>> I was wondering what policies others have regarding accepting x-outs on
>> multiple board arrays. We currently allow 50%, with the intent being we
>> would use a two up array with one x-out, but don't want to see 10 x-outs
>> on a 20 up array. We are contemplating going to something different,
>> say 5% or 1, whichever is greater. Or possibly no x-outs, but I'm
>> concerned about the cost ramifications of this (cost is a function of
>> yield, right?). I would appreciate any input from both fabricators and
>> assemblers.
>> Regards,
>> Pat Bailey
>> Zytec Corporation
This is a good one! Your 1 or 5% sounds pretty darned reasonable to me, and yes
there's going to be cost ramifications. Just how much, is going to depend on
what your fab vendor is able to yield, and that all depends on how much the
board you're asking him to build for you pushes the limits of his capabilities.
I had a talk with a gentleman from Bureau of Engraving named Wayne Hanson about
this one time. He told me that what he see's happen is that companies will come
in with a panel that uses every square inch to pack as many PCB's into the
available real estate as can be done, and try and get as many invidual boards
into a panel to drive the board costs down. But with larger arrays, it's tougher
to make all the boards good. So there's a cost trade off. But it goes beyond
just the price of the PCB. There's other costs that's going to be incurred
beside the cost of the bare PCB. Consider:
1. Solder paste - Yeah, you'll be printing a few grams of paste on a board that
is going to be cut out and thrown away, but that's not much money. But over a
years time if you build LOTS and LOTS of product, it could add up. At one of my
past employers we were in the memory business and built anywhere from 500,000 to
1,000,000 modules a month...I'd like to have the money from the wasted solder
paste!
2. Through-put and machine time - There's lot's of machines out there that have
the capability to use vision to look for X-outs, but that's going to add to the
machine cycle time, and there needs to be a mark on the X-out at the same place
on every X-out if it's going to work...9 times out of 10 what I've seen is a big
black "X" drawn on the board and it's not very repeatable as far as position.
When we would use the bad-board recognition function on the machines, an
operator would have to go through every board and put a masking tape dot at the
same place on every PCB that was bad so the camera could see it...maybe if the
fab vendors would standardize where they put a marking it could help save some
time.
If we really wanted good through-put we wouldn't use the pick-and-place
machine's vision recognition function, but separate the panels into groups that
all have the same bad PCB's, and then edit the pick-and-place program to not
load that PCB. That saves the time the machine had to spend looking for bad
boards before it started spitting parts out. That has it's downfall too, someone
has to spend their time separating the panels into their proper groups, and us
humans being what we are, will always make a mistake now and again. We would
build the bad boards once in a while because a mistake was made and a panel was
in the wrong group.
I guess the biggest message I got from Wayne Hanson when we talked about this
is, work with the fab vendor to have him help you layout the panel so it'll be
easier for him to get some good yields and you won't have to worry about X-outs.
But don't make the mistake of being so concerned with bare board costs that you
wind-up with a panel that is tough for your fab vendor to build without X-outs.
There's a lots more dollars that'll be spent on the manufacturing floor building
the product than you thought you were saving on bare board costs....
__\/__
. / ^ _ \ .
|\| (o)(o) |/|
#------.OOOo----oo----oOOO.-----#
# Steve Gregory #
# [log in to unmask] #
# #
#________________Oooo.__________#
.oooO ( )
( ) ) /
\ ( (_/
\_)
***************************************************************************
* TechNet mail list is provided as a service by IPC using SmartList v3.05 *
***************************************************************************
* To unsubscribe from this list at any time, send a message to: *
* [log in to unmask] with <subject: unsubscribe> and no text. *
***************************************************************************
* If you are having a problem with the IPC TechNet forum please contact *
* Dmitriy Sklyar at 847-509-9700 ext. 311 or email at [log in to unmask] *
***************************************************************************
|
