Not properly soldered boards

[buenos]

Hi

I got 8 prototypes of my new board design, and 6 of them have intermittent problems.
I suspect that the soldering was not done properly.
This is a double sided SMT board with 7000 components and maybe 60 BGAs.
In the morning when the board is cold those 6 don't start up, they fail with memory initialization. If I heat the memory with an airgun (maybe 150C), then they start up.
I suspect a reflow soldering issue, as I have actually found QFP/SSOP components with unsoldered pins.
What do you think, do the symptoms look like soldering issue?
Can we manually reheat the memory chips with a soldering-airgun (to 280C) to fix this?
How do you adjust the reflow process (th.profile, paste ball size, stencil thickness...)?

 

[tpetar]

Some chips can only be used a few times in the reflow process. Preheating is needed.

Maybe its best to use chip datasheet or contact manufacturer to get reflow informations.

If you have 6 prototypes boards with that problem, I think its no problem to test one board.


Best regards,
Peter

 

[andre_luis]

buenos,


It happens usually if production scale is small, such as prototype, and occurs to be not enough margin to calibrate production process.
If the problem is not caused due to insufficient solder, you can try fix it by performing a hot air blow over components.


+++

 

[buenos]

"you can try fix it by performing a hot air blow over components"
-I am thinking on it, but is there anything to consider to maximize the chance of success and avoid further damage?

 

[andre_luis]

I confess it is not parametrized for the guys on repair center were I work, and is guided for personal feeling of each one, due not only the temperature must be considered, but also distance from the pistol and even the geometry of assembled PCB, when certain components must be avoided to heat.

The only remark is that heat must be enough to meld solder, but must take care to do not fuse internal gold wires of IC´s which connects external leads to internal pads of silicon chips.

The procedure adopted is to perform a slow overall heating over entire board up to a temperature relatively small ( let´s say, few more than 80oC ) and suddenly apply a overheating focused only on target components, during few seconds.


+++

 

[123jack]

If you have evidence for solder issues you should contact your manufacturer.
Any reputable company will be more than happy to sort out the issue.
Unless you went to some cheap backstreet hack shop (and why would you for 8 complex prototypes)
my guess would be more likely design issues - especially with such a complex sounding board.
Second guess would be a bad batch of a specific component - again talk to the board maker.
Simply reheating a board with 7000 SMT components does not sound like a good way to fix this to me.
(and of course you cant go back afterwards if your inexperience with soldering ruins them)