kicad-developers team mailing list archive

[Lorenzo Marcantonio <l.marcantonio@xxxxxxxxxxxx>]

On Sat, Mar 30, 2013 at 09:07:45PM -0500, Karl Schmidt wrote:
> One problem is that to do it the messy way -  one wants to draw
> lines of no solder - but we have to draw the inverse of an 'X' with
> round ended lines etc..

CAUTION: the following is for mass production... if you assembly you
board and don't worry about reworking you can do as you want:P

You are very ingenious :D simply it's not industry practice (for
mechanical reason, mostly). The big pad is here for a good reason
(choose one or both: thermal dissipation or keeping the part in place).

If you make less soldering surface you lose 'some' of these properties
and the assembly become unreliable.

Practical example: QFN is not made that way just to use less board space
than a QFP. The leads are removed to reduce lead capacitance and
inductance (and the kovar contact interface) and have faster signals.
Faster signals give more power dissipation so the thermal pad is needed
(also you don't have heel joints so you need more mechanical strength).

You need to reduce the paste to have the 'right' amount of solder on the
pad ('easily' checked xraying the board:D) but the copper solder
interface surface remain the same... if you cut the thermal pad to route
other stuff you lose interface surface, which is bad. So, it's better to
keep the big pad intact.

There are various way to correctly reduce solder paste but nothing
standardized, so you have to hope the datasheet gives a pattern or ask
your fabricator for advice on what works better on his machines...

One other thing: the suggested approach of reducing the paste at the
edges only works for small pads... bigger ones (I'd say about 1cm
square) need the 'grid' approach to avoid gas lifting: the channels give
space for the flux reaction fumes to escape, otherwise the part will...
jump around XD it's fun to see (but not when you're installing the
part:P)

-- 
Lorenzo Marcantonio
Logos Srl