QFN/VQFN work

[Simon]

So i have designed a project that had to be quite small so I used VQFN packages although I am confused about whether or not VQFN are the same as QFN as it seems some manufacturers have their own names for the same thing.

So first assembly attemp has not gone well. I used a stencil, paste and a small IR oven. I had too much paste and many pins have shorted together. Has anyone any general advice for working with these parts?

[SMTech]

What kind of stencil, what thickness, what squeegee, what ratio between aperture & pad size?
Equally what device/package exactly... all helps

[Simon]

the stencil is 127µm, I just hand applied the paste with a credit card.

[SMTech]

Nasty unframed freebie stencil?

[Simon]

Nasty unframed yes, free no, I ordered them from PCB train with the boards. Stencils alone cost £41.50. I don't have any equipment to put a framed stencil into - should I have?

[SMTech]

It certainly helps to be able to pull them at least a little bit tight (a proper framed one is tight as a drum), the finer pitch you are pasting the truer that becomes, its not just the fact it can bend/flex/shift out of a frame but lifting it away from the board is more complicated too when it isn't rigid. Thats not to say its impossible, but printing by hand is already more likely to result in excess paste so every little helps. I haven't seen an off the shelf frame that helps with these stencils, however if your stencil is tiny it matters less, you just introduce a new problem- nowhere for the paste to start or end its journey.

If your VQFN is 0.5mm pitch a 0.1mm stencil may have helped.
Your paste should be Type 4 or higher
Your paste should be well mixed (check the datasheet)
For a squeegee get an old stencil and clamp it between two old pcbs leaving 20-30mm of "blade" unsupported this way you know the edge isn't deforming around the shape of your grip. (credit card would work just as well here probably)
Print with the squeegee held around 60degrees, the paste should form a roll as it you print. Don't "push" it through the apertures it will spread and force its way underneath

[T3sl4co1l]

some manufacturers have their own names for the same thing.

Yes, this.  "QFN" is a class; I don't think anyone specifies that it need to meet something.  The IPC calculator for example accepts any dimensions (body size, pins, pitch, pad size..) and simply applies the necessary expansion values to give a reasonable footprint (assuming reasonable inputs are provided, of course).

Names don't mean a thing; always check the drawing!  (Worst offender I've seen: an FTDI chip that says QFN but is actually a pullback (PQFN) so must be drawn for LGA style pads (no toe)!)

As for paste and reflow advice, the usual applies; try different adjustments until you nail down a process.  Tedious, takes a lot of mistakes, such is life.  Personally, I've only reworked QFNs with hot air, and it seems to go fine if I tin the pads then smoosh the chip into it.  Obviously, too much paste is too much paste, so, think about how it's applied (is excess being forced in?), or how high (mask too thick?), or wide (consider reducing pad-paste expansion and producing a new mask?).

Tim

[Simon]

It is a small stencil. I line it up by hand and tape it down with the board held between two other PCB's top and bottom. Yes I think the problem is the paste getting under the stencil. I usually take multiple swipes as the paste does not always fully fill the stencil hole. In the past this has been a problem as my 0805 pads were further apart and i needed lots of solder but after some issues I have brought them closer and I'm sure it will work with less solder anyway so I can go for one pass.

Type 4 paste? never heard of that. I'll have a look at what the distributors are offering.

[Simon]

some manufacturers have their own names for the same thing.

Names don't mean a thing; always check the drawing!  (Worst offender I've seen: an FTDI chip that says QFN but is actually a pullback (PQFN) so must be drawn for LGA style pads (no toe)!)

Would that be the QFN with outer terminals or not? i spoke to someone that we buy bespoke products from and he made this comment. Which is easier? I guess if the pads came out at the side the solder would have somewhere to go.

[Simon]

I've ordered some of this in a hurryjust to get something better in for tomorrow but apparently it's no good for printing? https://docs-emea.rs-online.com/webdocs/136a/0900766b8136abac.pdf

[georges80]

Just apply some flux and run a soldering iron tip along the 4 sides of the qfn and remove the excess/shorting solder. Unless you are building lots of these things (at which point I would just go with an assy house), touch up is easy enough. You do own liquid flux and a soldering iron, yes??

cheers,
george.

[T3sl4co1l]

Names don't mean a thing; always check the drawing!  (Worst offender I've seen: an FTDI chip that says QFN but is actually a pullback (PQFN) so must be drawn for LGA style pads (no toe)!)

Would that be the QFN with outer terminals or not? i spoke to someone that we buy bespoke products from and he made this comment. Which is easier? I guess if the pads came out at the side the solder would have somewhere to go.

Yes.  (The good old "inclusive yes".)

The actual part looks just like a QFN (with exposed wraparound metal), until you get a magnifier on it and see that there's plastic molding at the very corner.  So there's nothing for solder to wick up around.

The solder joints on that particular prototype looked very peculiar (but everything connected, so it was okay I guess).  Solder was bulged up in the toe area (because there wasn't a toe fillet at all).  It looked like it should be touching the side of the package (because the metal was exposed around the sides, like usual), but it just didn't, and poking it with the soldering iron couldn't make it reach, either.  The plastic corner was blocking it, but I couldn't see it in-situ.

Regular (no-pullback) QFN is easier because you can inspect the joints for a solder fillet.  Pullback is worse because you can only x-ray the joints.  If your assembler has x-ray capability, it's not much of a problem (what can I say, I'm spoiled?... ).

In your present circumstances, you'd probably put on such a part blindly, and just rework it if it fails testing, in whatever way you are able to pinpoint it as the culprit.

Tim

[Simon]

Just apply some flux and run a soldering iron tip along the 4 sides of the qfn and remove the excess/shorting solder. Unless you are building lots of these things (at which point I would just go with an assy house), touch up is easy enough. You do own liquid flux and a soldering iron, yes??

cheers,
george.

So you think the shorting is only down the sides and not underneath?

[jmelson]

the stencil is 127µm, I just hand applied the paste with a credit card.

Yikes, that's way too thick for fine-pitch stuff.  I use .003" stencils (about 76 um), but the fine-pitch parts have the apertures reduced down to 50 - 60% of the pad area.  I often use apertures of about 0.3 x 1 mm, while the pad might be .4 x 2mm.

If your stencil has apertures the same size as the pads, that **IS** your problem.

Jon

[jmelson]

I've ordered some of this in a hurryjust to get something better in for tomorrow but apparently it's no good for printing? https://docs-emea.rs-online.com/webdocs/136a/0900766b8136abac.pdf

I have found a bunch of solder pastes that do NOT work well.  The best stuff seems to be from Warton Metals (hard to get outside the UK) and Henkel GC10 (sold in the US by Loctite).  it doesn't make a small difference, it makes ALL THE DIFFERENCE!  I've now done a number of runs with the GC10 and have had an AMAZING reduction in solder defects.

Jon

[jmelson]

So you think the shorting is only down the sides and not underneath?

The shorts are almost certainly underneath, as well as some on the sides.  If you have too much solder, then it will GLADLY short under the part, and it can be HELL to get the bridge under the part to let go.  These QFNs need a VERY precise amount of solder on each pad to get reliable connections without shorts.  When the part is placed, the solder "brick" squishes out sideways.  To try it, apply paste, place the part firmly, and then lift the part off.  Examine the solder paste with a microscope.  If it is just a square ring of paste all around the terminal area, you are BOUND to have shorts.  What you WANT, is after placing the part, the paste is still almost entirely on the PADS, ONLY.  Really hard to do.  But, until you get this right, no sense to put boards in the oven.

Jon

[Simon]

the stencil is 127µm, I just hand applied the paste with a credit card.

Yikes, that's way too thick for fine-pitch stuff.  I use .003" stencils (about 76 um), but the fine-pitch parts have the apertures reduced down to 50 - 60% of the pad area.  I often use apertures of about 0.3 x 1 mm, while the pad might be .4 x 2mm.

If your stencil has apertures the same size as the pads, that **IS** your problem.

Jon

I thought the only options were 100, 127 and 150 microns.

[georges80]

Just apply some flux and run a soldering iron tip along the 4 sides of the qfn and remove the excess/shorting solder. Unless you are building lots of these things (at which point I would just go with an assy house), touch up is easy enough. You do own liquid flux and a soldering iron, yes??

cheers,
george.

So you think the shorting is only down the sides and not underneath?

Yeah, it'll more than likely be only on the sides. I've reflowed 100's if not 1000's of QFN parts and the surface tension will pull the solder to the sides and bunch (if too much solder paste). It *IS* very rare for the solder to short under the pad, regardless of what others may say

cheers,
george.

[KL27x]

^even if it's underneath, the iron will likely remove the bridges if you use flux.

Some manufacturers use a prefix when they make different sized QFN in the same chip. The VQFN are the microscopically small ones, in my experience, vs say UQFN.

My general advice: try different soldering pastes and different stencil apertures. Keep experimenting.

To fix them, I suggest you try fluxing the chip then wiping around the edge with as big a bevel tip as you can fit in there. If the chip is lifted too high and the bridges don't come out (or if connections get broken), flux it and press down on the center with tweezers while you hot air reflow. This will squish the excess solder out and form bridges around the side, but after taking an iron around the edges, the bridges should be gone.

You might also try manually soldering the ICs after stencil and reflow the rest of the board, but I would suggest you use ENIG finish. Many other finishes will oxidize during the reflow and will take a bit more time and effort to tin. Tin pads, place part, hold it down with tweezers while doing hot air reflow.

Another tip: the center pad might not need to be soldered. In this case, hand soldering with an iron can be faster/easier than doing reflow by hand. (Depends on your equipment and skills, I suppose). If it's not dissipating a large amount of heat, it's probably not necessary to connect the center pad, even if it's a ground pad.

[Simon]

How do I hand solder VQFN? The pads are all under the chip.

[Simon]

Right, i have been having a dig around and find that my solder paste expansion in CS is 0 by default so I have the full pad size of solder. My pads are 0.25mm wide, the component pads could be 0.18-0.30mm with the nominal at 0.23mm, so how big do I make my footprint pad? 0.30mm gives just 0.20mm between solderable areas with no solder mask.

I can of course reduce the solder paste apertures, this has to be a constant retraction from the edges, different x and y values are not allowed. Stencils can be done in 100, 127 and 150 µm, I think with correct paste mask offsets the 100µm should work. Do I make my pads 0.30mm wide?

[techman-001]

How do I hand solder VQFN? The pads are all under the chip.

I hand soldered this 32 pin QFN with a microscope and plenty of liquid flux, I didn't even preheat the board, and it all worked fine first time tho I wouldn't use it for anything but prototypes. I added the tiniest amount of solder at the pads at the edges of the chip. With preheating and a hot air gun, probably flux alone would have been fine as the header pcb was pre-tinned from the factory.

http://hightechdoc.net/misc/qfn32-hand-soldered-3.jpg

[Kean]

How do I hand solder VQFN? The pads are all under the chip.

The key is to ensure your pads extend out beyond the DFN/QFN part enough that you can get the soldering iron tip in there and do some drag soldering.  The extended pads also give the excess paste an exit path rather than trying to form bridges under the chip.

You'll want to use lots of flux, or at least in comparison to the tiny pads.  This is one of the few times I use my sharp pointed tip on my Weller pencil.

[mrpackethead]

Its quite possible that the stencil maker, reduced the apatures for you. This is common practice if you supply the gerber with Zero reductions applied.     

I calculate all my apature sizes, and supply my gerbers as how i want them made exactly.  In this case you need to tell them this, otherwise you might get reductions on reductions..   

This is quite a good guide.

http://www.surfacemountprocess.com/a-guide-to-effective-stencil-design.html


I'd also say that unless you have a way to tension your stencil, then you are going to have lots of trouble, esp once you get into these smaller sizes.     I only ever use framed stencils.  You also need to make sure that they are electropolished.   

Everything in the SMT process falls over if your paste process is poor.

[KL27x]

Some or most qfns bave exposed sidepad. A bevel tip will get into the tiny step and jdrag solder with ease, holding enough solder for multiple ICs. a knife tip aint bad, either.

You often dont need to comnect the center pad, except for power management IC or the like.

[SiliconWizard]

If the packages have an exposed pad, and if you haven't already, check whether some pins are shorted to ground (which is probably what the exposed pad is connected to). If so, you'll have a very hard time getting rid of the shorts with just some manual touch-up.

The solder paste mask matters. Is your exposed pad area in the stencil a solid square, or is it several smaller squares (as shown here: https://www.allaboutcircuits.com/technical-articles/stencil-design-for-exposed-pad-packages/ ) The latter is a much  better approach to avoid shorts and flatness issues.