kicad-developers team mailing list archive

[Lorenzo Marcantonio <l.marcantonio@xxxxxxxxxxxx>]

On Mon, 30 Aug 2010, Alex G wrote:

> Which exact standard are you referring to?

IPC-7351A - Generic Requirements fo Surface Mount Design and Land
Pattern Standard

> Do you think you could  write a small summarry or wiki regarding these
> 12 pages?

Mostly what I said before... but pages 12 to 16 are table with the
'magic numbers' to be used as a target toe/heel/side/courtyard size.

I found that there's a new IPC-7351B release, maybe that's why the new
calculator isn't free anymore.

The IMHO difficult thing is implementing correctly the tolerance
computation.

The whole algorithm summarized is: decide the pad size and position so
whatever the component size, placement and pcb size is (statistically
into the declared tolerances) there is at least so-much copper between
the pin and the board to have the correct solder fillets. Where
'correct' is dictated by the above mentioned tables.

So you compute the 'maximum material condition' of the package (more or
less using a RMS mean of the tolerances) to have the envelope of the
package (i.e. all the space it *could* cover), then you enlarge these
'maxi pins' to have the required solder fillets.

For example (and I'm using quoting privileges :P), if you have
a gull-wing lead with pitch less than equal of 0,625 mm, using the
nominal size (i.e. Density Level B - Median) you should make lands so
that you have 0,35mm of toe, 0,35mm of heel, -0,2mm of side (yes, the
pad is smaller than the maximum material condition of the pin); the
whole module should have 0,25 mm in excess of courtyard (to support
reworking and inspection, also as a bounding box for the autoplacer).

These info are contained in 19 table (for different components) for each
of the 3 level of density.

After that you must check for copper clearance: for example a typical
TSSOP need a 6-7 mil clearance for the 'optimal' pads... if you need to
use an 8 mil process you need to trim the pads: you have less than the
statistically needed copper but yield should be good anyway. In fact, if
you use Density Level A (Maximum) you are *required* to do manufacturing
tests to check the yield is effective (this stuff is a nightmare for
Q.A. engineers :D).

After you have the required clearance you just round everything to the
nearest 0,05 mm (the easiest part :D).

> computer, and I can email it to anyone interested.
> (I'm curious just how many people will try to attack my computer now).

Done that, have fun

--
Lorenzo Marcantonio
Logos Srl