| I need some advice concerning reflow of a BGA-292. Since I am not ready to change my entire process to no-clean, I decided to place the lone BGA-292 on a proto run using our BGA re-work station. The pcb is a moderately dense, single-sided, 8-layer assembly. My problem is that one side of the BGA seems to "collapse" during re-flow. The part is truly lower on one side than the other. The "collapsed" side has shorts and the balls are about 25% larger (x-ray measurement). I started by single-site stenciling no-clean paste, placing and reflowing - result = collapse on one side. I used no-clean sticky-flux and let the eutectic balls create the joint - result = collapse on same side. Other side joints look OK visually and with x-ray. Any hints, pointers, or other assistance will be greatly appreciated. | | Mark | | Mark,
As you've pointed out, all sides must collapse equally. Using a rework capability is ok if it is done right. You seem to be doing it right with paste (printed with a micro stencil with apertures about 3 mils larger than pads for your application) or paste flux.
I've used recommended (by PHD, others, and industry) off contact reflow heater nozzle settings (anywhere from 60 to 120 mils above the board surface) and achieved excellent results. Normally, I would use this process, with a proven profile (about 100 second soak plus another 100 second equalization ramp and about a 275-295 degree C topside heater tmep and 250 botom side heater temp using the solder paste profile recommendation window [mine is 100 seconds max in the 183 window and no more than 45 seconds in the liquidous window at 205 C at a peak of 220 C, etc.]).
I say normally I would do this. However, the folks in control here say no. The think, based on dubious data, excessive intermetallics are formed using this process because the heat is exhausted from the nozzle's bottom to very closely adjacent parts. Anyway, I do it all the time in other places and have yet to have reliability issues crop up.
I guess, instead of going through all the rework stuff, I'd just go for it in production. I see new parts every day and mostly go unnoticed even while using X-Ray on every prototype run. If problems do arise, as those you indicate, they usually are associated with DFM problems. No matter, all sides have to reflow equally to be acceptable.
Earl Moon
reply »