Hi Jim,
In my mailbox Iarchived an earlier posting from Glenn Woodhouse that contained some very usefull information about the subject of AOI. I hope that Glenn doesn't mind that I repost his input here:
Daan Terstegge
Unclassified mail
> For the evaluation and selection of an AOI system there are a few key
> points
> that I suggest you focus on.
>
> 1. Solder Joint Inspection - Most AOI suppliers claim solder joint
> inspection capability but this is really one of the differentiating
> points.
> You need to decide how badly you want good solder joint inspection
> capability. Truly capable systems are more complex mechanically as well
> as
> in their programming thus requiring a higher level of expertise, as well
> as
> investment in time, for program development and optimization. For low
> volume-high mix and prototyping environments, it often takes more time to
> develop robust solder joint inspection that it takes to build and ship the
> product! Solder joint inspection will increase the false call rate of any
> system. Finally, it is important to remember that video based AOI systems
> can only operate in a line of sight fashion. Packages like SOJ's, PLCC's,
> and BGA's will not support solder joint inspection. There are x-ray
> "combi"
> machines available as well but you have to weigh the cost and throughput
> implications of a combi. If you make the required commitment in capital
> expenditure and engineering support, solder joint inspection can yield
> very
> good results and help to take your process to the next level. If you
> don't
> provide the appropriate level of commitment in support resources, then the
> investment in solder joint inspection capability will be a waste and cause
> frustration and confusion on the line until you ultimately turn off that
> portion of the inspection. Look for systems that use convergent software
> technology where the s/w keeps track of false calls and the conditions
> surrounding them as a means to automatically and statistically refine
> inspection algorithms as a means to reduce false calls.
>
> 2. Production Environment - As eluded to above, it is my opinion that the
> state of the AOI industry today does not provide a single machine that
> "does
> it all". If you are running in a high mix/low volume/high change
> environment then machine speed/throughput is less important than ease in
> program development. In this environment if you can't get a program
> cranked
> out and reasonably optimized in 2 hours or less you will spend all of your
> time programming and very little time running the machine in production.
> Look for off-line programming capabilities as a means to help free the
> machine up a bit more, but you still always need machine time for initial
> image capture and final debug.
>
> 3. Vector Based versus Pattern Recognition -
>
> Vector based machines will find the edges of packages and leads,
> determine centroids, compare this to CAD data, and except/reject based on
> numerical tolerance settings. This allows for the building of numerically
> based part data files, like with pick-n-place machines. Vector based
> machines can provide you with very accurate x-y offset data to validate
> and
> tweak in your p-n-p equipment. This keeps the AOI machine from "learning"
> a
> bad placement or condition as the "golden" model. There is also at least
> an
> order of magnitude greater complexity, difficulty, and time investment in
> the programming and optimization of vector based machines, but vector
> based
> machines generally offer greater flexibility over the long run for odd
> packages, special measurement applications, and fine pitch lead
> measurement.
> The vector machines I am familiar with (MVT and VI) are generally faster
> than the pattern rec. class of machines although some companies like CR
> Technology are steadily developing faster and faster machines, thus
> closing
> the gap.
>
> Pattern Recognition, or correlation based vision processing compares
> a
> bit map from the camera field of view FOV for the current board to a saved
> "golden" image for the same FOV. The user defines the degree of bit map
> correlation required to achieve the pass threshold. Bit map correlation
> algorithms are used to break up the FOV into smaller chunks, allowing for
> the detection and identification of a small defect like a skewed 0402
> although comparing FOV images that may be larger (~1" x 1"). The
> advantages to pattern recognition is that it is generally easier to
> program,
> using wizard style teach functions combined with cad data, and easier to
> understand. Great for a proto or high mix shop, or an environment where
> you
> don't have the luxury of throwing a high level tech(s) or engineer(s) at
> the
> process. The disadvantage is that often acceptable manufacturing
> variation,
> such as a different color tantalum cap off of the approved vendor list,
> will
> result in a reject. It then requires you to teach this as an alternate
> acceptable image. Depending on who is allowed to teach alternates, you
> can
> get in the situation where pretty soon you have dozens of alternates for
> many FOV's which can start to affect the speed of inspection and
> inadvertently create opportunities for false passes.
>
> 4. SPC - It amazes me how many AOI systems out there provide inadequate
> supporting SPC packages. Without a system that provides good real time
> SPC
> feedback up the line, the AOI machine is only a failure screen that helps
> you increase your ICT yields but does not reduce rework (assuming you are
> repairing AOI detected defects in advance of ICT). Look for systems that
> allow for web based live SPC reporting so you don't have to invest in a
> separate data server and set of applications/licenses requiring dedicated
> data terminals. Web based output allows you to leverage your existing
> plant
> PC infrastructure in the factory and at the desks of those who need this
> data. All you need is an Ethernet card in the machine which writes data
> to
> a network target, and a standard browser to run the AOI manufacturer's
> Java
> scripts for displaying the live data from that target. If you want to
> integrate the AOI data real time and automatically into your plant's
> existing quality tracking system, evaluate the data output structures and
> communications interfaces that the machine supports.
>
> 5. Company Stability/Structure - There is something like 20 players in
> this
> market today. Considering the current electronics industry market
> condition, and the natural tendency for consolidation in this type of
> business, I think we'll see many of these 20 disappear over the next
> couple
> of years either through lack of funding, lack of market share, or
> acquisition by larger companies. Work to understand the AOI company's
> financial health (private, public, venture capital funded, cash??, etc.),
> short and long term business strategy (IPO, acquire or be acquired),
> historical and current growth rate compared to competitors, and if they
> posses any differentiating technology that has inherent value over their
> competitors (this point often makes a small company ripe for acquisition
> which can be good over the long haul). The quantum leap in PC processor
> speeds, cheap RAM, and quality high resolution color CCD cameras available
> today is the primary reason for this explosion in AOI products, i.e. we
> can
> finally afford them due to affordable components! This levels the playing
> field so to speak from the hardware standpoint, making the software side
> of
> the product the key differentiating feature. Look for companies that
> recognize this and possess evolved machine algorithms, program development
> tools, and SPC packages. They should have a large software development
> resource with many experienced software engineers and developers. They
> should be able to demonstrate active software improvement projects and
> have
> a "roadmap" of additional features and enhancements they are working on.
> If
> they cant, and are focused on hardware and speed improvements only,
> beware!
>
>
> There's more to think about but I feel these are the biggies.
>
> Good luck,
>
> Glenn Woodhouse
> MCMS
> Advanced Technology Development
> (208) 898-2753
>>> "Marsico, James" <[log in to unmask]> 01/14 4:08 pm >>>
I'm looking for a low end ($65k) optical inspection system for SMT component
placement and polarity inspection. Anyone have any recommendations?
Thanks,
Jim Marsico
Senior Engineer
Production Engineering
EDO Electronics Systems Group
[log in to unmask] <mailto:[log in to unmask]>
631-595-5879
---------------------------------------------------------------------------------
Technet Mail List provided as a free service by IPC using LISTSERV 1.8d
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 delivery of Technet send e-mail to [log in to unmask]: SET Technet NOMAIL
To recieve ONE mailing per day of all the posts: send e-mail to [log in to unmask]: SET Technet Digest
Search previous postings at: www.ipc.org > On-Line Resources & Databases > E-mail Archives
Please visit IPC web site (http://www.ipc.org/html/forum.htm) for additional
information, or contact Keach Sasamori at [log in to unmask] or 847-509-9700 ext.5315
---------------------------------------------------------------------------------
|