SMT lands formulas -- how are gull wings manufactured?

[Mr.CRC]

Hi:

I have a document which I obsessively revise that contains my formulas
for SMT lands. There is one formula that I can never settle on. It is
for what I call the "land inner bound" (LIB) which is the distance from
the origin on one axis to where the land begins. The land outer bound
(LOB) I have no trouble with. The inner bound is tricky. I base the
LIB and LOB on the lead and/or body dims and tolerances to compute the
land size and location. I use my own formulas mainly because I hand
solder many prototypes, and prefer my own land designs to IPC and
manufacturers recommendations.

In order to come up with a logical argument for how to compute the LIB,
I need to understand more about the manufacturing process for these
packages.

I imagine that the "lead frame" is flat until just before or after the
injection molding process, at which point it is slammed against some die
which bends the leads into the wing shape, and simultaneously snips the
excess length.

The main issue is this: If there is a die, then I expect the tolerances
on the foot sizes to be rather independent of the tolerances of the
plastic body dimensions. In this case, any formula for the LIB should
be based solely on dims. and tolerances for the leads and feet.

Alternately there is a coupling of the dim. tolerances of the feet and
the body dims. I think this is unlikely.

Also, I expect there is a tight coupling of the lead length and the foot
length according to this manufacturing process. Ie., you would expect
that if the lead length is toward its max. tolerance, that the foot
length would be as well, and vice/versa. You would not expect that the
foot length was at its max tol. and the the lead length was minimum.

However, if there was a correlation between body tolerances and the lead
bending process, then the above paragraph might be incorrect.

Perhaps if someone can explain the chip packaging and lead forming
process, we will all be better informed and I might be able to make the
final revision of my LIB formula!


Thanks for comments.

 

[Mr.CRC]

--- I don't understand what you mean by "land inner bound".

can you post an sketch or a drawing somewhere, please?

Thanks for the input John. I was afraid of this, the "a picture is
worth 1000 words" problem. I'll try in ASCII, and if that doesn't help,
I'll try to make a drawing when I have some time to play with web
servers and all that...

Lets say the package center is at the left edge of the screen in a top view:

BBB represents the plastic body
LLL represents the lead
FFF represents the foot part of the lead
Dashes, pipes, and pluses represent the land.

+---origin
|
v

BBBBBBBB
BBBBBBBB +------+
BBBBBBBBLLLLLFFF |
BBBBBBBBLLLLLFFF |
BBBBBBBB +------+
BBBBBBBB
^ ^
| |
LIB--------+ |
|
LOB---------------+


Hopefully this makes things clear.

The only coupling to the pin dimensions is the overhang, which will
determine how closely you can space the chips end-to-end, and has
nothing to do with the pin-to-pin relationships.

The 1000 words problem again. I'm not sure what you mean.

--- Since your land pattern is going to be dependent on the pin
dimension and spacing tolerances, then the manufacturing process is
really irrelevant and you should be able to work from the
manufacturer's package drawings.

Here's a glimpse of my reasoning for the present formulae:

Note that one of the main difficulties with calculating the LIB is that
manufacturers aren't consistent about specifying the foot length. Some
give only min or max values, and no nominal. Some have a huge
difference between min and max, and an unspecified nominal.

Here's the LOB formula. This one is easy:

--------------------------------------------------------------------
2. Determine land outer bound relative to the origin from the greater of:

(max lead length from org.) + 0.016~0.020" [0.40~0.51mm]
OR
(max lead length from org.) + (max lead thickness)

The consideration for land outer bound is simply to account for both
the max lead length plus some room to apply a soldering tip for manual
soldering, or for a minor fillet to form for reflow. For large lead
thicknesses such as on power packages, then the extension beyond the max
lead length is dictated by adequate fillet forming, thus the formula
switches over to being based on the lead thickness itself.
--------------------------------------------------------------------

The LIB is the hard part:

---------------------------------------------------------------------
3. Determine land inner boundary (LIB) from the greater of:

(nominal lead length from org.)-(nom foot length)-
(nom lead thickness)
OR
(min body length from org.)-(nom lead thickness)

Justification:

There are two worst case possibilities for the LIB:
A. min lead L - max foot L
B. min body L (ie, lead is crushed against a minimally sized body)

#A is highly unlikely because it is unlikely that lead L and foot L are
uncorrelated. Ie., if you have a short lead L, you probably also
have a shorter than maximum foot L. The problem is that we don't know
by how much to assume that the foot L is less than max. if the lead L is
small.

Thus, we could wave hands and say it's probably a safe bet to modify the
formula to: min lead Lo - nom foot L, because this is still likely to
produce a LIB that isn't too far out for any likely tolerances. This is
justified as well because this LIB is certain to be closer to the origin
than the calculation resulting from purely nominal values.

#B is also highly unlikely because I doubt the lead forming
operation allows for the lead to be pressed directly against the
package, especially in the case where the body L is minimum. I think
the lead forming die sets an inner bound for the foot bending corner
distance from origin, and the body tolerance fluctuates independently of
this due to injection molding variances. If this reasoning is correct,
then the body dims really shouldn't factor into the formulas at all,
except for a sanity check in some strange case where a value was
computed less than the min body L. In fact, this would argue that a
value less than the max body L would even be suspect!

Unless of course if my mental model of the process is incorrect.