Rectangular uBGAs with 1 raised edge after reflow

Are there other components on the board? If so, how does solder quality look on them? Are they all reflowing? Are the sections of the uBGA�s sitting on a ground plane? Sounds like un-even heating. You should attach a couple thermal couples and see how the area is heating. attach one at the end that is low and one where it is high. Just looking at your temps it doesn�t look like the board would be above liquidus very long you only have one zone where the board would actually exceed 183. Remember the temps you set your oven to does not indicate what the board temp will reach. Run a profile and see what your ramp rates, soak time and time above liq are. Good place to start. Your paste supplier should have a recommended profile.

Good Luck

reply »

Good problem description.

Your profile seems reasonable. Where did you take it? Thermocouples on: * Reflow oven internals * Pads were the solder balls are not collapsing * Another location

Intel publishes a BIG pdf on CSP flash memory for gear heads at: http://developer.intel.com/design/flcomp/packdata/csp/298161.htm

It seems easy to guess to shoot => slowing-down the conveyor to allow the solder to heat these balls more. But that may not be the answer, because if the flux is exhausted, more heat will not help. Then if the flux is exhausted, the answer is to crank the belt. Two different answers to the same question, hmmm.

reply »

Gregory, I think that the profile is too radical, i.e. zone 6 to zone 7 temperatures are too far apart. I would guess that the solderjoints have not all reached 183C in zone 6 which is why the uBGA's do not have a symmetrical ball collapse. You may want to increase zone 5 by 5 degrees and zone 6 by 5 or 10 degrees. Keep in mind that the set point for zone 7 (260 degrees) may be too high and may yield excessive board temps. Follow the solderpaste specs. and the BGA case temp. specs., typically 215C and 220C max. respectfully. An aggressive profile is not desirable for BGA's. You may need to slow down the belt speed and lower temps to yield a lower temp slope of around 1 degrees C/second. This will also prevent the BGA's from "potato-chipping" where the package becomes warped causing opens or shorts at the corners. Keep in mind that the paste may dry out if the profile is not within the mfgrs. specs. We have found that some pastes can withstand pushing their envelopes, thus widening their spec. windows a good solderjoint quality throughout all areas of the board. Another factor that just came to mind is "Has the oven been on for a long time?" Even though the oven temps might have been reached, the drive chain, etc. may not be up to temp. Convection ovens have cold spots as well as the substrates (pcbs)being reflowed due to heavy ground planes, etc. Our oven's fixed rail at the edge of the oven is a few degrees cooler than the variable one in the middle. Try rotating the board before reflowing it. Another factor, PCB thickness needs to be considered. A normal pcb (.062" thick) will heat up faster than a thick pcb (e.g. .120"). That's all folks! Happy Baking! Jay

reply »

Hi Greg,

Have you resolved your problem yet? Daves reply to slow the belt may be your best bet. It appears you are not heating the BGA's uniformily. The temp in zone 6 seems to be a bit odd, are you trying to prolong the soak? If so, try to increase this temp and lower zone 7 a bit. it's possible you are drying out the paste. you may also speed up the belt a bit and raise the temps in the 5 & 6 zones.

Greg

reply »