SMD BGA joint failure

[theAnonymous1]

Anyone here have any personal experience with BGA (ball grid array) solder joint failure?

I have an ATI video card that was acting up and causing artifacts to show up on the monitor. I could cause more artifacts if I just barely wiggled the video card.

After removing it numerous times and visually inspecting it, I could only assume there was a failure in the joints of one of the BGA memory chips.

Since I figured I didn't have anything to lose, I fired up my mini butane torch. I went over each memory chip until I thought it was about to start smoking, and then immediately stopped.

I let the board cool and then reinstalled it. To my surprise not only did the card still work, but the problem seems to have disappeared.

So, anyone else hate BGA?

[mightydub]

It is possible for BGA solder joints to fail, especially with mechanical stress. Your problem could also have been a crack in a pcb trace or via.

Things that can cause BGA solder attach to go wrong:

- incorrect thermal profile for the reflow oven
- not enough solder paste put down on the pad
- mechanical stress causing the joint to crack
- board or package not planar (warped)

Generally BGA joint failure is very rare. The manufacturer who assembles the board will run a series of boards to make sure the reflow thermal profile is correct. Manufacturing qualification for the chips includes shear strength and pull tests on the solder balls. 3D xray will be used to check for proper solder joint formation, and sometimes board cross sections are done. I have seen BGA failures due to rough handling, and too much heatsink pressure. But there's no better way to do high pin count packages, and the manufacturing process is well developed and understood.

Sounds like you managed to do a brute force reflow, I'm surprised you didn't cause other damage in the process!

[theAnonymous1]

This isn't the first product I've owned with problems from a BGA device. The other product had a high failure rate attributed to a BGA package IC so the manufacturer switched to a leaded version of the same IC.

You can find plenty of videos on the web of various problems from failed BGA solder joints. I think the problem is less prone with leaded devices because the leads provide a little stress relief since they are somewhat flexible.

Originally posted by mightydub
Sounds like you managed to do a brute force reflow, I'm surprised you didn't cause other damage in the process!

That's what I was going for. I didn't care if I ruined it, it was already on it's way to the trash. I'm glad it worked though, saved me around $100.

[dozer]

Originally posted by mightydub

Things that can cause BGA solder attach to go wrong:

- incorrect thermal profile for the reflow oven
- not enough solder paste put down on the pad
- mechanical stress causing the joint to crack
- board or package not planar (warped)

Generally BGA joint failure is very rare.

Two corrections to add here....(from someone who's recently retired from 30+ years in the electronics industry).....

1) Those causes listed above are all valid, but minor causes statistically.

THE number-one cause of BGA seperation is thermal-cycling. Not just running the parts/boards hot....but -cycling- them between hot and cold. The repeated expansion/contraction cycles pull the bond apart. Running hot does exacerbate this because the solders used tend to crystallize when held at higher temps for long periods.

2) I'm sorry, but generally BGA joint failure is very common indeed!

Entire series of products exhibit -exactly- this type of failure; by the -millions-. The #1 cause of failure in the otherwise excellent Thinkpad T40/T41/T42/T43 laptops is exactly BGA failure under the ATI video chip.

One might even suspect the ATI package to be the culprit....except that BGA-bond failure also occurs in volume with so many other products and chips.

The thinkpads.com forum has several very extensive threads on the failures themselves; and also numerous techniques for home-reflow.

It's a "nothing to lose" situation, as the OP himself noted....so why not give it a try?

uhhh.....but note that an industrial heat-gun is more likely to do the job, and more controllable in re peak-temp...just a -bit- safer for your chips...

Richard

[mightydub]

Richard,

You bring up a good point, I had forgotten to mention thermal cycling. Perhaps because in all of the products I have been involved with over the years, there has been thorough testing and engineering/manufacturing qualification to make sure that thermal cycling is not an issue. I have released multiple board and chip level products to manufacturing (just completed the release cycle on a 900 ball BGA and several associated boards), and never had any field failures due to BGA thermal cycles. That's because of the test and qualification cycle - thermal shock, temp cycle, mechanical shock, autoclave, ball pull strength, ball shear strength, etc. to make sure that any failure mechanisms are identified and corrected. A high quality contract manufacturer helps too.

As you point out, it does happen, especially in high volume consumer products. Unfortunately the attitude toward these products is that they are disposable, and the goal is to get new models out on a 6 to 9 month cycle so they are often rushed to production to meet a product cycle deadline without sufficient qualification testing. There's one large consumer electronics company known for its particular sense of style that is especially guilty of this and has the warranty return rates to prove it.

Congratulations on your retirement - I'm not quite to 30 years in this industry yet (27) but getting there. Now you have more time for DIY, I hope you're working on something big!

[dozer]

thanks for the congrats mightydub.

I wouldn't have done it voluntarily at this time; but I've been having a degenerative problem with my eyes for the past several years.

I finally admitted to myself this winter that it's not going to ever get better. I can't effectively do fine-detail work any more; and these days it's ALL fine-detail....lol. So....retirement time.....and thus the reason you see TE for sale in my sig.....no sense keeping it any more...when others could make good use of it.

But I'm enjoying the hell out of retirement. We're now 'watchmen' caretaking a gold-mine in Oregon....outdoors all the time now....in the mountains....working on 70,000 lb excavators and dozers and such....doing some welding and machining....exercising a whole different set of skills which had languished for a long time.

I hope your own retirement in a few years brings you as much joy as mine is proving to,

Richard

[theAnonymous1]

Surprisingly it wasn't the ATI GPU, it was at least one of the 8 memory chips. That makes sense though, they have a much more extreme thermal cycle as there are not any heatsinks on them.

Originally posted by dozer
but note that an industrial heat-gun is more likely to do the job, and more controllable in re peak-temp...just a -bit- safer for your chips.

I don't own a heat gun, or even a hair dryer (no need, no hair). I actually kind of expected that I would break the thing even further, so I was really shocked when I turned it on and it worked.

[mightydub]

Originally posted by dozer

But I'm enjoying the hell out of retirement. We're now 'watchmen' caretaking a gold-mine in Oregon....outdoors all the time now....in the mountains....working on 70,000 lb excavators and dozers and such....doing some welding and machining....exercising a whole different set of skills which had languished for a long time.

[/B]

That sounds really cool. The way things are going in Silicon Valley I may be taking early retirement, packing up my tools, and driving north to join you.

(off topic, I know. )

[BWRX]

Originally posted by theAnonymous1
Surprisingly it wasn't the ATI GPU, it was at least one of the 8 memory chips.

One day my laptop started displaying strange artifacts on the LCD for no reason at all. Sometimes it would be fine for a couple days then start doing it again. I eventually narrowed the problem down to the VRAM and assumed it was a solder joint that went bad on one of the memory chips. I tried the hot air reflow method with no luck. Unfortunately, you need an x-ray machine to check the integrity of the BGA solder joints. Ended up paying more than I wanted to for a used logic board but it was cheaper than a whole new computer.

Glad to hear you ended up fixing yours!