I think Ryan is spot on. I had done some research into the calibration of
Pick and Place equipment at my last place of employment. I was in the
process of trying to come up with a cost effective attack plan to insure
that the pick and place equipment stayed in calibration.
Down time was a problem. Three lines, two Fuji's and one universal with
multiple change over meant that there was no window of opportunity for
calibrations or worse unscheduled down times. This is what I came up with.
The line would be calibrated professionally. Data collection would start as
soon as the line went operational. Data would be collected on anything and
everything nozzle wear, miss picks, table crash anything. From there it
would go into a data base where the data would be used to try and trend the
issues. The data would be used to set up a maintenance program that followed
a scheduled down time, which is less stressful as a unscheduled down time.
The line would then be recalibrated using glass plates. I left before I
could put this in place and I think the largest hurdle was the ME's, Line
supervisors, line operators or what ever you may call them have to resist
the need to fix it and go. They hate documentation because it slows them
down. The other worth sharing is that all the tweaks that have been taking
place on your line to keep it going will have to be adjusted out. If you can
collect clean data your down time will go down and your accuracy will go up.
It will just take some time to collect it and process it. This is not a
quick fix but a long term solution.

Ted

----- Original Message ----- 
From: "Ryan Grant" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Tuesday, June 26, 2007 10:40 AM
Subject: Re: [TN] SMT placement test PCB


Hi Ioan,

PCB's are poor indicators of Pick and Place machine accuracy since the PCB
is not dimensionally stable.  In other words, if a land pad on the left end
of the board is used as an origin, the land pads on the right side of the
board can vary as much as ±5mil.  For this reason, the dimensionally stable
glass plates must be used.

However, I know your pain!  The Fuji's will slowly drift out of calibration.
At a former employer, we would constantly battle this until we organized
ourselves.

1.)We made sure every program was generated with the fiducial coordinates
from the gerbers.  No guessing here!
2.) Pick and place offsets were only made as global offsets to the primary
fiducials.  If a component is off during the proto, then the component
center is corrected in the Gerbers and the program regenerated.  No guessing
here!
3.) All pick and place programs are stored on a networked server and changes
to the pick and place program coordinates are not allowed to be uploaded to
the server.
4.)Once programs are "known golden", pick and place global offsets on the
machine file are only allowed up to a reasonable amount before the machine
must be "quick recalibrated".  In other words, a full calibration might take
an entire shift, but bringing the machine back into reasonable calibration
only takes 30 minutes or less. (If I remember correctly, it's a software
offset at the machine).  Determining the amount of "shift" can be done using
live product that is currently on the line, so no special boards are needed.
But a full calibration, and the recording of placement performance must use
a "calibrated" substrate, or the drift in the pad locations will make the
pick and place machine placements appear to be off.  While a photo process
is accurate, I'm not convinced that the glass/epoxy laminate is stable to
the dimensions required for placement equipment calibrations.

By the way, since Fuji charged an arm and a leg for their glass calibration
plates, we had custom glass plates manufactured locally for a fraction of
the cost.

Thanks,
Ryan

---------------------------------------------------
Technet Mail List provided as a service by IPC using LISTSERV 15.0
To unsubscribe, send a message to [log in to unmask] with following text in
the BODY (NOT the subject field): SIGNOFF Technet
To temporarily halt or (re-start) delivery of Technet send e-mail to [log in to unmask]: SET Technet NOMAIL or (MAIL)
To receive ONE mailing per day of all the posts: send e-mail to [log in to unmask]: SET Technet Digest
Search the archives of previous posts at: http://listserv.ipc.org/archives
Please visit IPC web site http://www.ipc.org/contentpage.asp?Pageid=4.3.16 for additional information, or contact Keach Sasamori at [log in to unmask] or 847-615-7100 ext.2815
-----------------------------------------------------