AT91SAM9G25 Hangs - CPU unresponsive - Requires powercycling

Go To Last Post
3 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi, We are using AT91SAM9G25 MPU and its hangs now and then. We use a 2.8V super capacitor to power VDDBU. Just resetting the processor is not helping at all.

 

Hanging issue can be recreated by power-cycling the board many times in a second. Once the CPU enters into a hang state, NRST is held low and SHDN pin will have the VDDBU rail voltage, here 2.8V.

In other forums I have seen that another single power-cycling makes the CPU come out of the hang condition. But in this case it was not happening, it took long time, say 1 hr to get out of hang condition. So we suspected the VDDBU rail which still has power from the Super-Cap.

 

We discharged the Supercap while the board was off with a 5ohm resister for 2 sec, during which period the voltage on VDDBU went below 1V level and came back to 2.8 V after removing the discharge path. This time the cpu came out of hang condition the moment VDDBU was cycled. we were able to recreate the case many times. Every time only when we bring the voltage on VDDBU low, along with all other power rails off, the CPU came out of hang condition.

 

In practical scenario, in our case, VDDBU is supposed to be there virtually all the time (to save RTC data). Is there any explanation for this behavior of CPU? How can I prevent this hang conditions? ( the board has enough capacitors and transorbs). Is it a bug in POR section of ATMEL devices? I have seen in forums, many people reporting same hang cases with different Atmel devices.

Last Edited: Tue. Dec 1, 2015 - 06:38 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

hello,

did you observe all the hints regarding power-up/down of the schematic check list?

 

regards

gerhard
 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks,

 

Found the problem. Sorry that I am very late to post it.

Trigger for the problem - field technicians used to screw the live DC power cable to the board. This caused transients and the POR circuitry of the processor gets lost some where and processor is always in reset state (even if turned off and on). Input filter and caps were helpless. 

Reason for not recovering after turning board off and on: Board had a supercap for on-chip RTC (VDDBU rail). The MPU virtually never losses power on this rail and the normal power cycle is not sufficient to reset the POR circuitry.

Solution:

Though the power sequencing is just two stage, we staged all the power supplies to the processor with the help of PGOOD and Power enable of the regulators used. All the regulators used the PGOOD of the supply preceding it. The tactic was to delay the core voltage and at the same time make use of the regulators' inherent transient blocking property, thereby reducing the effect of transients at the final, core voltage and this worked.

Last Edited: Tue. Aug 7, 2018 - 09:45 AM