iMX28 hang-up when voltage holes in VDD5V

[JuanmaAlvarez]

Hello,

I've a problem with iMX28.

Doing EMC test of my design I discover that iMX28 hangs-up when there are voltage holes in VDD5V, then it goes to a indeterminated state forever.

RESETN line has no effect, it only works again unplugging the power supply and plugging it again.

See the attached image. Yellow line is main Vcc 12V, blue line is 5V that feeds the iMX28.

I can reproduce the hang-up in all my prototipes even in the EVK. It happens at any step: bootlets, kernel linux boot, linux running.

So I can deduce that I can't solve this changing the register configuration. Maybe changing pin configuration???

My design is based on the EVK schematic, but I use VDD5V only, without battery.

I'm not clear about the use and the right connection of BATTERY pin and DCDC_BAT pin in VDD5V only design. The Freescale's docs is not clear about this.

In my design that pins are unconnected.

Using an external watchdog that controls the on/off of the VDD5V maybe works, but that be last action (that makes more expensive my design and imply a re-design)

Any ideas?

Thank you!!!

PD: See the attached pdf for schematic about power-cpu

[x10]

Hi, Juan

I got similar problem. In my design, 5V is regulated and it's output is connected to DCDC_BATT and BATTERY pin as well, so that the 4P2 linear is not work, which makes chip quite hot. If the regulator is implemented in LDO chip, there will be a voltage fall in VDDA(1.8V) to as low as 1.5V, the data in DDR2 could be damaged which cause system corrupted. The LDO can drive 800mA and no voltage fall is captured on its output. So I think that the problem may be cuased by DCDC within imx28, is that possible?

BR

Cheng Shi

[JuanmaAlvarez]

Hi Chen Shi,

I think I've located the problem.

DCDC block switches to DCDC_BATT pin thinking worngly that he is conected to a battery when there is a hole in 5V because there are a bypass of current from a serial to usb converter connected to debug port and an external 3V3 regulator in my design.

I think that to connect 5V to DCDC_BATT or BATT pins is an error. That pins are for 3.6V from battery. (Max 4.2V)

Try to connect DCDC_BATT and BATT with a pull-down 4k7 or 10k. That way the DCDC switcher won't change to battery if you have any hole in your 5V supply.

In the reference schematic that pins are tied to VDD4P2 pin with a 1k pull-up. It also works.

Regards

[JorgeRama_rezRi]

Hi Juan,

Please find attached a reference schematic on how to connect the power pins of the MX28 when using 5V only (page 4). It says MX23, but the power management unit is exactly the same, so it applies for MX28 too. Also, please note that you have to install a patch in your SW to be able to use this configuration. This patch is already published on the web.

Best regards

Jorge.

[chris321]

Hello Jorge

do you mean the 5V only patch? Do that mean that a 5V only configuration is only possible with this patch?

Best regards

Chris

[Rodrigue]

Hi Chris,

Yes. You need to patch the power_prep.c to disable swictching to the DCDC converter.

br,

Rod

[JuanmaAlvarez]

Hello Jorge,

Yes, I know that schematic.

I don't understand why the superCap is connected to BATT pin instead DCDC_BATT pin.

AN4199 (Using the i.MX28 Power Management Unit and Battery Charger. Getting Started with Power Supply) sais that BATT pin is the output of the battery charger.