Troubleshooting on device

Hi guys,

I know this question is too general but I want to see if I get some good idea.
There is an equipment which is a CAN bus dedicated monitoring unit with screen and always fails on same way, after around 10 years it start failing in this ways:
- Sometimes it is resetting
- Sometimes it switches off and not powering on again after long time
- Sometimes lines on screen and freeze
- Sometimes error message on screen and reset

The problem always come from a pluggable processor board on top of motherboard. We have been repairing other old generations of this boards but on this last generation we can´t find the problem.
We have been testing internal power supplies with logic analizer when problem occurs but everything seems OK, we have been reballing 4 BGA chips (processor, flash and 2 RAMS), we have been replacing some chips but we don´t find the problem.
We have some units all with same problem. Strange thing is if you leave a unit which doesn´t switch on for long time without power sometimes when you try again it works for hours or even one or 2 days.
We have been thinking about a problem on PCB design itself but in this case I´m lost...

May be some general light on this issue from your side? They are very expensive so I would like to find a solution.

Did you try replacing the processors with new a one or just reballed them?
I see similar symptoms in some automotive modules. Its hard to really say what it could be since I've no idea what your system is or the components used but in my cases, it has generally been capacitors or diodes contributing to those type symptoms and I have also seen MCU's failing like this. Sometimes just heating the mcu up would allow it to work and sometimes discharging it would allow it to work for awhile.

Thank you. We can´t replace processor by a new one because it is boot programmed by manufacturer. Here is a list of main components on board:

RTPXA270C5 (BGA) (Processor)
K4M513233C-DN75 (BGA) (Dinamic RAM)
SEC 130 EF70 K6F4016U6E (BGA) (CMOS SRAM)
M-Systems Flash disk Pioneer, Disk on chip G3, 0637KFD1JK2122, MD4832-d512-V3Q18-X-P-JP (BGA) ( Flash)
MIC2026-2YM (Dual channel power distribution switch)
3BUB5 (3BLBL) (SMD 6) (Mosfet P?)

AC244 x 2 (octal buffer and line driver)
HCT244 x 2 (octal buffer and line driver)
MAX704S ESA (4,40V supervisor)
MAX1626ESA (DC/DC step down) from 5V to 3,3V
E8564 (Real Time clock)
WZD2 = NC7WZ132 (Dual 2 input NAND gate with Schmitt triggers)
C11R (SMD 6) = SN74LVC1G11DBVR (single 3 input positive AND gate)
LCX245 x 2 (low voltage bidirectional transceiver)
MAX3223EAP (RS232 transceiver)
CBT3384 (10 bit low power bus switch)
71ACRZ = ISL6271A (Integrated Xscale regulator)
Altera EPM7032AE TI44-7N

The strange thing is that point about when you leave the unit for long time, after that it starts. If you start it during a while it fails more and more until it doesn´t start anymore.
About capacitors there are a few only, usually capacitor problems comes from electrolitic and affect the power supply, but in this case we don´t see that.
We also played with heat and cold on board to try to discover things but it is too random. Also very difficult to measure because there are 4 BGA chips and board is multilayer double face components and has to be connected to main board so no access to bottom, only soldering wires...
In an older generations I found a similar problem on a component which doesn´t exist on this PCB version.

I odnt know if its true however I heard in the 80's & 90's there were some Ic's, delay lines was one that degenerated.
A problem I once had, circuit glue, nasty yellow stuff that after a while turns brown & becomes conductive.
Another thing that springs to mind is crystals and oscillators, vibration could be getting to them.