Hi,
in a new product we use the DA9061-66 with an iMX6ULL. The product is finished and we already had market launch.
Generally everything works fine (passed all EMI tests with above the norm levels without problems, no problems during our development phase) but unfortunately in our long time tests we experience problems. After ~1-21 days on some of our devices we see that the DA9061 shuts off the Buck3 regulator which supplies our DDR3 RAM (we use one 16-Bit DDR3 IC). This leads to a freezing device from which there is no other way out than turning the power off and on again. So market launch is stopped until we resolve the problem.
At the moment we have no idea why it happens.
Here is what we already tried:
——我们没有电报I2C的监控which shut off the regulator when the device has a failure.
- On a device in failure mode, we read out the registers of the DA9061 with the development board, I will attach the file to this post as .pdf file, but it is a .txt file. We found nothing unusual. Buck3 is enabled. Maybe we overlooked something.
- On this failing device we monitored the current on the Buck 3 rail and there was no overcurrent event.
——Vref DA9061设备在冻结是1,19 v
- Vcore of the DA9061 was 2.5V
- Only disconnecting the power seems to get the DA9061 back to normal work
- It seems totally random when it happens. On one device it happened after 30 minutes, next run after 20 days. Same software, same environment and so on
- We tried to force the error with different approaches but until now we didn't find a way to reproduce it. It only happens completely random, but it seems as if it always happens on the same devices.
- Our system logs (we use Linux) don't show anything. The system just stops without any error message probably when the RAM voltage fails. Until now we didn't get a trigger to the falling voltage, we are still trying to. Its a bit of a problem with the long time between the error occurence.
At the moment the most plausible explanation to us is that the DA9061 chips on this devices are somehow damaged.
We already submitted schematics and layout to Dialog but got no feedback to them now for 24 days, which is very disappointing to me.
Has anyone here ever had a comparable thing with this PMIC or this PMIC family? We are at a dead end here as it seems.
Thank you very much,
Best regards,
Torben
Hello Torben,
I will reach out directly to you.
Kind regards,
Elliott
Hi Elliott,
looking forward to it.
Hello Torben,
we are in a very similar situation as you are. We are using a DA9063 device in a product that will be released in a few weeks.
In some products the DA9063 devices now show the same behavior as you already described - they suddenly shut-off the BUCK-Peri DCDC after it has been running normally for several days.
Our investigations results are as follows:
- The DA9063 internal logic/sequencer seems to be still running as nRESET pin and other voltages remain active while BUCK_PERI is turned off
- No I2C communication with the DA9063 when the BUCK-Peri DCDC is shut-off (one thought was, that the regulator is switched off via I2C)
- We connected an I2C sniffer to the devices that had switched off the BUCK_PERI regulator to read out the status and control registers:
- After a powercycle BUCK-Peri is working correctly again
Until now we have not been able to force the failure.
- We applied different loads and overload to the DCDC (including sudden load steps)
- We applied external voltages to the device (below and above the DCDC output voltage)
To answer your question Torben, yes we see the almost exact same behavior as you described with a device of the same family. And we are running out of ideas for the further investigation.
@Eliot: I would appriciate very much if we could get in contact and work on a solution toghether. This is a very serious issue for as as we are very close to the release of our product.
Thank you very much for your help.
Best regards
Florian
Hi Florian,
that sounds very much like a comparable problem, except we do not measure any voltage left on the disabled regulator, this seems different.
We are still working on the problem here. No luck forcing the error until now here either.
Best regards
Torben
Hello Florian,
I will reach out directly to you.
Kind regards,
Elliott