Test Your Know How: PCB Separation

In reply to this we would like to publish our results:

   1. Speed => A. Scoring

   2. Cost => C. Lazer Cutting

   3. Stress => A. Scoring 

Scored PCBs are separated using dual "pizza slicer" blades. This action happens very quickly in one swift linear motion. These cuts are generally not selective and cannot reach some areas of the pcb. Complex pcb shapes must have some pre-routing performed on the pcb during fabrication. The final cut following component assembly, happens on the contact areas that remain between adjacent pcbs. These pcb borders must be partially pre-cut into thin v-groove using similar "pizza blades". This makes the final pcb separation very fast and the cut line is a little bit rough on the pcb edges

Lazer cutting is considered an expensive process and is not considered to be fast. Cutting by lazer typically requires many multiple passes over the same cutting line. Each pass removes a thin layer of material in a process known as ablation. There is some risk of conductive carbonised residue on the pcb edges from this. As a result of machine cost, this is generally considered to be the more expensive solution in the current market. It is more typically found in assembly processes that involve thin pcbs such as flex designs. 

Scoring wins ( or loses) again for component stress. In many industries component stress during separation is considered critical. The highest risk components for stress are typically ceramic capacitors which can crack during pcb separation. However, long resistor bodies, such as 1206/1210 or similar, can also crack under such mechanical stress. If mechanically sensitive components are located close to the cutting border ( <5mm or so) then it is advised to pre-route the board in these locations to reduce the stress applied by the scoring blades. 

Note that separation by milling is also considered stressful. However, typically milling tabs are placed in specific locations evenly around the pcb which reduces the total stress impact on the board. 

If you have any other experiences, comments or observations ( or even disagreements) on the points listed above,  then we would be very happy to read and reply to any comments in the section below. Looking forward to hearing any reactions!  

Hi,

Typically we never use panelized boards. The reason is an extra step is required in post production to de-panel the board. That process can introduce flexing of the board which can cause component fracture in a worst case. In the past years ago as with most designers we had used mouse bites and scoring but modern SMT Placement equipment can literally place components at the edge of the board if needed, not that you would want to do that.

If the board is super small say 1 inch square or slightly larger, Typically a panel will be required. In this case mouse bites work well. On very small boards we usually opt to use a thinner PCB Laminate as opposed to using standard 62 mil FR4. The advantage here is thinner material is easier to route out and de-panel if needed.

I have found that if using an assembly house it is a very good idea to get an idea of the SMD Equipment they have. A good question to ask is do they have a preferred minimum on how close they can place smd components to the edge of the board. In most cases armed with this information it is possible to design a board where the actual finished board is the panel.

Best regards.

Hi,

Typically we never use panelized boards. The reason is an extra step is required in post production to de-panel the board. That process can introduce flexing of the board which can cause component fracture in a worst case. In the past years ago as with most designers we had used mouse bites and scoring but modern SMT Placement equipment can literally place components at the edge of the board if needed, not that you would want to do that.

If the board is super small say 1 inch square or slightly larger, Typically a panel will be required. In this case mouse bites work well. On very small boards we usually opt to use a thinner PCB Laminate as opposed to using standard 62 mil FR4. The advantage here is thinner material is easier to route out and de-panel if needed.

I have found that if using an assembly house it is a very good idea to get an idea of the SMD Equipment they have. A good question to ask is do they have a preferred minimum on how close they can place smd components to the edge of the board. In most cases armed with this information it is possible to design a board where the actual finished board is the panel.

Best regards.

Hi Excellon,

You have piqued my interest. I have been involved in many hundreds of board designs and it is very rare to be able to design one without an array (as I prefer to call it). I have visited numerous fabricators and, again, it is very rare to see boards that don't need to be singulated on their production lines.

The cost of the extra step to singulate a board, and the cost of the extra laminate, adds up. Not having to singulate would be a big cost saving.

I spoke to my production engineers about your experience and they are sceptical. Can you please elaborate a little more on how you achieve this nirvana? It would be  a great conference presentation.

Cheers,
Dave

Hi Dave.

The boards we do are plain vanilla for the most part with rounded edges on the corners, needless to say they can be of various sizes. Our preference is to ship the boards to the Assy House. On the boards they do incorporate space between the parallel
edges of the board and components. The minimum space we use is typically 150mil from the board edge, sometimes this can be 200mil if the board is double sided. Basic idea is to have the breakaway rail incorporated into the board so the SMD placer can handle it. Or put another way have margin on the board design so no breakaway rails are needed.

If we are unable to design a board to meet the minimums, then a fixture would be needed to hold the boards This fixture then gets slid into the SMD placer. Typically there is a cost adder if a fixture is needed. Some machines today work on the principle of work holding in that the variable is the board size "Width" Provided the board can be held in place and presented to the SMD placer then the width of the board is a non issue. If I am not mistaken Panasonic equipment has such a feature. The work holder is the conveyor.

If we have to use breakaway rails and we have in the past, typically these are 400 mil in width. They also contain tooling holes and fiducials.

You make a very good point on the savings with respect to extra laminate. If this can be minimized then there will be a cost saving.

I think there is not a one size fits all here, Its going to be down to the assembly house and the equipment they have. Certainly it is good to ask questions in advance of a build to see if any savings are possible or possibly a new approach in how the boards are designed could get you there too. One can never ask enough questions on this stuff as the equipment today is light years ahead of where it was 10 Years ago.

One last thing on this. It is possible to design a board that does not have any real margin around its edges. On the board 2 or 4 holes are included for mounting purposes. The Board goes into the SMD placer and stops. Then swage pins come up to hold the board. The placer then moves the conveyor rail away from the board and the Placement begins. These type of placers can handle some very esoteric shaped boards such as round, or odd shaped etc. Memory is a bit foggy on this, it might be a universal machine. 

Best regards.

Hi Dave,

Just a quick follow up. I found a video on how small computer motherboards are manufactured that might be of interest. Jump on youtube and do a search for  Inside China's Mini PC Production: How Tiny Computers Are Made

Go to about 8.10 in the video. At that point it evolves where by the board is mounted in a carrier and then put into the SMD Placer.
The video clearly shows that the board is de-paneled. The carrier method means no breakaway rails etc are needed on the board.
Perhaps that might be something to help reduce cost and increase reliability since the stress on the board is out of the equation.

Best Regards.

Thanks,

That's a fascinating video for a number of reasons.

We use pallets for very thin boards and flexes but still use arrays for those.

Food for thought.

Cheers,
Dave

Hi Excellon1 & Dave, great discussion!

I am back from holidays and just wanted to add in some thoughts on this. Agree that we need to keep asking these questions, discuss and learn from everyone - especially with both our assemblers and fabricators!  Also, vanilla is a very underrated flavour; but I guess we are debating here if we need to pay for extra sprinkles or not! 

I really enjoyed the video thanks Excellon1! The wave soldering conveyor belt was very interesting to see; I noted it had custom pre-cut openings for that product to match the connector positions. That was a nice touch I hadn't seen before!

On the negative side, I see an exceptional amount of manual intervention required for the assembly flow. I am more used to an automated processes that minimises physical hands. One major strategy to achieve this is to minimise or eliminate the amount of wave soldering on the board. Ideally smt side 1 and side 2 only is best, fastest and therefore cheapest. Eliminating those pallets is actually a cost improvement; which take extra time, manual intervention and cost to load then unload; not to mention the bespoke design and material cost of those things...

 I have more thoughts if you are interested to discuss further! Would love to hear anyone else's experiences also!