Don't Assume Anything When It Comes To DFM

Nov. 28, 2006
Design assumptions can have devastating effects, especially now as the electronics industry struggles with the transition to lead-free assemblies and OEMs seek the help of qualified EMS providers familiar with lead-free and RoHS compliance issues.

Design assumptions can have devastating effects, especially now as the electronics industry struggles with the transition to lead-free assemblies and original equipment manufacturers (OEMs) seek the help of qualified electronics manufacturing service (EMS) providers familiar with lead-free and Restrictions on Hazardous Substances (RoHS) compliance issues.

Innocent assumptions can result in unnecessary high costs and lost time-to-market. As design engineers, we're taught precision, exactness, and discipline, as well as careful evaluation of necessary tradeoffs to reach our design goals.

However, assumptions occasionally creep into our logic. When they do, they create a domino effect with an assortment of negative consequences, including lost engineering time, finger pointing, records backtracking, frustration, headaches for both the EMS provider and OEM, and large money losses.

Once that assumption is uncovered, the engineers usually say they were too busy or doing too many projects. When it comes to efficient design for manufacturing (DFM), though, those excuses aren't acceptable.

Something like this happened to us recently. A customer engineer assured us a ball-grid array (BGA) socket pin assignment on a specification sheet was the same as a previous assignment. We learned later that this customer statement was based on an assumption, rather than a fact.

The customer provided us, the EMS provider, with a wrong component footprint as part of a pc-board design layout specification. Instead of viewing the BGA socket from the top, the customer engineer captured the BGA view from the bottom. So all of the pin assignments were 180° from where they should have been.

Our engineers thought this arrangement was strange and questioned it. We were doggedly persistent in trying to get a correct customer answer. A series of back and forth phone calls and e-mails followed with the responsible engineer. Each time, the customer engineer was sure the footprint was correct.

Finally, our persistence paid off. The issue was kicked upstairs, and customer engineering management subsequently issued a mea culpa after a detailed screening of the original product specification sheet and reviewing all of the e-mail chains. As a result of this miscue, the OEM had to re-spin the pc board at an added $20,000 cost.

This expensive lesson is an example of how a small error that could have been avoided easily can cost thousands of dollars. Can you imagine the cost of a bigger mistake? If you're running a batch of 50 to 100 boards, for example, this type of assumption can cost up to $100,000.

It's difficult to correct human behavior, and Murphy's Law is bound to occur when you least expect it. But OEMs and other companies working with EMS providers can keep from falling victim to design assumptions. One way is to create a comprehensive checklist. EMS providers also should maintain stringent design checklists to verify OEM orders. We applied a 10-step verification list to the footprint assumption described above to uncover the customer discrepancy:

  1. Pay extra attention to the socket footprint data sheet for all dimensions and related information. If anything is missing, ask the customer to provide the missing info.
  2. Check if it's the top view or the bottom view.
  3. Check for pin pitch.
  4. Check for pin numbering.
  5. Check the socket pin locations, pin population, and numbering with the schematic pin location and numbering for the socket.
  6. Verify if the socket is through-hole or surface-mount (SMT).
  7. When the socket is SMT, make sure that the data sheet specifies the SMT pad via and tolerances. Make the pad size exactly the same size as specified in the data sheet.
  8. When the socket is through-hole, make sure that the data sheet specifies the through-hole drill via along with tolerances. Make the drill size exactly the same as specified in the data sheet and add the tolerances in the drill chart.
  9. Make sure to correctly establish the center line of the whole socket. It may or may not be the center of the socket pins.
  10. The socket designer should have the other designer check the socket for any mistakes.

The checklist included another eight key points to follow at the component placement stage:

  1. Make sure the positions of the Nexlev connectors, VHDM connectors, mounting holes, and other items that are part of the board template are glued/fixed and aren't moved at any time during placement or any other stage of the board unless the customer specifically asks to move them.
  2. Verify pin A1 orientation of the socket on the board with the customer.
  3. Components should be placed to optimize the routing and use the shortest routes possible.
  4. Follow customer instructions specific to the placement of critical components.
  5. Some components need to be spaced at customer-specified distances for testability. For example, spacing between MINI-PROBE sockets should be at least 250 mils center to center in dense situations. If space allows, make it 300 mils.
  6. It's best to set the silkscreen text labels/reference designators and their sizes at the placement stage to avoid any space issues later on for silkscreen label placement.
  7. If reference gerbers are sent as a reference, for creating a new design template for critical placement of connectors, make sure to overlay your board outline gerbers with the ones sent by the customer to ensure everything matches perfectly.
  8. Ask the customer to verify the placement by sending the pc-board file at different stages during placement to make sure everything is on track and according to the customer's requirements.

In short, checking once, double checking, and then checking any questionable design aspects again will lead to prudent and correct decisions and prevent costly and time-wasting errors.

Sponsored Recommendations

Comments

To join the conversation, and become an exclusive member of Electronic Design, create an account today!