stencil/paste technique?

[awallin]

I have a board with a 3-mm × 3-mm, 16-Pin QFN package - so I ordered the PCB with a stencil (multi-cb).
Their standard stencil is 100um (or 120um?) thick.

Seems I get quite a lot of paste on the board (see picture) with this setup. I taped the board to the table and then taped the stencil on top.
Any good techniques for getting less paste onto the board?
-thinner tape? thinner stencil?
-minimize buckling and bowing of stencil - how?
-paste-application technique? I ran a sharp knife-edge with paste tilted at ~45-deg across the stensil..
-paste type? I think this is chip-quick (don't remember part-nr...)

I hope I got the QFN soldered OK - after hot-air gun I applied a bit of flux and ran a soldering iron across the pins to remove solder-bridges on two of the four sides...

thanks for suggetions!

[Kasper]

That does look like you've got a bit too much paste.

You can use a knife to remove some paste but that is a little tedious.

You may be better off with a thinner stencil. Polulu sells 0.0762mm thick stencils for cheaper than anyone else I've found.

You could reduce the pad sizes on the paste layer of your pcb design.

I recently had a fab house recommend removing silkscreen from near my 0.35mm pitch QFN when I reported a solder bridge. They said the silkscreen is thick enough to lift the stencil enough to cause extra paste to be deposited.

This is what I do:
Use a bare pcb as my squeege. Digikey sells squeeges that work well also.

Swipe a few times, pressing harder each swipe.

Don't tape anything.

Use thinnest stencil polulu offers.

If paste is cold (stored in fridge) then allow paste to warm up before use.

[TassiloH]

Sounds like you have an unframed stencil. The thickness should be ok. I'm not sure how exactly you did this - where did you tape the stencil to: to the table, to the board (how, the stencil is typically bigger than the board?!), etc.? I would suggest:

• Tape board to table
• Tape scrap board pieces next to board on opposite sides
• Tape stencil to scrap pieces, on once side over the full width, on the other side just with a piece of tape so that the position is fixed. Roll up a bit of that tape so that it is easy to remove and the stencil can be flipped up without much smearing.

These steps should make the stencil as flat as possible. At least I had some success when I was still using loose Mylar stencils, but it is not as reproducible as with a framed stencil.
When applying paste, do it in a smooth motion so that you don't push extra paste unter the stencil and it remains flat on the board (this part is easier with a framed stencil or the stencil mounted in a frame).

[SMTech]

That looks like poor gasketting to me.
I'll contradict one previous suggestion, One firm confident pass with the squeegee (using well stirred paste) when you don't have a framed stencil. Any other approach will increase the risk of smudging it or push more paste through the apertures. More passes when you are not a robot, is quite difficult to do unless you have a lot of practise and have the pressure and angle down to T (IMHO 60 degrees would be more like it).

Hold the PCB in place with old PCB tooling strip/scrap of the same thickness or slightly thinner (L shapes are perfect, worth creating intentionally even)
Tape the stencil foil to the bits above on one edge so you can lift it away easily.
If necessary raise the PCB slightly so that it is either flush or slightly higher than the total height of the bits holding it in place+the tape holding them in place (sheet of paper or some masking tape underneath perhaps)
Squeegee blade should be flexible and for that device 100/120 should normally both be fine.
Global aperture reductions certainly help as do "homebase" or other aperture patterns but neither help a great deal/apply on a device like the QFN. Every pad in your photo has too much paste

[jmelson]

Nobody else mentioned making the apertures SMALLER than the pads.  The smaller the components and lead pitch, the greater a reduction should be applied to the pad area to give the aperture area in the stencil.  I do use .003" (roughly .075 mm) thick stencils, but ALSO reduce the apertures to about 75% for SOIC ICs, and even smaller for finer pitch chips.

Jon

[ANTALIFE]

jmelson is on point. What is your paste pullback set to? Also are you able to show us what the paste and copper layer look like in your ECAD program?

[SMTech]

Nobody else mentioned making the apertures SMALLER than the pads.  The smaller the components and lead pitch, the greater a reduction should be applied to the pad area to give the aperture area in the stencil.  I do use .003" (roughly .075 mm) thick stencils, but ALSO reduce the apertures to about 75% for SOIC ICs, and even smaller for finer pitch chips.

Jon

That is a savagely thin stencil and a very aggressive aperture reduction, however there's way more to fix for the OP before starting to optimise stencil apertures.
A decent paste applied properly with 1:1 aperture ratio should still look like a decent print with sharp edges, and that print should stay that way for hours if left in the rack. The OPs print is awful before reflow even takes place, aperture reduction isn't the solution if they haven't got their printing process at least good enough that it looks like a nice print even if there is too much paste height.
We typically only apply weird stencil aperture shapes or reductions if we know there's something odd about the footprints and how they relate to the actual devices. Most of the time 1:1 is exactly what works in a tensioned stencil, a reduction helps for unframed but I would still expect a better result than that when I am forced to use one.

[mikeselectricstuff]

Are you using paste designed for stencilling, as opposed to dispensing? The latter is more runny and will flow too much.

[mrpackethead]

Im picking a few thigns;

(1) the paste is not suitable for stenciling..  I use Henkel GC10 which is great for stenciling and terrible for dispensing with a syringe.   
(2) the stencil is not gasketing properly..

Your squegee needs to be more like 60 degrees. 

Is this a stainless or plastic stencil?

[jmelson]

A decent paste applied properly with 1:1 aperture ratio should still look like a decent print with sharp edges, and that print should stay that way for hours if left in the rack. The OPs print is awful before reflow even takes place, aperture reduction isn't the solution if they haven't got their printing process at least good enough that it looks like a nice print even if there is too much paste height.

I got some expensive lead-free paste from Indium Corp. that did just like that.  When the stuff was cold out of the fridge, it wasn't too bad, but 20 minutes later sitting on the board, it was massively smeared/spread out.  Of course, reflow performance was NOT so good!

I then got some Loctite GC10 paste, and it is MUCH better.  There is still some slight slumping of the "bricks", but they mostly hold shape until the parts are placed and into the oven.  Several other pastes also worked well, but that Indium stuff was JUST TERRIBLE.

Jon

[mbless]

-minimize buckling and bowing of stencil - how?

TassiloH touched on it, but I have to say this is the problem since I've done that myself. When the stencil isn't flat on the PCB, the paste fills in height gap and results in too much paste on the board.

For unframed stencils I bought the stencil supports from OSH Stencils. It's just 2 pieces of plastic that are the same thickness as the PCB so the stencil will lie flat.

[jmelson]

TassiloH touched on it, but I have to say this is the problem since I've done that myself. When the stencil isn't flat on the PCB, the paste fills in height gap and results in too much paste on the board.

I'm still doing unframed stencils here.  If you apply enough pressure on the squeegee, it presses the stencil against the board, and the paste does not flow out under the stencil.  The idea is a stiff metal squeegee is pressing much harder on the stencil than the pressure it applies to the paste.  This seems to work pretty well for me.
Sometime, I will have to try out framed stencils, though.

Jon

[mrpackethead]

The biggest difference with a framed stencil, is that you should get much better release. 

[girts]

1) Solder mask swell and paste mask shrink settings - both between 1...3 mil. Because there must be mask between pads. Just to break whiskers.
2) According to your design QFN most likely will float to upper right corner after soldering...

[mrpackethead]

there does'nt need to be mask between pins.  If you are having  bridges between pins you've likelky got too much paste.   It is MUCH better to have no mask, and a flat surface than to have mask and a surface that is not flat, and get poor gasketing.