• Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Repairing CPU board

2PAC Mafia

Hi guys,

I´m trying to repair some CPU boards which have random failing behavieurs, sometimes they are not booting and is hanged in different steps consumptions, sometimes they boot and show message about exception, sometimes they boot correctly and after minutes reset, sometimes... They are based in BGA PXA270 processor and Disk on chip BGA memory with Windows as operating system and customized application. I don´t want to extend a lot about this, I understand with all these data I give you is very difficult but I would like to know your opinion about this problem, may be some new idea and do you think about this:

Sometimes a board which is in one state, in example, not booting, after weeks without checking it then it can boot and opposite situation, if you use a lot one which is booting finally it stops booting.
We have done reworks in BGA chips, processor, flash and RAM x 2.
We have changed and test mainly all components with no success.
We have changed the processor and RAMs with no success.
I have tested and compared all the internal power supplies and all is OK.
Sometimes we have discovered swapping the flash memory that the behavieur also swapped.

I have 128 points recorded with scope in a good booting and a bad bootings on all I/O pins of this CPU and I can say when it fails I see how some GPIO´s from processor are with strange measurements, floating like ground is lost (I get sinusoidal measurements, only at some of the GPIO´s of microprocessor, not all) and this point is sending me all the time to the PCB itself which is between 6 and 8 layers, to the processor (which we already changed in 2 of them) and to the flash memory.

About PCB failure of GND internally, it is very difficult to check, if you measure points on grounds all are OK, all interconnected and there is no access to measure at processor because it is BGA (0.30 ball).
Now we are trying to read the firmware, not easy process, to put a new programmed chip. Is it possible that a failure in the flash can make microprocessor GPIO´s outputs floating?

I would say the problem comes always on this PCB´s from the same point, even with this random behavieur, but I don´t know how to solve it.

I have had problems before with similar boards where the flash was damaged but it always was similar behavieur, system not booting.

In this case it changes a lot ramdomly.

Thank you.

dr pepper

Well-Known Member
Most Helpful Member
If I had all the data I'd probably write a diagnostic program to check out the various parts of the system, but I guess you cant do that.
I had an issue on a machine that had remotely similar problems, turned out its graphics card was on the way out, and I think it was pulling the supply down causing problems elsewhere.
These issues can be hard to find.

Latest threads

EE World Online Articles