SMT by hand

The problem with QFNs is the center pad and getting just the right amount of solder on it. There isn't a lot of room for error here. Too much and the part will "float" on the center pad so it won't align properly.

It should, though it's a lot harder to get right than a BGA, for instance. The center pad works against you, where in a BGA all the balls work in your favor. Since the pad is huge, in proportion to the pins, it can get dicey. Even with automatic PnP, we have far more problems with QFNs than BGAs. Of course the crappy RoHS processes make things far worse.

That is usually solved by putting several uncapped vias underneath the QFN. The vias also allow to solder the center pad manually. Preheat the board using a heat-gun and put solder in one of the vias. Wait until the solder comes up from the other via. Then you'll know the center pad is soldered.

Nope. If you make the pads a bit larger you can hand solder the part with a very fine soldering tip. I usually wrap a piece of thin solid copper wire around the soldering tip to solder the pins. Copper is still the best material for soldering tips.

Ofcourse, you'll need flux as well.

IF the pad is the problem I can just leave it off then. At this point it is not an issue(chips have thermal shutdown and I don't plan on running them hot). I have vias that I can inject some solder or thermal paste if I do have some issues with heat.

Be sure to check that the pad isn't (the only) ground. I've seen this more than once.

QFNs are a great idea but what a PITA. QFNs and interboard connectors are our two largest problems.

the

to

end

I don't have any problems hand-soldering QFN chips. I extend the pads a little and drag-solder the chip connections. I use a large via for the central pad and feed solder in with the soldering iron tip. A good soldering station like the Metcal system I use helps a lot.

Leon

is

The pad should always be soldered. If you don't, and you drop the board, the chip is likely to shear off, taking the pads with it.

Leon

For hand-soldering, put a really big untented via under it, so you can solder it from the other side later.

Jon Slaughter schrieb:

Try to avoid that. Once I had to do 20+. Not fun.

Mine were of that type:

I took a picture from each side with a quite simple digicam with a good macro-setting to check soldering.

I found that paste between pads caused the most trouble. It can be removed using a sewing needle.

Yes, it will.

If you do not plan to show it to others, dead bug "design" might be an alternative.

All hand soldered and working:

Falk

For non-production type work (e.g. repair and prototype), I tin each PCB pad so there's a small bump of solder. I make sure the bumps are approximately the same height. If you have a thermal pad in the center, only apply enough solder to raise the part slightly. Too much solder on the center thermal pad will cause problems. I use leaded solder since it's easier to handle and has a lower melting point than unleaded solder.

If you're reworking a board that used unleaded solder, I flood the pads with leaded solder which dilutes the unleaded solder enough where it isn't a problem. If you have leisured time, you can remove the unleaded solder with solder wick or a vacuum desoldering station.

Next step, coat the bottom of the IC with flux.

Place the part.

Preheat the board to 150 deg C. I use the Madell QK853 hot air plate to preheat boards.

Once the flux has thickened enough (I use RMA flux) use hot air soldering station to solder the part. You will see the part self-align when the solder melts. If you use water-based fluxes, be sure to wash the board as soon as possible as water based fluxes are highly corrosive. Water based fluxes are also conductive.

Hot air systems can be had for under $200 for cheap Chinese equipment (See Madell QK857D). They are fine for occasional repair work or prototyping.

Other ways of soldering are toaster ovens and, my favorite, a waffle iron with flippable plates (waffle or flat).

I have used this technique down to 8-pin parts in 1.6 x 1.6 mm packages.

Thanks guys. I'll have to get some paste and see how it turns out. The main problem is that a simple toaster oven or waffle iron is too small for my boards. Maybe with some type of flat pan iron it will be possible or using a hot air gun(after all, it takes them off pretty easily)?

Anyone know where I can get some solder paste on the cheap and easy and possibly non RoHS(i.e., leaded)?

Well, i have never tried them, but IF the particular units in question have the lands extended to the side of the package, then i would first tin the PC pads, drop the part on top, align as best possible, and use iron on a "corner" pad to "drop" (or align) in place. The capillary action forces by the solder should align the device reasonable sized parts. One can see if the part rotates some / too much; use a toothpick on "far" end to prevent that; let it cool. Inspect, if all OK then work a pad on the "far" end to "lock" the part in place. The rest of the pins will be a piece of cake.

Maybe..this will also work if no extension (no metal on package sides).

Are you in the states? How about a 100g syringe of Sn96.5Ag3.5?

I wouldn't design those packages in. They are hard to get right by the assemblers and you can't check the joints visually.

Try this website:

Select "articles" Select "RF PCB Design"

There is a PDF there titled: "PCB Design and Assembly for QFN Packages"

Good luck, Dave

I have a naive question. Do I still print solder paste on the PCB with a stencil for BGA's? (Given that the balls themselves are solder.)