Hi,
Thanks for the reply guys, and this is interesting information. I had assumed that if we had a paste spec, and this was the spec for the solder to reflow correctly, then if I achieved that spec, the solder should perform correctly. However this seems not the case. Oops!
For some more info, we are using SAC solder, lead free and the BGA is very large with balls around the outside only, which makes it less mechanically strong, and prone to high stress forces around the BGA edge when the PCB is warped in any way.
We are profiling by using high temperature soldering of the thermocouple to the BGA ball via a hole in from the bottom side of the PCB. I am not sure if we have covered up the hole on the rear and I am going to check that. (I am wondering if this could be showing the thermocouple hotter than it is by heat energy leaking in via the hole from the back side.)
We have a whole panel of product we ran and then thermocoupled that board, so we should have an accurate simulation of a real product running. Everything is RoHS.
The failure was a clean snap of the BGA part off the PCB, with all the solder on the BGA itself with the pad on the PCB clean (gold) without any solder on it. This is what made me think we must be too cool on the profile, as the gold should be consumed in the joint, and to see the gold pads exposed like that worried me.
However we have been distracted and thought there was other causes of the failures because when I looked into the problem the guys said they had dropped the profiler in the oven at one time previously, and did not know if it was still accurate. It was annoying to find that out just when you needed it to solve a problem.
We then questioned the accuracy of the profiler. So to eliminate this from ever being a problem again, I purchased another identical model profiler, so we would have a "standard" one, and the one that's used. Then we could check them any time. However this was not an issue, as they were both matched within 1 deg/c, so we knew we had an accurate profiler.
We also have only a 5 zone oven with passive cooling. This was really struggling with RoHS and the boards we do. Some boards have over 2500 parts per board in a very small space.
We have been working that problem for about 8 months building a new factory, so there was a thinking that things would just become better once we had the new oven up and running. We are just about to move into a new factory on Monday that's got the space for the shiny new 10 zone oven with active cooling.
However when I saw the way the BGA's were breaking off in customer returned product, I realized this was not the cause, as we should not be getting the clean break like this if the solder had melted into the PCB pad correctly, and I realized something else was wrong with our profile itself.
It's funny how sometimes these issues can be caused by multiple problems at the same time, and these issues were important, but distracting us from a more simple and obvious problem.
It was simply our procedures, and my personal knowledge of how to reflow profile a product. I embarrassingly realized I had something wrong with the profile itself. I don't have a process engineer, so have to do all this myself, and sometimes I miss things because I have to run the company, and cannot spend enough time thinking about the issues in production as I should.
This is why after experiencing this, I though it was a good idea to post on SMTnet, so I could get peoples advice because in these things I am a bit isolated. I also have questions about reflow profile development, as I am obviously missing some knowledge here.
I will check out the archives for Hybrid Profile and see what shows up. Thanks for the guidance.
Regards,
Grant
reply »