Confessions of a PCB Designer — on Assembly Sub-Panels
The electronics business is a small world. To scale up, we have to gang the PCBs to facilitate handling. The assembly panel is an under-appreciated chance to save on the cost of goods sold.
Few PCBs are so large that only one is made from an 18 x 22-inch fabrication panel. Not many assembly lines are big enough to handle that much board anyway. More often, we see boards that are only a few inches per side. A lot is being asked of these little boards, so it’s not often that they are sparsely populated. Even when the form factor does not require the minimum board size, the trend is to shrink it anyway; less material, lower weight, more features are all immutable trends.
The assembly machinery — pick and place, reflow, AOI, test, etc., needs some flat space to hold the board along its edges. Those edges typically are taken by the components. Adding a break-off strip to two sides of the small board so that the machines can secure it for processing is essential.
The discard material may be the only place left for tooling holes, fiducial marks, impedance coupons, tracking tags, and so on. Taking efficiency a little further, we can create an array of boards that are joined by conventional break-off tabs.
Board size and shape will dictate the method of panelization. If the singulated PCB has an odd shape, then a routed outline with break-away tabs, (so-called “mouse-bites”) along the sides may be necessary. Edge connectors, mid-mounts, gold fingers, and anything else that meets or exceeds the board envelope needs to be accounted for in the sub-panel design. These component types almost always require a routed edge away from the break-off tabs. Many boards will be busy enough that a full frame around all sides is needed.
The spacing of the tabs is typically 2 to 3 inches. The weight of the components, along with board thickness and width, will drive the optimum spacing. Too many tabs will stress the board during singulation, and not enough tabs will produce a panel that is too flimsy for processing. Learning all of this the hard way can be quite expensive. Don’t be afraid — your vendors know their equipment and want you to succeed.
You can still wing it if you have the time and money and somehow managed to miss out on the “fear” gene.
IPC-7351 chapter 3.4.8.3 provides a bit of bravado on this subject. Seriously — work with your partners in fabrication and assembly.
Straight edges can leverage V-score techniques which minimize the amount of waste material. The edges left behind by V-score can be too jagged for some applications. Since they are straight edges, cleaning them up isn’t too hard. For a quick and dirty test on fixtures or other places where you’re not required to have a unique outline, four sides with 90-degree corners may be your best friend. Even if there are only two straight edges, a combination of V-score and routing may be the most clever solution.
Pro tip: It is much better if you can use .125" diameter non-plated holes for mounting hardware. Most fixtures use 0.12" or 0.10" (2.5 mm) dowels for alignment. Hardware with 2–56 UNC threads will comfortably fit into these hole sizes. Reusing the diameter of the tooling hole saves a drill bit change-over. These little things add up to big numbers in mass production.
The size of the individual board and the equipment used for manufacturing will allow for some engineering choices. Occasionally, boards can be nested, and the degree to which you can optimize the number of boards per fabrication panel will show up on your bottom line. An extra board or two isn’t worth it if the panelization plan causes damage to the completed assemblies during depanelization.
Placing the mouse-bites in a way that minimizes stress on the PCBA takes some insights. It’s worth the time upfront to go through the “what-if” scenarios with the fab and assembly houses. Better fabricators are willing to create a sub-panel and share the array of artwork with the stencil and assembly vendors. That said, it may be wise to keep the assembly panel geometry in-house so you don’t run into an awkward situation when you use a second source.
One scenario that has been successful is rotating and mirroring the boards in pairs. Your fabricator may not like this approach since they now have to generate multiple drill files with different versions of everything. There’s an upside though — we put a little extra work into the fabrication if it makes the assembly process boring and uneventful. This is one of those things.
To start with, there’s the two-up sub-assembly panel. The ordinary way of doing this is to step and repeat the first board for an A-A panel. The extraordinary plan is to make the second board a mirror image of the first one by flipping it over for an A-B panel. Then you have both a top-side and a bottom side facing you. This is no help with a one-sided board where you would stick to the A-A presentation.
The beauty of the A-B flip is that you can turn the assembly panel over and see the same pattern. One X-Y placement file will work for both sides. The time spent populating one side of the panel is the same as the other side of the panel, which reduces bottlenecks in assembly. Since both sides are the same, there’s no reconfiguration of the component reels on the pick and place robot as you would find if all the boards face-up. You have one stencil instead of two — this alone streamlines the process.
There are downsides to this process, of course. All of your components need to be able to withstand two trips through the IR oven. Heavy parts may need to be glued with under-fill, so they don’t fall off when they go through upside-down. That creates a rework headache. If your test plan is to probe from the bottom only, you’d only test half of the boards on a panel at a time.
However, if your parts are robust and light-weight, this approach may pay off on your bottom line. Three tooling holes in the panel break-offs are normal. This method requires four, and they need to be symmetric. The added tooling hole goes against the principle of “poke yoke” mistake proofing, so adding big arrows to both sides of the break-off tabs that indicate placement orientation and direction of travel is advised (don’t make it perfectly square!).
As the boards get even smaller, a two-up panel isn’t cost effective, so we take the two-up and repeat it or mirror it on the other axis to get four different orientations. The deciding factor between mirror and repeat is minimizing wasted material. From there, using step and repeat, we can create 8-up, 12-up, and so on in multiples of four.
Odd board shapes become more common as we pack more electronics into less space. If this is the case, get out some paper, a pen, and scissors, and mock-up different placements to see if they can nest and flip while minimizing fabrication panel usage. You can try your mock-ups in the CAD (Computer Aided Design) software if you’re handy with the drafting tools (you should be if you want to make a career out of PCB Design). In any case, the board shapes will reveal their most advantageous orientation as you do the study.
It’s difficult to put a dollar figure on this as we balance the additional NRE (Non-Recurring Engineering) expense for the extra CAM (Computer Aided Manufacturing) work against reduced assembly set-up and overhead. It’s not for prototypes, but pre-production runs that might be a good time to test drive the A-B or A-B-C-D flip ahead of mass production.
A little work up front buys you that nice boring assembly line.
The assembly always has the greater risks to schedule as well as the budgetary exposure because of all of the piece-parts in play. While there are some dependencies, it’s worth a look.
A deeper dive if you want to wade in:
http://www.electronicdesign.com/boards/pcb-designers-need-know-these-panelization-guidelines
http://www.pcdandf.com/pcdesign/index.php/editorial/menu-features/9351-dfm-1406
