Hacker News Reader

Comment by femto | Hacker News Reader

A nice video, though it missed one of the most interesting aspects of BGA packages: their ability to self-centre during soldering. If the misalignment is less extreme than in the video, surface tension will pull the BGA package into alignment. Here is an example

https://www.youtube.com/watch?v=dz7ltWBDm7U

Notice how in the second part of the video, once the solder has melted, the body of the BGA moves so it is correctly centred over the PCB pads. This is happening by itself, without an external influence pushing the chip into position.

View on HN

This is true for most surface mount packages, the surface tension of the solder will usually center the parts in their pads during reflow.

This. In my experience QFN-like packages self center better than comparable BGAs, it’s just that BGAs in general are big so they are easy to line up.

Small BGAs can be hell to hand solder.

I find SOICs to be by far the most forgiving; off by anything less than half the pitch and it tends to jump into place.

In fact, the force of self centering is more a problem for small 0402 parts!!!!

The force is so strong, it can 'Tombstone' the piece during reflow. This is partially why 0402 is used rather than the more electrically convenient 0204. It's easier to tombstone on the long edge.

-------

Also, if you don't care about tombstoning but need the better parasitics of a 0204 capacitor or the better heat capacities of a 0204 resistor, try them out!!

Surface tension for the win.

In the extreme misalignment example they also did not use flux so I guess they were purposefully showing what NOT soldered looks like. They did show how the solder flows in the "too much solder paste" example.

Yup, when we rework BGA boards it's sufficient to go off the silkscreen outline of the package, if the board has one.

Also the cool-down BGA effect they showed is a lead-free thing, you don't see that on leaded solder balls unless something is wrong.

As someone who doesn't solder or really do much with hardware at all, this answers a longtime question of mine. I always wondered how people got all their stuff aligned perfectly ;)

you have to have the right amount of solder on 100% of the pads for this to work correctly though

Do you? I'd imagine you can have too little on some of the pads, such that the melted material doesn't make contact.

As long as others pads have enough they'd pull the BGA into place, and at that point those pads would make contact and complete the solder.

And if it doesn't work on all pads, you may need an X-ray to figure it out.

Live by the balls; die by the balls.

Most of my soldering errors are from incorrect amount of solder, so, BGA tends to work better than other footprints.

In BGA soldering, the concept of partial alignment doesn’t really happen. Here’s why: In a 2D plane (which is the PCB), only two points are needed to define alignment. If two solder balls are aligned with their pads, surface tension in the reflow process ensures the entire BGA package aligns with the rest of the pads automatically.

Surface tension acts on all the solder balls, pulling the package into place along both the X and Y axes. So once the first few balls are in place, the whole thing “snaps” into alignment. Misaligning only some pads while others are perfectly aligned isn’t possible unless something’s seriously wrong, like a warped board or damaged package.

It’s an all-or-nothing situation—either everything aligns, or nothing does.

> Misaligning only some pads while others are perfectly aligned isn’t possible

It's possible when your pcb-making process is on the odge of what designer wanted. Some pads on some boards may be not perfectly aligned in such cases. Or the board was heated not enough or in too short time and not all balls melted properly. Or it cracked under stresses because designer put too much vias in one place near the chip.

> It’s an all-or-nothing situation—either everything aligns, or nothing does.

In theory, practice and theory agrees. In practice, it sometimes does not.

I believe you forgot to quote "unless something’s seriously wrong". Your counter examples seem to describe exactly that: something going seriously wrong.

No wonder computers are binary.

I've heard of head-in-pillow defects where the solder may touch but fails to wet the pad on the chip or board. BGA probably makes it almost impossible to nudge a leg to check for mechanically solid connections. I don't know if bad joints from the factory are a problem in practice, or they usually only crack in use.

The failure mode that I worry about and would need X-ray to detect is that temporary misalignment will merge/bridge some of the solder balls.

I like to "wiggle" the chip with tweezers (under hot air) to see surface tension in action and judge whether the balls have all melted, but a tiny bit too much wiggle causes that problem.

Indeed, the only common issue we have on BGA rework is bridged balls. If it's on the edge and it's a larger BGA, we can often wick it out with desoldering braid, cut up some fine solder wire, push it under, and "reball" it in-place. If it's on an inner row you're removing the BGA.

Reply on news.ycombinator.com