reflow often solder QFN 0.4 pitch

[M@X77]

Hello, I've some problems to solder QFN 0.4 pitch, maybe the problem is my (bad) reflow oven (an updated T962A), when I try to solder a KSZ8775 the result that has short circuits all pins, I tried different profiles and solder paste (also bismuth with 138°C), at the eyes it doesn't seems there are phisically pins jointed, but I also tryed with desoldering braid and liquid solder paste to remove solder alloy excess, also tried SN+AG+CU and classic SN+PB, nothing in all pcb the result is the same...
Then I also thinked is solder temperature, is the reason that I tried bismuth alloy, but nothing, no changes...
Do you have an idea?

Thanks
Max

[mars01]

You should check the solder stencil that you used for that board. Perhaps you are adding too much solder paste on the pads?

[kayvee]

Stencil is the most important thing here IMO.  Results can vary dramatically if this is not ideal.  Be sure to follow the datasheets recommended values for paste mask reduction (Except for the center pad I have found).

Excess paste on the exposed center pad causes the device to float during reflow whilst not making contact for the other pads.  I generally aim for 30% to 40% coverage.

I solder 0.4mm QFN's no problem, using an unmounted laser cut stainless foil, by hand applying paste through the screen.

[M@X77]

My vendor for pcb' s and stencils is jlcpcb, are steel stencils, laser cutted and electropolished (raccomandad with pin distance <0.5mm. To apply solder paste I use a flex steel spatula, the solder paste thickness is the stencil thickness is 0,12mm.
KSZ has not central pad. Looking with a 4x magnifying glass it seems there aren't short circuit pins.
Could be a layout problem? I don't have any ideas how to check it.
I attached chip layout.

[vern]

I solder 0.4mm pitch, also 0.4mm BGA which is even more difficult with no problem.
I use leadfree paste with a fine particle size, SENJU S70G - Type4.
You don't need much solder paste, a stencil with 120um is to thick, you get shortcuts.
I only use 100um, that is perfect.
I also found some critical areas on your board, you might get shortcuts there if the solder resist is not perfectly covering these areas.

[optikon]

In addition to the suggestions already,  Have a look at your paste aperture for each pad. You posted the copper layer showing the pads, but the thing to examine is also the mask openings (allowing paste)

You should not have it cover the whole pad. Allow for a tiny line of mask in between the pads if possible.
Additionally, this is one reason design rules exist in the CAD tools to warn us if there is insufficient mask barrier between pads. Did you have a design rule check for this problem?

One other tip, if there is no way to get solder-mask in between, perhaps you could shrink the pad width just a tiny amount if the pad is already larger than the pin.

[M@X77]

Yes I used DRC rules based on jlcpcb suggests
Dimension of granules is 20-40um.
Here is image with layer stop (where there isn't solder mask)

Thanks to everyone!!!

[vern]

one more hint after looking at your pcb again:
I would never make a pin a part of the ground plane or any large copper area, I would always separate it and connect it with one or two traces.
The reason:
If you apply solder paste with a stencil, the pad is exactly the same as the others, it gets the same amount of solder paste as all the others.
But when the solder paste melts it spreads out until it hits the solder resist, which makes the actual pin area larger than the other pins and therefore makes a thinner solder coating, which can lead to failure.
The other thing: the ground plane conducts heat very well of course, when you try to solder the pin by hand it is really hard to heat everything up. Same with desoldering with a hot air gun, you have to heat the whole grund plane to get the part off.

[doublex]

For me it works the below solder paste for DFN situations like yours. I believe that the problem is the balls of your solder size inside the paste that you use. Check the below video for more details. At the below link you can see the part number from the solder paste that I use. I hope that this will help you.

https://www.youtube.com/watch?v=uSb1fgCMUug

https://www.digikey.com/product-detail/en/chip-quik-inc/SMD291AX10T5/SMD291AX10T5-ND/3972568

[token0]

I also thinked is solder temperature, is the reason that I tried bismuth alloy, but nothing, no changes...

I'd avoid soldering with bismuth, it's good only for desoldering and light tinning.
Could be expired solder paste or damaged packaging (should be airtight) or storage conditions violated, check it's expiration date. Flux base could become deteriorated.