BGA replacement success rate

[Ice-Tea]

Hello folks,

Quick question: I was told a supplier can't swap BGA's (35x35 balls or so) because he expects a 30% failure rate. Which he considers quite good and claims his competitors would do only 50%. I can't seem to find any statistics on BGA replacements but this seems... High. Any experiences you can share?

Ice-Tea

[Chriss]

Hmmmm.
If I understand your question, the bga part have 35x35 balls and that part must be replaced, and you wish to know how big is the risk to success the rework?
If this is the question I will rate it around 50% of success if you don't have much experience. If you have the success will be much higher.

I do changing bga parts in the last 15 years and today I really have high rating of success.

Sent from my GT-I8260 using Tapatalk

[tszaboo]

Once I replaced a BGA memory chip with a bigger one. The first chip had its balls crushed, and I scrapped it. Second worked. So 50%.
I've used a chinese hot air station, and I have a sample size of 1.

[AndyC_772]

BGA replacement using a high quality, fit-for-purpose rework station, by a trained operator in an ESD protected environment, is generally a very reliable operation. There's no reason why it shouldn't be.

The kind of 30-50% failure rates you're being quoted would apply to someone with a pair of tweezers and heat gun. There are a lot of people out there who claim to "know" how to rework a BGA, who don't follow anything even resembling a controlled process, but who nevertheless get away with it often enough that they don't understand why what they're doing is wrong.

The difference is between the service you'll get from a reputable contract electronics manufacturer vs a back street phone and games console repair outfit. There's a difference in price, of course. If you need the board to work afterwards, go to a good CEM. If you're on a budget, and just want to see if you can rescue some piece of consumer electronics from the bin, then the repair shop might be the only option that makes economic sense.

[wraper]

If using decent equipment and not chips from E-waste (which often is the case with repairs), then success rate will be close to 100%.

[RoGeorge]

I think the failure rate here seems big because the talk was about the failure rate measured a few months after the BGA replacement. I'll say the failure rate is even bigger, like 80-90%.

Replacing the BGA is easy, and is almost 100% success, so short term failure rate is close to zero. Long term failure rate is something different. Most of the time it won't last more than a few weeks/months. I am thinking here about reworking/reballing repair shops for laptops and phones.

Usually BGA solder balls fail because of thermal stress, so most of the time it's a thermal design problem. Being a design problem, it won't go away by simply replacing the chip, because if it failed once, it will probably fail again.

[wraper]

I think the failure rate here seems big because the talk was about the failure rate measured a few months after the BGA replacement. I'll say the failure rate is even bigger, like 80-90%.

Replacing the BGA is easy, and is almost 100% success, so short term failure rate is close to zero. Long term failure rate is something different. Most of the time it won't last more than a few weeks/months. I am thinking here about reworking/reballing repair shops for laptops and phones.

Usually BGA solder balls fail because of thermal stress

BGA solder balls fail very rarely. If we talk about things like laptop GPU rework or similar, then reball don't fix the issue because usually it's BGA chip what is faulty, to begin with. Or using salvaged reballed chips, removed from the boards which failed because of this faulty chip. Heating just rises it from the grave for a few weeks until it dies again.

so most of the time it's a thermal design problem. Being a design problem, it won't go away by simply replacing the chip, because if it failed once, it will probably fail again.

It failed because BGA chip had design issue. And repair does not last because no new replacement part is used. Even if the same with issues but NEW chip is used in those cases, repair will last at least a year or two.

[mikerj]

I think the failure rate here seems big because the talk was about the failure rate measured a few months after the BGA replacement. I'll say the failure rate is even bigger, like 80-90%.

Replacing the BGA is easy, and is almost 100% success, so short term failure rate is close to zero. Long term failure rate is something different. Most of the time it won't last more than a few weeks/months. I am thinking here about reworking/reballing repair shops for laptops and phones.

Usually BGA solder balls fail because of thermal stress, so most of the time it's a thermal design problem. Being a design problem, it won't go away by simply replacing the chip, because if it failed once, it will probably fail again.

Sounds to me like you are thinking of bad GPU repairs (e.g. on Xbox and laptops) where reflowing the devices can provide a short term fix.  Correctly replacing a BGA device certainly shouldn't cause high failure rates.

[Ice-Tea]

These are new boards and the replacements are factory-fresh. The replacements would be done at the many site, so I'm guessing they have decent equipment and decent passable operators.

From what I read here, failure rate should be much much lower than advertised, right?

Thanks for all the feedback

[Chriss]

Think about who is advertising?
With respect to these person's, but backstreet phone repair "guru's" who maybe use a car halogen bulb for rework?
Or some blog writer who even don't have practice in a real situation?

If you give your equipment to a company with reputation and of course with a normal price ranking I think your repair will be fine without worrying.

If your budget can't handle the high-end rework company then the risk to get back a bad or in worst case scenario a much more killed equipment (because of overheating the whole pcb ) is a must happen situatin.

Your chance to win is in your hand.

I hope these infos here will help you to take the right path.

My best regards.

Sent from my GT-I8260 using Tapatalk

[Rasz]

I think the failure rate here seems big because the talk was about the failure rate measured a few months after the BGA replacement. I'll say the failure rate is even bigger, like 80-90%.

Replacing the BGA is easy, and is almost 100% success, so short term failure rate is close to zero. Long term failure rate is something different. Most of the time it won't last more than a few weeks/months. I am thinking here about reworking/reballing repair shops for laptops and phones.

Usually BGA solder balls fail because of thermal stress, so most of the time it's a thermal design problem. Being a design problem, it won't go away by simply replacing the chip, because if it failed once, it will probably fail again.

whaaaa?
what you describe is morons reheating factory defective GPUs, and early days of iphone no touch replace the chip without reinforcing pcb/m1 jumper AKA BAD repair, with zero correlation to bga soldering failure rate

These are new boards and the replacements are factory-fresh. The replacements would be done at the many site, so I'm guessing they have decent equipment and decent passable operators.

From what I read here, failure rate should be much much lower than advertised, right?

Thanks for all the feedback

are we talking China small manufacturing run here? and what makes you think those BGA chips need replacement? this 30% failure rate might be accurate from the mfg line foreman perspective when client asks if replacing random big components will somehow bring bad boards back to life.

[Ice-Tea]

are we talking China small manufacturing run here? and what makes you think those BGA chips need replacement? this 30% failure rate might be accurate from the mfg line foreman perspective when client asks if replacing random big components will somehow bring bad boards back to life.

The delivery of the right BGA is taking a bit longer than expected. They don't want to postpone production so suggest to start production of lab boards with a pin and function compatible component (long story, not really relevant). I agreed but said the components should be swapped at the earliest opportunity. And then they flashed the 30% number. Which in my opinion means they are incompetent or hiding something. Or both.