RTC battery

Robotics & Edge Computing Jetson & Embedded Systems Jetson AGX Xavier
1

Hello

I’ve learned after reading lots of date-time issues about the RTC clock and it’s batteries, that:

By default, AGX Dev Kit has a RTC battery, called supercapacitor or C512. Normally it should long for weeks.
That a 2pin CMOS battery can be used, just as a backup.

Is that correct?
If I disconnect even a minute the AGX loose it’s time. I need desperately internet access to get ntp to maintain time and don’t comeback to 1970. This no only happens with one agx , it happens with 3 AGX we’ve got.

I notice this issue happens to a lot of people, and to all xavier I’ve tested. No matter what Jetpack (4.4 to 5.1) I use.

Is this a factory issue? Or is this capacitor in fact like a battery that should be replaced from time to time? It’s super tiny, and I don’t see how to extract it and replace it.
(Jetson Xavier system date and time resets to Jan, 28 after each reboot. - #3 by albertr)
By the way, I don’t find it in google with reference c512.

Regards

4

You should check if C512 is mounted on your board. It is not super tiny as below showing.
C512

If it is not mounted, the following backup cells may be attached to this pin:
Super Capacitor (gold cap, double layer electrolytic)
Standard capacitors (tantalum)
Rechargeable Lithium Manganese cells

5

Maybe we’ve got different tiny size in mind :)

The schematich is not the same. But 512 drawing is quite similar. Looks like a “small” removable cell battery. Take a look:

IMG-6532
IMG-65321920×2560 351 KB

6

Have you probe its voltage level? If it is >1.65V, then should be no problem from hardware side.

7

I’ve verified in 4 jetsons, touching

  • one of the soldered pins to the board
  • the top of the battery (finally is this a battery or a capacitor?) . By the way, I thought the voltage of the top of the battery and the top “holder” that is on it would be the same.

Result of measurement:
2.9 V
When I disconnect power, it goes to 0 V in 30 seconds. :S

8

Had the same issue, The battery drains pretty quick and the clock drifts away. Why don’t you try to make an NTP server with a GPS or so source if internet is not always available? (depends on the application) if not anything makes life lot easier when parsing logfiles :D

1 Like
9

My application doesn’t requires internet, and has not access to it while is making inferences - It’s in a vehicle -. Do I need 4G because jetson RTC battery is a joke?

10

Just use GPS if you are worried about data or security,

Internet is your oyster

A Guide To GPS NTP Servers | Network Time Servers | TimeTools.

Also if you ask chatGPT it will litrally find the steps and the code to start the daemon reconfigure the gpsd.service

11

Thanks for the info. It’s super useful, and maybe I will take an approach in other project I’ve got to do. But in this case, I don’t use GPS.

12

cheers,
it is super cheap because you don’t need anything like RTK,

for simpler things that don’t mind a slight drift we even send GPS over CAN so our onboard edge nodes can correct their internal clocks and have time coherent internal logs etc.

13

I see it a little bit over engineering here? :P

In any case , this is not the solution for a problem with the rtc clock. Every computer I’ve bought in my life keep the clock for a long long time. Every embedded device I bought too. But an Nvidia product of thousand dollars can’t? I mean,Ive got 4 Xavier AGX, if the failure is confirmed by Nvidia, I can assure this is not a bug or issue. It’s a design problem. Hard to digest and understand.

With the price tag is awful that: I have to learn and configure lots of things , buying a gps I shouldn’t need, and dedicate time and effort to implement it and maintain.

I hope @Trumany give me better news.

15

Hi masip85,

Are you using the devkit or custom board for AGX Xavier?

Do you mean the RTC time would lost after a minute of power off?

Could you help to provide the serial console log for further check?
and provide the result of the following command on your board.

# ls -l /dev/rtc*
16

Yes, nvidia dev kit. You can see the picture I posted.

Do you want the serial log,at which moment? The c512 has voltage, and it’s lost when power is turned off in seconds:

Result of measurement:
2.9 V
When I disconnect power, it goes to 0 V in 30 seconds. :S

I obtained:

lrwxrwxrwx 1 root root 4 Mar 2 13:58 /dev/rtc → rtc1
crw------- 1 root root 250, 0 Mar 2 13:58 /dev/rtc0
crw------- 1 root root 250, 1 Mar 2 13:58 /dev/rtc1

Do you mean the RTC time would lost after a minute of power off?
→ even if I power off shorter than a minute

17

Please provide the serial console log from powered up.
NVIDIA Jetson Xavier - Serial Console

18

Hello

This is the boot log:

capturefile (33.5 KB)

19

Do you disable the kernel log? It seems very less messages of kernel.
Please help to capture dmesg instead.

20

I don’t know even how to disable that.

Here I’ve attached /var/log/dmesg.0
dmesg.0 (81.6 KB)

Regards

21 
[    2.919986] kernel: tegra_rtc c2a0000.rtc: registered as rtc1
[    2.922877] kernel: tegra_rtc c2a0000.rtc: setting system clock to 1970-01-01T00:00:29 UTC (29)
[    2.930866] kernel: tegra_rtc c2a0000.rtc: Tegra internal Real Time Clock

It seems you are using rtc1.

Please refer to the following thread to use rtc0 and check if it could help.
RTC can't hold time the after system rebooted in L4T 35.1 - #19 by ShaneCCC

22

It’s a fresh Jetpack version. Also happens with old ones. Do you mean that if you want it redirecting is mandatory?

Regards

23

Just modify that through kernel config to use rtc0 instead of rtc1.