Steve Hughes - PCB Fabrication

Previous Next
7

Welcome everyone to the PCB Fabrication blog.

I consider myself fortunate to have left my decision late to become an electrical engineer, having dabbled in a mechanical engineering career in my early working years.

That’s because when I started working in electrical/electronic design in the mid 1980’s, it was the advent of CAD tools for PCB layout and I made the decision right there and then that I would never want to spend endless hours cutting up strips of red tape and sticking them on pieces of paper large enough to carpet my living room, in order to provide the appropriate data needed to get a PCB fabricated.
I worked alongside those crusty old guys who had spent their entire careers in manual PCB design and even in my mid to late 20’s I was still considered a bit “flash” and an upstart just because I was the one using the new fangled computer software programs to design the PCB in a lot less time then they did.

However, when I look back I’m not certain that I ever really managed to get my PCB fabricated, assembled and back to the lab for testing before they did. After all I was new to this industry and still learning what a Gerber file was, Silk what? Drill Table: I thought they had those in the machine shop right? And of course no one was forthcoming in helping me figure this out.

So, I decided early on that the more data I sent to the fabricators the better and would generate every conceivable output that the CAD system was capable of at that time. Along with a Gerber file of every electrical layer, I provided drill and dimension drawings, Silkscreen, Solder Mask, Solder Paste layers and some component placement information, the list went on and the cool thing was that I generated it all on magnetic media (1/2 Inch tape reels at that time) no need for those living room sized pieces of paper.

However this was not a fool proof process, as more than once I wrote “all” my data to tape, then drove to the PCB fabricator to proudly deliver my PCB only to find that I forgotten something. Sometimes it was the tape recorders fault of course and certain Gerber files were not written to the tape, or maybe it just me rushing and I forgot to see the message regarding “please insert a 2nd tape” or something like that, or I failed to get a work order or pre approval from the buyers or other necessary documentation was missing.

So has anything really changed since then?

What are the typical data sets you provide to the PCB fabricators today?

It’s a fool proof process today right?

How many times do get a call from the fabricator to inform you that that cannot fabricate the board because of missing data or because critical design failures? Or that they can get the job done if you allow them to modify some of the supplied data!!!

Over the past few years I’ve worked quite closely with some customers and their fabricators to understand some of the issues regarding the critical design failures, after all there is nothing I can really do to help if you simply forget to provide a contract review or obtain financial approval.

The critical design failures appear to be fairly consistent and could be categorized as follows:

  • Unable to add etch compensation (At the fabricator, due mainly to the ever decreasing feature sizes)

  • Unbalanced build of the PCB design (The stackup simply does not jive)

  • Annular ring failure (That smallest drill size will just not work with a board stackup this thick!)

  • Unbalanced copper (On each of the PCB metal layers)

  • Thermal annular ring failure (One 10 mil spoke will never be enough on a Ground pin on that negative plane layer)

  • Plane layer clearances (Your minimum design clearance will just not suffice)

  • Minimum track width failure (I know you had to neck down to squeeze those few traces in, but if they cannot be fabricated!)

Anyway, you may be familiar with some of the above problems or you may have others of your own? Why not share those and you resolve those issues with this audience?

My question is: has the advent of more sophisticated PCB CAD tools and fabrication/manufacturing formats made the process any more fool proof?


P.S. For those of you who already know me and noticed in my avatar that I was dressed in a suit and tie and wondered, “what’s up with that?”, well I recently became a US Citizen, (the photo was from the Naturalization Ceremony) so now I can do anything you can do, except become President of the United States of America. But watch out Bill Ritter, if it Arnold Schwarzenegger can become Governor of California, why can’t I become the Governor of Colorado. 



Aug 26, 2008 3:37 AM Click to view PeteBrant's profile PeteBrant

A more advanced PCB CAD tool may allow you to place more constraints, or more complex DFM checking, but neither will do you any good if you source information is incorrect.

There is no substitue for knowing exactly what your fabricator can and cannot do, what will and wont add extra cost.

The second part of the equation is the diligence and abilities of you fabricator in checking the data your send them, some will accept what you send them as being "gospel" and will assume that you really did mean to drill a 3mm npth hole through the middle of that collection of pads. Other will question it.

Of course, after a period of time with a given fabricator, you build a raport, and they will know the nuances of you particualr designs, but this sort of thing can only be learned over time.

Of course, people should use DFM checks, checklisting and so on, but it os no good checking whether there are any traces under 0.1mm when your fabricator can only handle 0.2mmm

Aug 29, 2008 2:44 PM Click to view steve_hughes's profile steve_hughes

Pete,

You bring up some good points here and prompted some further questions from me.

I've started a secondary discussion here:

http://communities.mentor.com/mgcx/thread/1253?tstart=0

Steve

Sep 12, 2008 12:06 PM Click to view Jack's profile Jack

You brought back memories...! I started in 1987 on PCAD for DOS. Why? because none of the other designers there wanted to learn CAD, so the company advertised a position for "Anyone who knows how to use a PC". funny, huh? That's how I got my first job.

Anyway, I couldn't help thinking that all of your bullet points are issues caught by the front-end engineering. I mean, if the designer had thought of them, they wouldn't be in the list of problems, right? Would you believe that only three years ago I visited a board fabricator that had absolutely NO CAM dept. They would send your Gerbers out to be plotted to film, and what you got was whatever you sent. They are out of business now (go figure) but I was absolutely speechless during the tour. How can you work with a partner like that?
So my main comment is how much better some of these CAM guys are getting, and how much we rely on them to evaluate our designs before streaming them onto the fab floor.
I just spent about a week (5 days over several weeks) with our CAM guy and wrote up some notes (maybe a potential article) called "One Circuit Board Designer's Introduction to Front-End Engineering" which I could share if you're interested.

One question you asked was: What are the typical data sets you provide?
I would say "the typical ones", either GERBER/DRILL/DWG/NETLIST or some comprehensive file like ODB++. But that doesn't really tell you much. and your next statement is hilarious "It's a foolproof process today right?" Ha. Maybe I'll take the bait and try to respond to that one, but I'll have to write back later (maybe this weekend)

Just wanted to say hello for now (gotta work, too!) seeya

Sep 15, 2008 12:43 PM Click to view steve_hughes's profile steve_hughes in response to: Jack

Pete, Jack & All,

So continuing this theme here’s my thoughts and questions for today.

It appears to me that most folks are probably either:

• Relying on the fabricator to “prepare” the board for fabrication

• Using 3rd party tools to perform DFF and other fabrication related checks themselves

Of course this does not mean that the PCB engineer has not spent considerable time ensuring that fabrication, test and assembly related requirements are already engineered into the design. But a PCB layout tool can only do so much right?

Some of my thoughts on this subject came about recently when I spent a few days with some customers discussing DFF related issues for PCB design.

What I found was that many of them build time into their schedule for prepping the board for fabrication; we never even touched on test and/or assembly 
Typically that time was around a full (work) week, or 5 full days after they had completed the layout.

Now I realize that each customer has a different set of circumstances and that many don’t have either specialized DFF/CAM engineers or tools in house and rely on the fabricator to ensure that the designs are optimized for yield without compromising the electrical performance.

However even in the case where those tools and expertise are available in house, there still appears to be a rather inordinate time required to process and analyze the designs. In addition, several round trips may be required before the potential problems are resolved.

Remember that those 5 days were devoted just to identifying and addressing those DFF related issues.

Let’s not ignore the cost here as well. Do the fabricators really offer discounts if the board data is something other than good old Gerber, e.g. if you provide ODB++ data?
Are you paying for their time to run multiple DFF analyses and provide feed back? Or do they incorporate those costs in their original bid?

Does anyone else out there have any comment on this subject?

Regards,

Steve

Sep 15, 2008 12:47 PM Click to view steve_hughes's profile steve_hughes in response to: Jack

Jack,

Sorry, I almost forgot to ask if you would be kind enough to share the document "One Circuit Board Designer's Introduction to Front-End Engineering" with the rest of the audience here.

I'm sure that other users will find it useful and enlightening? :)

Regards,

Steve

Sep 26, 2008 12:58 PM Click to view steve_hughes's profile steve_hughes in response to: steve_hughes

Jack,

Where did you go?

Did you manage to figure out a way to share that document?

Thanks,

Steve

Sep 29, 2008 8:34 AM Click to view Jack's profile Jack in response to: steve_hughes

Where did I go?

I was at the IPC MidWest Conference
in Schaumburg, IL. Wasn't everyone?
(still catching up...)

OK, here is a link to my personal journey
into the heart of the CAM process, with
guidance from my friend Mike Tucker at PDC

http://frontdoor.biz/PCBportal/CAD2CAM.pdf

Jack (aka "the new guy")