RT1052 dcdc failed

[SinSera]

We use RT1052DVL6B as the mcu in our products.

In the recent 7*24 uninterruptable power test, there will occasionally be several device screens that do not display.
According to our investigation, we can conclude that the 1.2V voltage of DCDC has no output.

This phenomenon occurs in about 1-2% of the devices tested.
This usually occurs after more than 24 hours of electricity.
We suspect static electricity or voltage fluctuations, but we have no way to prove it, and no good way to test it.

We can't find the reason why 1.2V is no longer output.
Because the core is powered through this DCDC, there is also no way to read any clues to anomalies.

We have tried to use some methods to allow the device to automatically restart in this situation, but so far we have no idea.

Please help！

Best Regards!

[MultipleMonomials]

One thing to be aware of, which I learned from the Teensy 4 schematics, is that the DCDC_PSWITCH signal is edge sensitive, not level sensitive, and it must see a rising edge after power to the MIMXRT is stable in order to turn on the DCDC. If you have a power glitch such that power to the MIMXRT is momentarily lost, but the DCDC_PSWITCH signal stays high without a rising edge, you can end up in a situation where the MIMXRT chip has power but the DC-DC converter is off.

Really the official EVK design should be changed to include a $1 power rail supervisor on this line, because the default RC timer circuit has such a big issue with this...

[mayliu1]

Hi @SinSera ,

Thank you for your interest in the NXP MIMXRT product, I would  like to provide service for you.

First, I would like you check your hardware design, especially design of the DCDC converter.

Please refer to MIMXRT1050-EVKB board design.

https://www.nxp.com/webapp/Download?colCode=MIMXRT1050EVKBHUG

https://www.nxp.com/webapp/Download?colCode=MIMXRT1050EVKHUG&location=null

Please double check  the selection and placement of inductors , capacitors , and other components.  Please confirm that all decoupling capacitors are correctly installed and are placed as close as possible to the MIMXRT1050 chip power pins.

Second, Use an oscilloscope to monitor VDD_SOC_IN and DCDC_IN, observing the startup process and voltage stability during long  time operation. Check for the presence of overshoot, undershoot, or ripple.

Third, Try injecting voltage anomalies (voltage fluctuations or spikes in the input voltage of the DCDC converter may cause the converter to fail), or put your board under high - temperature conditions

Please check  if the issue can be reproduced and whether a patterns can be found.

Finally, I recommended that you can add  watchdog.

Wish it helps you.
If you still have question about it, please kindly let me know.

BR

MayLiu

[SinSera]

Hi, @mayliu1 @

Thank you for your reply!

We already had mature equipment before, and this new device was developed by replacing the display panel and driver chip on this basis.Therefore, we suspect that the introduction of new components has led to changes in the electrical performance of the equipment.

Before the panel was replaced, there had never been a DCDC power failure during normal operation.Similar situations have not occurred in other products using the same MCU either.

By the way, on the equipment using RT1052 as the MCU, there is an extremely rare DCDC power-on failure when the device is powered on for the first time. But this is relatively acceptable, and we are constantly modifying and testing this issue.

It is extremely difficult to use an oscilloscope for tracking tests. We do not have the ability to monitor the oscilloscope 24/7, nor do we have enough oscilloscopes to monitor all the equipment. We can record one or two oscilloscopes, but it can be imagined that it is extremely difficult to find potential voltage fluctuations.

We have tried disabling the on/off button and extending the debounce time to 500ms.

SNVS->LPCR &= ~SNVS_LPCR_BTN_PRESS_TIME_MASK;
SNVS->LPCR |= SNVS_LPCR_BTN_PRESS_TIME(3);
SNVS->LPCR &= ~SNVS_LPCR_DEBOUNCE_MASK;
SNVS->LPCR |= SNVS_LPCR_DEBOUNCE(2);

It seems that the probability of DCDC power loss has decreased.

Currently, during our code review, we found that we are using an old SDK, with the DCDC voltage set at 1.25V and the core frequency at 600MHz. We are wondering if increasing the DCDC voltage to 1.275V or reducing the frequency to 528MHz would bring about any improvement.

Looking forward to your reply.

Best Regards!

[mayliu1]

Hi @SinSera ,

Thanks for your update information.

Question: We have tried disabling the on/off button and extending the debounce time to 500ms.

SNVS->LPCR &= ~SNVS_LPCR_BTN_PRESS_TIME_MASK;
SNVS->LPCR |= SNVS_LPCR_BTN_PRESS_TIME(3);
SNVS->LPCR &= ~SNVS_LPCR_DEBOUNCE_MASK;
SNVS->LPCR |= SNVS_LPCR_DEBOUNCE(2);

It seems that the probability of DCDC power loss has decreased.

Answer: I think that the code you modified involves changes to the time for controlling the MIMXRT1050 to enter the SNVS mode, as well as the amount of debounce time for the BTN input signal.

If you suspect that the RT1050 has entered the SNVS mode, you can wake it up through the wakeup pin. The SNVS mode is a type of low power mode and will not cause the MCU to crash.  

Question: Currently, during our code review, we found that we are using an old SDK, with the DCDC voltage set at 1.25V and the core frequency at 600MHz. We are wondering if increasing the DCDC voltage to 1.275V or reducing the frequency to 528MHz would bring about any improvement.

Answer:  These measures may bring some improvement.  I also suggest you use the latest SDK demo. 

Wish it helps you.
If you still have question about it, please kindly let me know.

BR

MayLiu