Re: 2 Fried G210's in a single day :-(

Jon,

Good morning! This is OT probably, but you might be interested how it
works here.

Each panel (12 drives, 3 up 4 across) takes 1850 seconds to surface
mount. It takes 3 minutes to remove a finished panel, replace it with
a new one, recycle the pick-and-place to the fiducial marks and hit
the "go" key. It is a 12 hour day for 250 drives. It runs 4 days a
month for what we need.

--- In CAD_CAM_EDM_DRO@y..., Jon Elson <elson@p...> wrote:
> mariss92705 wrote:
>
> >
> > That means in 12 months over 1 million parts were assembled using
> > 3.25 million solder joints.
>
> Yes, it's amazing that anything works when you figure out how many
> solder joints are in there.
>

>
> Yikes! I was envying you with your pick-and-place machine and
soldering
> ovens, but now I'm not so envious. Do you use glue dots to hold the
> parts until reflow? Some of my boards have 800+ parts, so it would
be a
> nightmare!
>
Envy is not the word you would use if you owned Ursula, the evil and
possessed chip shooter from Hades. She is a 4 axis, servo driven,
tooth-belted, DOS operated diabolical tormentor forged in the bowels
hell by the Prince of Darkness himself. Other than that, she's OK.

The stated accuracy is +/-0.0004", which is creative writing worthy
of Shakepeare himself. When she is in a bad mood (most of the time),
actual "accuracy" can be as bad as +/-0.012".

We use Kester R596L, a premium water soluable flux solder paste. A
500gm tub is enough to do 1,000 drives. The panels are manually
stenciled with the paste using a squeegee and a .006" stainless steel
stencil. The paste is sticky enough to hold the parts in place
(wherever they may land; thanks, Ursula).

The mounted panels are accumulated over the course of the day. When
the day's run is finished, they are run thru the conveyor re-flow
oven. The oven takes 45 minutes to reach operating temperature and it
takes 40 minutes to run 20 panels thru it.

The saving grace for poorly placed parts is surface tension. As long
as a part lands somplace on the pads, surface tension pulls it back
into place when the solder melts. In fact, it is so strong that it
will stand a part up on its end (tombstoning) if the part lands on
only one pad.

Glue dots are used if the board will have SMT components on both
sides of the board. You don't want the parts on the bottom to fall
off as you solder the ones on top. I have avoided double-sided
mounting because 5 cycles instead of 2 would be needed; glue dispense
cycle, top mount cyle, oven cycle, reverse side mount cycle and
another oven cycle.
>
> Hmmm, well power electronics is different, of course, but I have
approximately
> 0% component failure in the stuff I build. A VERY occasional
component
> failure in the field, other than obvious abuse, but almost 100%
(I'd say,
> maybe 99.75% if I had to assign a number) of all failures are
soldering
> problems or connectors that are not making proper contact.
>
Surface mount parts see stresses that thru hole parts don't. They are
heated to 335 C during soldering and also have mechanical stresses if
the cool-down cycle is not correct.
>
> If you wanted to be really inventive, you could build a test stand
that recycles
> the power output from the drive back to the test stand. This might
not
> actually be that hard.
>
Possible, but I wouldn't want to do 50 times!
>
> Jon

Mariss