I have been having reliability problems with SOD-323 diodes on my boards.I used the following document for my original footprint for the SOD-323 package: http://www.diodes.com/_files/datasheets/ds30097.pdf
This has a 0.59x0.45mm pad - too small in practice
Another document from the same vendor http://www.diodes.com/_files/zetex_files/pack/SOD323.pdf has 0.63x0.83mm, twice the size
NXP have http://www.nxp.com/documents/outline_drawing/SOD323.pdf recommending 0.6x0.6mm
Any thoughts on what this should be
There is a good discussion about this in the element 14 froum:
https://www.element14.com/community/thread/4449/l/sod-323-sod323-land-pattern-which-one?displayFullThread=true
Apparently most manufacturer show the outdated IPC-782 land patterns. Looking at the more current IPC-7351A land patterns, something like the one suggested by Diodes Inc. is now preferred. I can see why, as there is more pad for the heel of the part’s bent lead to grab on to, and more at the toe to ease re-work. I settled on 1.35mm x .7mm, pads spaced 2.4mm from center.
1 Like
Yes, I’ve seen that thread. The SOD-323 seems to have become a problem footprint on other CAD systems as well as KiCad.
What would a IPC-7351A nominal pad measure? Your 1.35x0.7mm is really large
I use 0.71 x 0.40 for the pads, centers spaced 1.99 apart, can’t remember from what device I got those from.
Special is that I included bigger copper pads underneath for better heat sinking, one of the Chinese fabs naturally screwed up the solder mask and the device then usually slips during reflow into unwanted orientations (but that only happened once, other boards came out fine as intended and work well).
The maximum lead width is 0.4mm, so a 0.4mm aperture looks a bit narrow, add some tolerance on the aperture and you have the resist lifting the lead off the copper, especially if the resist lifts with solder underneath it as sometimes happens
Found the reference I used - according to Diodes it’s 0.35:
1 Like
Fairchild recommend slightly larger for their SOD-323f 1N4148WS as they go up to 0.4mm lead.
https://www.fairchildsemi.com/datasheets/1N/1N4148WS.pdf
1.05x0.55, 2.8mm centres
It looks like different brands are not fully interchangeable physically
You guys know that you are mixing SOD-323 and SOD-323F footprints?
That is part of the problem, the makers seem to be using the same part number 1N4148WS randomly between flat lead and gull wing parts. Ideally I would like to be able to deal with either.
It’s even worse (see here: http://nl.mouser.com/Search/Refine.aspx?Keyword=1N4148WS), the through hole part also has the 1N4148WS manufacturing number 
I guess you end up making three footprints, and even a few more if you make footprints for both reflow and wave soldering 
If you follow the link and check the datasheet, it’s actually a SOD-323F.
Mouser etc frequently have the wrong thumbnail image. 
Worse, you get this sort of thing: http://diotec.com/tl_files/diotec/files/pdf/datasheets/1n4148ws
and http://www.eicsemi.com/datasheet/1N4148WS.PDF
Both clearly identified as SOD-323 but drawn as a flat lead
I just checked some old JEITA papers and the SOD323 seems to be similar to an SC-76, where as the SOD323F seems to match an SC-90.
I’ll probably make the pads of my F a bit wider (0.5 instead of 0.4) just to lean on the safe side.
1 Like
