Electronics > Manufacturing & Assembly

Board layout of QFN for more reliable soldering

(1/1)

kcuzner:
I recently finished up a PCB design for one of my hobby projects. I was feeling adventurous and decided to try out some new IC packages, since I have had success hand-soldering some 10-DFN and 32-QFN packages. One of these was a 28-UFQFPN STM32 microcontroller. This package is a little strange for a QFN in that it doesn't have a large central pad. The PCB was only two layer, so I had ground pours on the same layer as my traces. Here is what I ended up with for a layout:


Note that the plane is covered in solder mask.

Now, when I attempted to assemble this, I ran into some problems when reflowing it with a hot air gun. My procedure was to wet all the pads with solder (less is more in this case), throw down a bunch of liquid flux, and reflow it. This works great with the DFNs. However, I needed to reflow it three times before it finally connected all the pads (and even then there are some uncertain ones and I'll only really know once I program those parts of the firmware):


The center pin (indicated with an arrow) had a difficult time soldering, though it finally connected. The far left pin has a slight misalignment, which normally doesn't happen during these reflow operations with tiny lightweight chips from what I've seen. They tend to snap into place. Anyway, three tries is a little much and if I have to do more, I'm not sure at what point the IC will stop working or the board will delamaninate.

I think that the plane invading the center of my IC is to blame here for the difficulties soldering this. It probably provided a nice sticky surface for the bottom of the chip to bind against and I also suspect that it prevented the chip from sinking down somewhat. I'm not sure what the thickness of the solder mask is, but I've learned that generally the few mils that it takes up aren't negligible when it comes to these packages.

What do you all think? How could I improve my layout so it is easier to solder these things on the first try?

T3sl4co1l:
There might be a thing in the soldering / footprint section about removing copper or mask beneath the device.  QFNs are definitely sensitive about flatness.

This is also sometimes seen on gyros/accelerometers and other MEMS stuff like that, where additionally, the strain from a bump under the device can cause performance problems and errors.

Tim

KL27x:
Yup. Silkscreen is too thick.

You can remove the unused ground pad underneath the chip. Or you can use a regular footprint with the pad there, even though you're not going to use it. (But I'd lose the connection you put between the ground pin and that center pad, if you don't have a center pad.)

In meantime, if there is no central pad, you can very simply solder these with an iron. Knife tip would make short work of it.

kcuzner:

--- Quote from: KL27x on June 18, 2017, 01:40:51 am ---In meantime, if there is no central pad, you can very simply solder these with an iron. Knife tip would make short work of it.

--- End quote ---

Ironically, this is actually a controller board for a Weller RT-based soldering iron I'm building because I'm fed up with the limited options I have with my WLC100 (especially the lack of temperature control).

Navigation

[0] Message Index