6
\$\begingroup\$

I made a breakout board for an LGA-16 part (LIS3DH) and two attempts at soldering failed with a solder bridge between two pads. Top view of the PCB:

enter image description here

The land pattern was created according to ST's recommendations in TN0018:

  • PCB land length = LGA solder pin length + 0.1 mm
  • PCB land width = LGA solder pin width + 0.1 mm

A first attempt with solder paste applied manually (with a toothpick) failed with a pair of shorted pads (the leftmost two):

enter image description here

I then tried a second time, with a kapton tape stencil:

enter image description here

But the result was pretty much the same:

enter image description here

The part was also lifted from the board by this bridge, resulting in adjacent joints having no contact at all.

I'm now looking for advice regarding the land pattern design and solder paste application. Maxim has a land pattern recommendation (document 90-0396) that extends the pads to the outside of the package (the package dimensions are identical as per drawing 21-0660). My guess is that those extended pads would allow excess solder paste to flow away instead of creating bridges.

I could also try to cut the kapton stencil to allow for less solder paste, or use narrower pads. I also considered taping the chip down to the pcb with a piece of kapton to avoid the lifting problem, but that wouldn't allow the part to align itself during reflow.

Soldering was done in a selfmade vapor phase oven.

\$\endgroup\$
  • 2
    \$\begingroup\$ The interesting portion (the actual pads) of your images is less than 1% of the total, and not much more than some squishy pixel blob where you can't see much. So I go for the standard answer when you have solder bridges: try again with less paste. \$\endgroup\$ – PlasmaHH Dec 16 '15 at 10:23
  • \$\begingroup\$ If I could I would have supplied better pictures, but I don't have a microscope. Sorry. \$\endgroup\$ – Christoph Dec 16 '15 at 10:25
  • \$\begingroup\$ The recommended pads probably assume very precise stenciling. I'd want a little more space between pads if you know that you're going to homebrew that step, at least on the soldermask. Having the pads extend a bit past the package would greatly increase your odds for correcting the problem with an iron, or an iron a bit of braid to eat the excess solder. I guess you could try that approach but with a hot air rework station if you have one. \$\endgroup\$ – Ecnerwal Dec 16 '15 at 12:52
  • 1
    \$\begingroup\$ Worth trying to blow the bridge with a hot air pencil. Or suck it with wick and resolder with pencil. (I would agree with both extending and narrowing the pads to help both hand soldering and rework) \$\endgroup\$ – Brian Drummond Dec 16 '15 at 13:08
4
\$\begingroup\$

This should be doable without a stencil.

  1. Using a flux pen, like TechSpray 2507-N, wipe the pads.
  2. Using your soldering iron, carefully mound solder on the pads, being sure there are no bridges, and not too much solder.
  3. Flux the IC and the mounded solder well.
  4. Reflow. Once the IC and board are to reflow temperature, poke the top with a tweezer or similar to make sure everything is floating properly, it should self center at this point.
  5. If there are bridges, it would be common for the solder to bridge to the outside, so you can remove these with desoldering braid, and reflow again if necessary.

I do similar with dspic33fj12MC202 qfn-32 parts for fixing.

enter image description here

\$\endgroup\$
  • 1
    \$\begingroup\$ Wouldn't it be hard to "balance" the chip on the solder? I'd use solder paste to pre-solder the pads. It might be hard to poke the chip during reflow when it's in the oven, but I can just bump the whole oven instead. \$\endgroup\$ – Christoph Dec 16 '15 at 13:36
  • \$\begingroup\$ Its not that exact of a science. You just sort of perch it on there, and once its hot, it may be one tooth off, but you just push it into place, and it will suck onto the pad. I do similar all the time with fixing these things, I just use the solder that is already on the pads. \$\endgroup\$ – Erik Friesen Dec 16 '15 at 13:40
  • \$\begingroup\$ It may be necessary to push on the part, when you aren't using paste, it tends to help seat everything right. \$\endgroup\$ – Erik Friesen Dec 16 '15 at 13:42
  • 1
    \$\begingroup\$ Use a hot-air reflow tool rather than your oven so that you can poke the chip. \$\endgroup\$ – brhans Dec 16 '15 at 16:04
  • \$\begingroup\$ Well I don't have a hot air tool so I'll go for a hot plate. I can still poke it then. \$\endgroup\$ – Christoph Dec 17 '15 at 8:27
2
\$\begingroup\$

Erik's answer has all the important points, I just want to add what I actually did in the end to get an apparently correctly soldered part:

  • Solder paste was applied to the LGA pads with a tape stencil. I just removed four rectangular pieces from the tape with a scalpel, filled the gaps with paste and then wiped excess paste away with a toothpick. That resulted in about equal amounts of paste per pad: enter image description here

    I had to add some more to one pad but that didn't turn out to be a problem. Ugly cropped photo: enter image description here

  • The board was heated with a hot plate until the solder paste reflowed, resulting in tinned pads without bridges: enter image description here

  • Flux gel (Edsyn FL22) was added and the part placed. I also added solder paste for the passives and the passives themselves and put the board on the hot plate again. When the solder began to melt again, I poked the chip with tweezers, but much too hard: it travelled a distance of about two pads. This was fortunate because it helped spread the excess solder from that one pin I mentioned before. I pushed it back into place and removed the board from the plate.

  • I could measure no shorts, and a few MOhms between all I/O pins and Vdd or ground. I have not actually tested if the board works, but it should. enter image description here There is still a visible blob of solder at pin 6 (right edge of the chip, camera end), which I could just push away with a toothpick. The chip is also reasonably flat on the board.

Thank you to all who provided help.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.