These days, when it comes to high-speed PCB designs, you may not know precisely where problems can arise. There are a variety of potential problematic areas that can be based on manufacturing or process-related problems, components, or fabrication-associated issues, or something else that’s hard to detect even with the most advanced PCB inspection.
A high-speed board may be perfect as far as BGA assembly. All the balls properly collapse; all the thermal profiles are accurately determined and performed. All soak temperatures, pre-heat, soak and cool-off periods are all within manufacturer limits and ranges. A sample thermal profile is shown in Fig. 1. Yet, this high-speed board fails at high speed at the time of system functional level testing in the system.
It becomes considerably worrisome when PCBs fail at certain high-speed level bands. Investigation as to the reason(s) why they fail include checking out the process and dissecting it step-by-step and scrubbing the bill of materials (BOM) all the way from A to Z to make sure there are no issues during high speed verification testing.
To learn more about this particular issue, read our SMT Magazine article.
In the meantime, here are some tips and hints to help you understand what’s important in tracking down problems associated with high-speed PCB designs.
- A mixed material board with special processing and temperature treatment may require an even closer scrutiny to determine if it may have triggered problems.
- Check out how your board was fabricated; what were the materials used; was it straight FR4, a hybrid, Rogers, or polyimide?
- Take another look at the coupons run at fabrication and the tolerances you, the OEM, provided. You want to check out if tolerances are adversely affecting impedance calculations.
- Get a good understanding of the components used on your PCB. There are countless questions that need to be asked. Find out if you’re using marginal devices. Those can be questionable for the required performance at high speeds.
- Lots of high-speed PCB designs use FPGAs. Find out if the FPGA design engineer may have inadvertently or wrongly programmed and designed the user-defined pins.
- If you suspect an FPGA is at the root of the problem, team up with your EMS Provider and go to the chipmaker to get it resolved.
In the case of a questionable FPGA, it’s in the best interest of the OEM to assure it is tested at their facilities at extremely high speeds at different permutations, combinations of different voltages, and different temperature settings, vibrations, cycling, and zoning. This way chances are all the faults relating to that FPGA are found during the OEM’s testing, validation, and debugging phase. By doing so, that same FPGA doesn’t have to go through the EMS cycle to locate those errors.