STM32F769, LTDC startup not always done, LTDC stuck

[Anton Olofsson]

We are using the STM32F769 SOC and have found a problem in that the LTDC controller is not started on all resets.

Only way to get it to work again is removing power completely.

Neither SW or PIN reset works, only POR will make the LTDC start again.

All other CPU activity is fine and we have double checked that display is OK and working.

The problem only seems to happen on startup, If a unit has started OK it will not stop working.

Its a fairly common issue on our design.

We would appreciate any hints or errata regarding this.

Best regards,

Anton Olofsson

[STOne-32]

Dear @Anton Olofsson​ ,

I would advice to get contact with your local ST FAE or distributor and to provide the exact sequence of LTDC SW configuration able to reproduce the issue

even from time to time on a failing unit or our evaluation borad/discovery It may happen a wrong initialization sequence, not inline with our Reference Manual recommendation such as start to do some actions on Areas/memory not yet ready. looking to the fact that only POR is recovering, that means a kind of freeze or non wanted lock-up !

Cheers,

STOne-32.

[Anton Olofsson]

Best @STOne-32​ 

Thank you for your fast answer and recommendations!

Will do!

I'm pretty sure our startup sequence is exactly the one generated from CubeMX, although a very old version as the project is mature (4.19.0).

We have only updated display-timings for our display-panel.

We will also try to update CubeMX and re-generate the code.

Best,

Anton

[Anton Olofsson]

Update on issue:

Even SWD re-program has been verified not to reset LTDC. Only POR is confirmed to reset it.

Stepping code while hanged gives no clues as of yet. Registers are exactly the same in hanged/working.

Attached gdb dump of LTDC registers when hanged (same as when working).

[waclawek.jan]

Can you try to power-reset the attached display, and then soft-reset the 'F769?

JW

[Anton Olofsson]

Thank you Jan,

This does not help. We have removed the display completely, re-attached it, soft-reset the SOM and still stuck.

Next task is for us to connect a oscilloscope and check all clock and data pins.

We should also check if the LTDC_IRQHandler is still active.

Anton

[Anton Olofsson]

Update on issue:

We have found that all signals are OK (clock, DE, pixels) from CPU with oscilloscope, so problem must be with display.

Sorry for unnecessary post on this ST forum.

Thank you everyone who helped! I appreciate it.

Best Anton

[waclawek.jan]

> all signals are OK (clock, DE, pixels)

How do you mean, pixels? This is not a MIPI/DSI display?

JW

[Anton Olofsson]

Jan,

Correct, this is not a serial interface, product uses parallel RGB TTL.

Best,

Anton