OLinuXino Lime 2 fails to boot after few powercuts

[Julien]

Hello,

We would like to use Lime 2 for an industrial project.

We are currently testing the board's power loss tolerance.
To do this, we are driving the power supply of a Lime 2 and generating power loss.
We are using an electrical relay.

For some unknown reason, the board stopped booting after a few power loss simulations. The image is not corrupted because if we restart the board (unplug/plug power supply), the board will boot.

The red LED near the socket is on when the board doesn't boot.

We are using an SD card (and we have tried several SD cards but the problem persists.

We have tried another Lime 2 board, but we have the same problem.

We are using the latest OlinuxIno image (A20-OLinuXino-bullseye-minimal-20230515-130040).

We think it's a power problem because Uboot won't boot (no logs on debug). Maybe our relay is generating a surge or something like that?

Have you ever had this problem before?

[LubOlimex]

How fast do you toggle the power supply? If you toggle too fast some capacitors would remain charged and conflict situation would occur.

It is generally not a good practice to unplug the power supply and re-plug it. Eventually it would lead to Linux corruption or SD card corruption.

If you expect power downs, my advice is to incorporate a Li-Po battery (with maybe Li-Po battery charge monitoring and software shutdown).

> For some unknown reason, the board stopped booting after a few power loss simulations. The image is not corrupted because if we restart the board (unplug/plug power supply), the board will boot.

I am confused. Does it boot or not? Did you still try with another card with fresh image?

Sometimes the card might look like it alright but tiny part of it might be corrupted and you experience that corruption only in corner cases.

[Julien]

Thank you for your response.

> How fast do you toggle the power supply?
We wait 10 seconds after switching off before switching on.

> I am confused. Does it boot or not?
For example, it boots 4 times, but the fifth time it doesn't boot.

> Did you still try with another card with fresh image?
Yes, but the problem remains the same.

We are trying with the emmc, I will keep you informed.

[LubOlimex]

As I said this is generally a bad practice. These tests can't end up positive result.

Sudden loss of power supply is not something that is recommended to happen to the hardware. Sudden power downs will certainly lead to some problems.

[Julien]

We've identified the cause of the problem, it's the power module that's not starting properly. There is no 3.3V when the problem occurs.


> As I said this is generally a bad practice. These tests can't end up positive result.

In fact, it's a problem, as you say, we need to use a UPS system.

[Julien]

We have updated our test, now using a Li-Po battery.

We detect that the AC power is off and shut down the system. After two minutes, the power is restored and the system starts 'normally'.

That's the problem: for unknown reasons, the system fails to start at random times.

The AXP209 doesn't supply any power.

I have to disconnect and reconnect the battery and AC power to get it to boot.
Or press the reset button (middle button).

Do you have any information on this?

Thanks you in advance !

[LubOlimex]

> That's the problem: for unknown reasons, the system fails to start at random times.

- When this happened, was the procedure you described followed (aka "We detect that the AC power is off and shut down the system. After two minutes, the power is restored and the system starts 'normally'.")?

- Or was it quick disconnect of power supply, attach power supply?

It is important to notice that in the design we have placed big capacitor near the power jack (C200 in most revisions). It serves to prevent small and quick glitches to the power supply, the downside is that it requires some time to discharge after power supply goes missing. And if you restore the power supply before it is discharged, there would be a problem. This problem will persist no matter if you have battery or not. You might try to improve behavior by unsoldering C200, this way power up and power down would be more responsive. The best test is empirical.

[Julien]

> When this happened, was the procedure you described followed ?

No. If this happens, the process is stopped and the power is on.

> It is important to notice that in the design we have placed big capacitor [...]

If the AC power is of for two minutes, do you think it is possible that the capacitor is not fully discharged ?

However, we're going to do the test by removing it.

[LubOlimex]

I wonder how do you apply the power supply? Via the power jack?

Did you try with more than one power adapter?

Does the issue persists with more than one board?