Does an disabled crystal hooked to a micro still wiggle?

[jnz]

I have an Infineon XMC1200 micro I'm trying to put into low power mode. I follow the reference manual for steps, I think I hit all the right registers, I get to sleep and I'm not at 250uA, but 1000uA.

I noticed the XTAL1/2 pins are still sin waving. I'm pretty sure they shouldn't be.... right? IIRC the XTAL is only part of an the whole oscillator circuit and there is a special driver in the micro itself, I assume this is where my power drain is coming from.

Thoughts?

[Doctorandus_P]

Microcontrollers have mutiple ways to save power.
In the lowest power (highes power save?) setting the crystal oscillator is usually stopped, and the uC can only be awakend by external events such as interrupts.

So if you uC crystal is still oscillating at it's nominal frequency then you have very likely not configured your uC for the lowest power consumption.

[mvs]

I have an Infineon XMC1200 micro I'm trying to put into low power mode. I follow the reference manual for steps, I think I hit all the right registers, I get to sleep and I'm not at 250uA, but 1000uA.

250µA is lowest consumption in deep-sleep with all kinds of efficiency hacks applied, like code execution from RAM only (flash disabled), all peripherals disabled, Pin Power Save feature enabled, etc.

I noticed the XTAL1/2 pins are still sin waving. I'm pretty sure they shouldn't be.... right?

DCO1 should be disabled in deep-sleep, but not in normal sleep.


PS How long lasts sleep phase in your application? Do you use some peripherals?

[jnz]

I have an Infineon XMC1200 micro I'm trying to put into low power mode. I follow the reference manual for steps, I think I hit all the right registers, I get to sleep and I'm not at 250uA, but 1000uA.

250µA is lowest consumption in deep-sleep with all kinds of efficiency hacks applied, like code execution from RAM only (flash disabled), all peripherals disabled, Pin Power Save feature enabled, etc.

I noticed the XTAL1/2 pins are still sin waving. I'm pretty sure they shouldn't be.... right?

DCO1 should be disabled in deep-sleep, but not in normal sleep.
PS How long lasts sleep phase in your application? Do you use some peripherals?

Yea, I'm looking for explicit DCO1 disable right now. Infineon support is really the worst, we found some "deep sleep" code on their forum, but it leaves the external on.

The deep sleep phase here is permanent. Once the device is done working, it shuts the whole thing down. The low power is so I don't run the battery dead.

EDIT: DCO1 is the internal. I want to disable both. Interestingly I just saw an issue where we did *SOMETHING* to the clocks that entirely broke the chip. Seems to crash the debugger entirely.

[Doctorandus_P]

250uA is also not real low power nowaday's.
some uC's go down to 100nA while still running @ 1MHz.

[jnz]

250uA is also not real low power nowaday's.
some uC's go down to 100nA while still running @ 1MHz.

Thanks, I'll keep that in mind when I re-design Infineon's line up.

[jnz]

Turns out there were a few issues.

The SCU_Analog register didn't read correctly.

You need to put the PASSWD register in before the clock will disable.

There is definitely something you can do to 100% brick these chips. I have no idea what we did, but broke 4 before rolling code way back. This will remain a mystery.