iMX535 fails to start

[jahlman]

I have run into a weird problem with my design that uses an iMX535. Essentially the processor does not boot (using SD card) after a successful power on cycle and POR. In addition, the crystal fails to start. However, if you simply touch the iMX with your finger it loads U-Boot, Linux, and seems to operate normally after that. Just pressing on the part with a non-conductive tool does not have the same effect.

I’ve verified the following:

• Rail voltages are well within spec (measured at the iMX pins). Using LTC3589-2 PMIC.
• Power on sequence meets the Freescale spec. POR asserted until ~7mS after last rail voltage is stable.
• I’m using GPIO for the configuration bits. Voltage levels on these pins meet the spec, and are set to boot from SD card low speed.
• Crystal is 24 MHz 80 ohms ESR with two 18pF load caps. CL is 10pF for this part, and I’ve tried a few different combinations with the same result.

Has anyone seen behavior like this? Any help will be greatly appreciated.

[igorpadykov]

Hi Jim

one can check both 32.768KHz and 24MHz crystals requirements

given by MX53UG Chapter 1 Design Checklist, and test signals

by oscilloscope. Try to prolong POR up to 1 sec. and check if this

helps. Also reason may be in low quality crystal or parasitic

resistance on board traces, so crystal can not start. For

load caps one can look at below picture (note this is applicable

to both 32.768KHz and 24MHz crystals)

Best regards

igor

-----------------------------------------------------------------------------------------------------------------------

Note: If this post answers your question, please click the Correct Answer button. Thank you!

-----------------------------------------------------------------------------------------------------------------------

[jahlman]

Thanks for the reply Igor. Unfortunately I misstated my problem slightly. The crystal does start, but the iMX does not start the boot process, or starts very slowly without touching the part with my finger. My initial assumption was that the stray capacitance was ~5pF, so I set the two load caps at 10pF each. I'm using a CTS425 series crystal at 24 MHz and 80 ohms ESR. Once started the crystal operates at 24MHz. As I stated in the original message I also tried two 18pF caps (assuming stray C was ~1pF), but this didn't help. I apologize for the confusion.

The first revision of the board seemed to work, so I fixed the errors and moved on.

Right now I have three boards with the new PCB that have problems:

One will boot on its own, but it is really slow during the U-Boot phase unless I touch the part with my finger. After U-Boot ends it seems to load Linux and operate normally.

Two will not start the boot process without touching the part. After that they seem to boot normally.

It seems that there's something that's not happy with the internal clock or PLLs that's keeping the part from operating normally.

I based my design off the Freescale Quick Start board, and followed the user guide closely (I thought anyway).

[igorpadykov · Accepted solution]

Jim

I think one can try to use other crystal.

Also one needs to verify that 32.768KHz crystal worked

properly, without it 24MHz oscillator may not start.

~igor