is not very tolerant. Perhaps your power source or the connector to the power source differ when cold. Perhaps put a large capacitor on the power input as close to your board as possible and see if it changes. If your power source is also cold, then try with the power source warm and the board cold.

Thanks for your response linuxdev. We are using a SPD3303X-E Siglent power supply which is not in the chamber. It is set at 12V and 1.5A. The only thing we made cold was the TX2 module. When testing to see if it boots, we remove it from the chamber and plug it into our carrier board and use the power supply to power our carrier board both are at room temperature.

We barely see any current being drawn from the power supply when the cold module is connected. I don’t see why the timing and FPGA logic needs to change if it works at room temperature. We have a suspicion that the threshold voltage level for certain signals changes when it gets cold.

It makes me wonder if there is some sort of thermal expansion getting in the way of how solid the connection is between the module and the carrier. You might try once with the carrier itself also in the cold (but not connected to the module), and then pull both out of the chamber and connect as usual. If there is a thermal expansion issue at the connector then the two should be more closely matched when both are cold.

Thanks for your suggestion we have given that a try already, when bot areh frozen and it did not work. The problem seems to lie with just the TX2 module, since if we place just the carrier board into the chamber and then later connect it with a warm TX2 module, the TX2 module does boot up. But if in any scenario the TX2 is cooled down even taking it outside, it fails to boot up.

Did you try pressing RESET button (RESET_IN pin) when power on fail?