Strange Core Voltage Problem

[davelund]

Helping someone debug problems with SAM7S256 board they have been building for many years. New issue though.

A small number of units seem to occasionally lockup. Conditions are as follows:

VDDIO/VDDFLASH and VDDIN all at 3.30V
VDDOUT/VDDCORE/VDDPLL sitting at 0.9V

NRST seems to be driving out low. I assume the power on reset has tripped and is causing this.

My guess is that perhaps the voltage regulator is being set to low power or Standby mode? This causes the voltage to collapse due to it's limited output capability and the system shuts down.

Circuit seems OK. It is a multilayer PCB with a full ground plane and a small plane for the core. There is a 10uF ceramic and 5x 0.1uF ceramics on the Core Voltage net.

While debugging things I glitched the core to ground and the unit spontaneously started up.

Any ideas?

Thanks,
Dave

[davelund]

I didn't expect a lot of activity here, so let me ask this question a different way:

The SAM7S data sheet specifies a core voltage rise slope of 6V/mS or better. My question would be, What if you don't meet that. Is there any information on what the symptoms are of not hitting this requirement? I believe the design meets this most of the time but I managed to catch it once with the core voltage stuck at 0.9V Is this one possible outcome?

Thanks,
Dave

[gerhardf]

The SAM7S data sheet specifies a core voltage rise slope of 6V/mS or better. My question would be, What if you don't meet that. Is there any information on what the symptoms are of not hitting this requirement? I believe the design meets this most of the time but I managed to catch it once with the core voltage stuck at 0.9V Is this one possible outcome?

not only the core voltage but also the 3.3V has to rise with at least 6V/ms!

regards
gerhard

[davelund]

Thanks for that tip. We are good on the rise time on the 3.3V also. What I have been desperately trying to confirm is WHAT HAPPENS if you don't meet these specs??? I have very unique failure mode where VDDCORE gets stuck at around 0.9V. It also happens very rarely. I cannot get units to fail in the lab. Any info related to tying the rise time spec to this particular failure mode would be a huge help. Atmel themselves hasn't been able to offer any help here either.

Dave

[davelund]

One further comment. The units fail from a running condition, not a cold start and not a wakeup. This leads me to believe it isn't a rise time related situation. Power never was interrupted. It is starting to look as though the code may be either crashing or programming something into an illegal condition that pulls a lot of current on VDDCORE...

Dave