Butchering QFN-64 Packages, best way to improve

[Glenn0010]

S I have absolutely butchered 2 QFN 64 packages on expensive pcbs for my project. I often struggle with getting rid of the last tiny solder bridge. The best method I've found to deal with this, is to add more solder a lot of flux and just wick it off that seems to help a lot.

Now my question is how to improve. I am think about ordering 5 atmegas (the cheapest QFN 64 ) I could find and a bunch of cheap PCBs for them to practice

Any other suggestions?

[Bassman59]

S I have absolutely butchered 2 QFN 64 packages on expensive pcbs for my project. I often struggle with getting rid of the last tiny solder bridge. The best method I've found to deal with this, is to add more solder a lot of flux and just wick it off that seems to help a lot.

Now my question is how to improve. I am think about ordering 5 atmegas (the cheapest QFN 64 ) I could find and a bunch of cheap PCBs for them to practice

Any other suggestions?

A stencil and solder paste are your friends.

[abraxa]

Are you using a hollow point soldering tip? For me, those work best when there's the right amount of solder in the cavity, just so that it runs out when I reach the end of the row. That way, bridges are a non-issue. They're still easy to wick away either way, so it's hollow point all the way for me.

[Glenn0010]

S I have absolutely butchered 2 QFN 64 packages on expensive pcbs for my project. I often struggle with getting rid of the last tiny solder bridge. The best method I've found to deal with this, is to add more solder a lot of flux and just wick it off that seems to help a lot.

Now my question is how to improve. I am think about ordering 5 atmegas (the cheapest QFN 64 ) I could find and a bunch of cheap PCBs for them to practice

Any other suggestions?

A stencil and solder paste are your friends.

I tried using a stencil but to be honest 2 opposite sides were perfect, while the other 2 opposite sides had a lot of bridges. Was my first time using a stencil so I probably did something wrong in my application, maybe too much paste?

[Glenn0010]

I am just using a bevel tip iron, I'll see if I can acquire one and give it a good, sound awesome in principle! Thanks for the tip

[Howardlong]

Do the chip’s pads completely wrap around bottom to side, or are they interrupted with plastic?

It’s much easier to consistently hand solder when the pads completely wrap around.

Don’t be afraid to make mistakes, it can take a while to get the right knack.

I had to do a ton of rework on some QFN devices that didn’t have wrap around pads, with about a 7% failure rate on the device. It was a PITA to get it right and I never really got a 100% foolproof knack after fixing hundreds of them, but it was good enough.

When I switched to a new device that has wraparound pads, I have a similar failure rate, but soldering them is far easier. As it’s rework, I re-tin the pads and powerpad, apply flux and reflow with a hot air iron, then drag solder to clean up as required. I’ve done thousands this way now.

For a new board, if I have a stencil, then that simplifies things, apply paste and reflow. For prototypes, I tin the pads, apply flux, reflow, then drag solder as necessary.

[Glenn0010]

They do complete wrap around, However I will try your technique my applying solder to the pads and re-flowing it with hot air.

[Howardlong]

To be clear, I don’t tin the device’s pads, only the lands on the board. I then use a flux pen on both the board and the untinned part, place the part, and reflow under the microscope.

[Glenn0010]

Definitely understood I'll give it a go

[KL27x]

A bevel tip is pretty good for QFN with the wraparound sidepads. I tend to use bevel tip or knife tip for handsoldering and fixing the bridges on QFN.

If hot air reflow, I like to pre tin the pads similar to HowardLong. But I find I can get enough of a bead of rosin flux on the pcb that I don't have to bother explicitly fluxing the chip. This is using a pointy plastic tipped syringe, not a felt tip pen. The one pearl I can offer is that if you don't put enough solder on the center pad, the chip will suck to the pcb and won't auto-center. I prefer the mistake of putting too much solder on the center pad, and just squishing the chip down with tweezers while reflowing. Once it flows, let go, and the chip will center. Any excess solder will have squeezed out and balled up (and possibly bridged) to some of the outer pads, which necessitates post hot-air inspection and rework with the iron. So I don't think your stencil/oven result was a disaster, at all. If the chips were centered, you did fine in my book.

It is not always necessary to solder the center pad on a microcontroller. If I can omit the center pad, I find it is usually easier just to hand solder the pads. Here, I much prefer the knife tip. One of the reasons is you can use the tip as a probe to hold the chip down while adjusting it with the tweezers.

I've just hand-soldered 50 chips (no center pad), and it's been a few months. So I have been paying particular attention while relearning how to do this.

After fluxing and placing the chip

1. Push the chip roughly into place with the tweezers.
2. Pin it down with the iron tip (this is one reason why I prefer knife tip for this... pointy tip, no solder blob left on the chip, no overheating the chip.)
3. Focus on a single pad. For me it is the front right pad.
4. While pinning the chip down with the iron tip, I adjust the chip with tweezers until that one single pad is lined up right.  (I tilt the microscope forwards by about 10 degrees so I can see the side pad is centered on the pcb pad. )
5. When it looks right, I switch implements. I pin the chip down with the tweezers, then I can remove the iron tip
6. I solder the single front right pad with the tip of the knife, bringing the tinned tip to the fluxed pads.
7. Solder the front left pad, tweaking the chip as necessary, to straighten it out. You can even walk the chip up or down by reflowing either end, if you got it grossly too far up/down.
8. Pin with tweezers, again, and drag solder the entire row/side.

After that side is down, then it's just a matter of drag soldering the other 3 sides. I turn the pcb for each row, so that I get the same view into the joints. If you got the chip too far out of whack and some of the connections are just not taking, you can hit the thing with hot air and watch it pop into place.

[Howardlong]

Here's a rework video I did a while ago.

From about half way through, you can see where I prep the board first and go through the steps I mentioned.

https://youtu.be/YCRiVzsZRVY

[AndyC_772]

What are your lighting and magnification tools like?

IMHO there's no better way to improve your technique than to be able to see, really clearly, what you're doing. I use a 3d microscope tilted at a slight angle to the vertical, and a bench lamp with two bright LED heads on goosenecks.

Once you know exactly where your solder is flowing and when, you'll be able to see for yourself what other changes (if any) will help achieve consistent results. Without good visibility, you're just guessing.

[Howardlong]

What are your lighting and magnification tools like?

IMHO there's no better way to improve your technique than to be able to see, really clearly, what you're doing. I use a 3d microscope tilted at a slight angle to the vertical, and a bench lamp with two bright LED heads on goosenecks.

Once you know exactly where your solder is flowing and when, you'll be able to see for yourself what other changes (if any) will help achieve consistent results. Without good visibility, you're just guessing.

I totally agree with this, and it oretty much reflects my setup, although I use a stereoscopic zoom microscope, not a 3D Mantis type, I simply don’t have the bench space for a Mantis.

The tilt allows you to look at the sides better, avoiding parallax error, but you can also achieve a similar thing (but a little less effective) by moving the object off centre. The problem of tilting is that in narrow depth-of-field situations, you risk loosing focus off bore-sight. There’s a compromise to everything!

[AndyC_772]

Sorry, I meant stereoscopic, not a 'heads up' unit like the Mantis.