Suspect BGAs

We are having a small number (~2%) of our boards pass our manufacturing diagnostics, fail once they reach the customer, fail when they get back to us and then start working again after handling them for a short time. We have eliminated a lot of items and the big suspicion right now is our BGAs (two 256-pin plastic packages).

We are getting ready to send some boards out to be cross- sectioned so that we can see what is going on under the BGAs. We are also thinking about just ripping a couple of BGAs off of a board. Is there anything we can tell from that? (e.g. on a good attachement should all of the balls rip PCB material with them, or will some have the solder joint fail?) Are there any other quick checks we can do to see if there are problems with our BGAs?

Thanks, Jeff Dahlin

reply »

| We are having a small number (~2%) of our boards pass our manufacturing diagnostics, fail once they reach the customer, fail when they get back to us and then start working again after handling them for a short time. We have eliminated a lot of items and the big suspicion right now is our BGAs (two 256-pin plastic packages). | | We are getting ready to send some boards out to be cross- sectioned so that we can see what is going on under the BGAs. We are also thinking about just ripping a couple of BGAs off of a board. Is there anything we can tell from that? (e.g. on a good attachement should all of the balls rip PCB material with them, or will some have the solder joint fail?) Are there any other quick checks we can do to see if there are problems with our BGAs? | | Thanks, | Jeff Dahlin | | | Jeff,

We are running and reworking, occassionally, 432 permiter super BGA's, PBGA's, micro BGA's, CCGA's, and most all else. Our electronic assembly development center has done lots of home work, as have many others in industry, and have come up with some fairly precise rules to prevent defect. Enough of that, though they are more often right than wrong - by high percentages.

When we receive customer returns (all our customers are ours from other divisions within the company), we undertake repair/rework. When electrical testing clearly indicates BGA shorting problems, as an example, we remove the part while running TC profiles, inspect the site after removal to determine profile effects (too hot/cold), clean the site, paste or not, place and reflow the part, and go immediately to X-Ray to confirm rework acceptability.

I think what I'm really trying to say is X-Ray laminography (the ability to "slice" through a BGA's solder joint vertically) first provides the ability not to have to X-Section (a very expensive but highly graphic process). We can see all the solder joints from part to board level in vertical passes. The machine points to pre-programmed acceptance specification violations as excess solder, insuficient solder, excessive voiding, cracks, etc.

I know where you are coming from concerning physically pulling a part. You might see a complete crack. However, you also might cause a crack resulting from internal solder joint voiding.

In conclusion, this is where a very good X-Ray capability has its advantages, provided you don't have to pay for one of the damn things. After X-Ray, and upon finding the suspected failure location, X-Sectioning is a logical next step if only to find the failure mechanism as entrapped flux, oxides, or other contamination.

There's more, but this is the simplest explanation I can provide after a long day of doing what you are proposing.

Earl Moon

reply »

| | We are having a small number (~2%) of our boards pass our manufacturing diagnostics, fail once they reach the customer, fail when they get back to us and then start working again after handling them for a short time. We have eliminated a lot of items and the big suspicion right now is our BGAs (two 256-pin plastic packages). | | | | We are getting ready to send some boards out to be cross- sectioned so that we can see what is going on under the BGAs. We are also thinking about just ripping a couple of BGAs off of a board. Is there anything we can tell from that? (e.g. on a good attachement should all of the balls rip PCB material with them, or will some have the solder joint fail?) Are there any other quick checks we can do to see if there are problems with our BGAs? | | | | Thanks, | | Jeff Dahlin | | | | | | | Jeff, | | We are running and reworking, occassionally, 432 permiter super BGA's, PBGA's, micro BGA's, CCGA's, and most all else. Our electronic assembly development center has done lots of home work, as have many others in industry, and have come up with some fairly precise rules to prevent defect. Enough of that, though they are more often right than wrong - by high percentages. | | When we receive customer returns (all our customers are ours from other divisions within the company), we undertake repair/rework. When electrical testing clearly indicates BGA shorting problems, as an example, we remove the part while running TC profiles, inspect the site after removal to determine profile effects (too hot/cold), clean the site, paste or not, place and reflow the part, and go immediately to X-Ray to confirm rework acceptability. | | I think what I'm really trying to say is X-Ray laminography (the ability to "slice" through a BGA's solder joint vertically) first provides the ability not to have to X-Section (a very expensive but highly graphic process). We can see all the solder joints from part to board level in vertical passes. The machine points to pre-programmed acceptance specification violations as excess solder, insuficient solder, excessive voiding, cracks, etc. | | I know where you are coming from concerning physically pulling a part. You might see a complete crack. However, you also might cause a crack resulting from internal solder joint voiding. | | In conclusion, this is where a very good X-Ray capability has its advantages, provided you don't have to pay for one of the damn things. After X-Ray, and upon finding the suspected failure location, X-Sectioning is a logical next step if only to find the failure mechanism as entrapped flux, oxides, or other contamination. | | There's more, but this is the simplest explanation I can provide after a long day of doing what you are proposing. | | Earl Moon |

Hi,

We have X-ray laminography and I totally agree with what is being said here except that Earl Moon has also stated that cracking can also be detected. I find that hard to believe, what FOV are you running at? If you have captured some images of defective bga joints I would be interested in seeing them. We have a moderate bga production but other than an occasional short I haven't seen any other types of defects.

Joe

reply »

| We are having a small number (~2%) of our boards pass our manufacturing diagnostics, fail once they reach the customer, fail when they get back to us and then start working again after handling them for a short time. We have eliminated a lot of items and the big suspicion right now is our BGAs (two 256-pin plastic packages). | | We are getting ready to send some boards out to be cross- sectioned so that we can see what is going on under the BGAs. We are also thinking about just ripping a couple of BGAs off of a board. Is there anything we can tell from that? (e.g. on a good attachement should all of the balls rip PCB material with them, or will some have the solder joint fail?) Are there any other quick checks we can do to see if there are problems with our BGAs? | | Thanks, | Jeff Dahlin | Jeff: Another alternative to consider when assessing "intermittant" problems like is describe is: ionic contamination. Good luck Dave F

reply »

Another test to consider is dye penetrant analysis. The technique involves flooding the area beneath the soldered part with dye (preferably a bright color, like red). After drying the board+dye in an oven, you peel or torque the part off and inspect what's left. The dye should have colored any brittle fracture (crack) surfaces, as well as any unwetted pads.

Check out Terry Burnette's posting on 1/21/99 -- it discusses "secondary reflow" of the BGA joints during wave soldering. I saw this problem recently when a wave profile was running too hot, resulting in solder opens and visibly tilted parts. Could this be the root-cause of your problem?

Good luck!

| We are having a small number (~2%) of our boards pass our manufacturing diagnostics, fail once they reach the customer, fail when they get back to us and then start working again after handling them for a short time. We have eliminated a lot of items and the big suspicion right now is our BGAs (two 256-pin plastic packages). | | We are getting ready to send some boards out to be cross- sectioned so that we can see what is going on under the BGAs. We are also thinking about just ripping a couple of BGAs off of a board. Is there anything we can tell from that? (e.g. on a good attachement should all of the balls rip PCB material with them, or will some have the solder joint fail?) Are there any other quick checks we can do to see if there are problems with our BGAs? | | Thanks, | Jeff Dahlin | | |

reply »

| | | We are having a small number (~2%) of our boards pass our manufacturing diagnostics, fail once they reach the customer, fail when they get back to us and then start working again after handling them for a short time. We have eliminated a lot of items and the big suspicion right now is our BGAs (two 256-pin plastic packages). | | | | | | We are getting ready to send some boards out to be cross- sectioned so that we can see what is going on under the BGAs. We are also thinking about just ripping a couple of BGAs off of a board. Is there anything we can tell from that? (e.g. on a good attachement should all of the balls rip PCB material with them, or will some have the solder joint fail?) Are there any other quick checks we can do to see if there are problems with our BGAs? | | | | | | Thanks, | | | Jeff Dahlin | | | | | | | | | | | Jeff, | | | | We are running and reworking, occassionally, 432 permiter super BGA's, PBGA's, micro BGA's, CCGA's, and most all else. Our electronic assembly development center has done lots of home work, as have many others in industry, and have come up with some fairly precise rules to prevent defect. Enough of that, though they are more often right than wrong - by high percentages. | | | | When we receive customer returns (all our customers are ours from other divisions within the company), we undertake repair/rework. When electrical testing clearly indicates BGA shorting problems, as an example, we remove the part while running TC profiles, inspect the site after removal to determine profile effects (too hot/cold), clean the site, paste or not, place and reflow the part, and go immediately to X-Ray to confirm rework acceptability. | | | | I think what I'm really trying to say is X-Ray laminography (the ability to "slice" through a BGA's solder joint vertically) first provides the ability not to have to X-Section (a very expensive but highly graphic process). We can see all the solder joints from part to board level in vertical passes. The machine points to pre-programmed acceptance specification violations as excess solder, insuficient solder, excessive voiding, cracks, etc. | | | | I know where you are coming from concerning physically pulling a part. You might see a complete crack. However, you also might cause a crack resulting from internal solder joint voiding. | | | | In conclusion, this is where a very good X-Ray capability has its advantages, provided you don't have to pay for one of the damn things. After X-Ray, and upon finding the suspected failure location, X-Sectioning is a logical next step if only to find the failure mechanism as entrapped flux, oxides, or other contamination. | | | | There's more, but this is the simplest explanation I can provide after a long day of doing what you are proposing. | | | | Earl Moon | | | | Hi, | | We have X-ray laminography and I totally agree with what is being said here except that Earl Moon has also stated that cracking can also be detected. I find that hard to believe, what FOV are you running at? If you have captured some images of defective bga joints I would be interested in seeing them. We have a moderate bga production but other than an occasional short I haven't seen any other types of defects. | | Joe | We draw cracking conclusions from excessive voiding and/or irregular voiding shapes. I recognize your question as one of what can really be seen with the type analysis presented. You can easily recognize voiding and void shape as well as size and volume. If voiding shapes are highly irregular (not spherical) it is possible to suspect cracking extending in or out of the void pocket.

Though not readily recognizable, using any field of view on a grey scale system, high resolution black and white X-Ray can also be employed after finding the affected area as well as X-Sectional analysis. This is true notwithstanding using TDR testing to determine impedance characteristics on test panels. All this saves anyone the pain and expense of X-Sectioning hundreds of solder joints.

We have done studies indicating certain void characteristics, under certain stress conditions most likely will produce cracks. Therefore, we have looked very close at these occurences and under what conditions, and what is exhibited during X-Ray laminography. Follow up has been done using X-Sectioning or additional reflow to "heal" the affected solder joint. This may be done during rework followed by 100% X-Ray of affected parts and boards.

As very few voiding occurences are found using rework sequences without solder paste, we key on those occurences using paste. Often this is correlated with data gathered during initial studies conducted during R&D before any release to production.

We use only two reflow profiles to produce over 1500 assembly numbers with a very wide part mix. Our success is based on doing the studies up front to prevent reacting to defect by using any measure let alone extreme.

Your point, when specifically pointed out as you have done so well, is well taken. I'm simply suggesting doing the work before production and use the most helpful tools without going to extremes. This requires the foregoing as well as using a well thought out plan with significant experience.

Earl Moon

reply »

| We are having a small number (~2%) of our boards pass our manufacturing diagnostics, fail once they reach the customer, fail when they get back to us and then start working again after handling them for a short time. We have eliminated a lot of items and the big suspicion right now is our BGAs (two 256-pin plastic packages). | | We are getting ready to send some boards out to be cross- sectioned so that we can see what is going on under the BGAs. We are also thinking about just ripping a couple of BGAs off of a board. Is there anything we can tell from that? (e.g. on a good attachement should all of the balls rip PCB material with them, or will some have the solder joint fail?) Are there any other quick checks we can do to see if there are problems with our BGAs? | | Thanks, | Jeff Dahlin | Jeff Take a Dremmel and slowly grind off the top of the part and stop just short of the solder joint's. You can then pull of the bottom layer of the BGA and see what is going on. Good luck Mike |

reply »