Stability problem with PDN, reset loop at startup

[jlumme]

We have created a board basing our design on Olimex maxi without the additional IC chip helping with power lines. Our configuration is 5V only without any battery.
We are having big difficulties getting the power lines up in (up-to date) u-boot. The same uboot build will boot Olinuxino Maxi fine.

The board sometimes manages to boot up (like 1 in 20), but its still unstable even then. It seems to behave better "when cold" (caps are fully discharged), but we're not sure if this is a reliable assumption.

Our PDN schematic looks like this:


On the scope the reset loops look like this:


So one can see that clearly our 2v5 initialization is failing, and the board boots. The 5V seems steady all the way, and there is a slight drop on 4v2, but the line recovers (?) before resetting.. 2v5 never reaches even close to it's intended voltage..

We have been trying many different configurations for the capacitors (ie. Freescale reference design for 5v operation only), but so far we are unable to get the PDN stable.

Should we put some load (ie 100k resistor) on the other power lines (VDDA, VDDD) ? Also we wonder why Olimex design is so radically different from Freescale's reference capacitor choises ?

What cap configurations have other projects used ? I read fairly a lot of PDN issues with imx233 around the net (well, here and Freescale forums), so I hope someone with experiences with this could help. Thank you, we really appreciate any advice!

[mbergmann]

Hi,

we have an Audio player based on iMX233 on the market as well as an TTS speech module and an Audio Development Board.

http://blog.bones-embedded.ch/imx233-audio-development-board/

I can share the power related stuff via PM if you are interested.

Regards

Mike

[jlumme]

Hi mbergmann,

We are a 2-man hobby-operation so most definitely all help is welcome

[Evgeny Boger]

jlumme, we have had similar problems with 2.5V supply rail with defective/counterfit Samsung memory chips.
That chips consumed too much current on startup which caused brownout protection to restart the imx233.

So I would suggest you make sure you are using the working RAM chips.

It's also very helpful to add some debug prints to u-boot power init code (arch/arm/cpu/arm926ejs/mxs/spl_power_init.c). This will hopefully allow you to know what command triggers the failure.

[jlumme]

Hi Evgeny,

Interesting. Actually you could be on to something here; we actually bought some Hynix chips over the internet from non-official distributor as well.. Hmm.

Regarding u-boot debugging yes indeed those spl files are the key here. I am always crashing at power init code - so some brownout is happening here..