On the eve of the industry’s largest PWB show, I have begun to reflect on what the American expats whom I will meet here are doing for their Greater China employers. I wonder what they could not have done in the United States, and the reasons why. Now the game has changed. Labor costs are up. Automation is advancing. Business continues to consolidate rapidly as “blue chips” have vanished. Others are in jeopardy. Those remaining are few and far between.
One thing is certain. The days when Asian companies would buy old, tired American facilities and fix them up as a potential method of market entry are long gone. They have learned that it is better to invest in the newest technologies at the outset and not waste their funds patching and fixing near-obsolete operations. This could be one of the motivations for some American corporations to design and install new state-of-the-art automated and “green” captive PWB operations. One such venture is even predicting a bare board cost reduction of greater than 30% over his currently outsourced panels.
First a disclaimer: I am not an electrical engineer. I am only fluent in my knowledge of PCB fabrication. In other words, much of what I am about to share is borrowed from well- educated and experienced EEs who don’t share my aversion to math. I am, in effect, jumping into the deep end of the pool, but only because I am wearing my (virtual) brain floaties! For this reason, I ask for your forbearance as I attempt to translate this left-brained subject matter into my right-brained mother tongue. Here we go…
Substrate copper application methods. I have been discussing surface finishes for the last couple of posts, and I would be remiss if I didn’t cover the crucial topic of copper surface roughness and how it specifically impacts high performance PCBs. Most substrates are copper clad with either rolled annealed (RA) copper, electrodeposited copper (ED) or reverse treated copper (RT). I have put some links below should you want to learn more about each type of copper and the resulting surface roughness of each.
When the copper surface is rough, even at microscopic levels, the effective conductor length grows and the resistance increased as the signal must move up and down with the topography of the copper surface. To the naked eye, copper-clad substrate appears very smooth but when you view the surface under magnification, the copper can look like something akin to the Himalayas! For this reason, some choose RA copper that is both smooth and consistent in thickness. RA can be more costly, however, and not an option for all. ED copper has the roughest surface but, depending on the application and speed requirements, may be perfectly adequate. RT copper is smoother, and doesn’t cost more than ED. Yet with RT copper you need to be careful about potential delamination and poor peel strength. Once again, we are back to trade offs and specific design needs!
Rogers has a “LoPro” laminate series that is extremely smooth while Taconic offers an extremely smooth “reverse treated/ED” clad laminate. These laminates aid in good, sharp etch definition as well. In some cases, these are smoother than RA copper and can be less costly.
Skin Effect Considerations
Michael Ingham of Spectrum Integrity shared the following illustrations with me in regards to the skin effect and how a signal flows through a conductor. In the first illustration, he shows a rough cross-section of how a very high frequency signal tends to run on the outermost areas of a conductor—creating the skin effect. However, he notes that this is only true when there is not a ground plane underneath.
In the second and third simulation illustrations he used a full 3D field solver for a top layer trace that has a ground plane beneath.
The second illustration shows how a current will flow through the entire trace cross section area at low frequency.
Finally, in the last illustration, he shows the current distribution at 20GHz, which is mostly at the bottom and sides of the trace.
Michael’s point is that when it comes to surface finishes and texture, the most critical issue appears to be the smoothness of the copper. He has raised the question that if the high-frequency current mainly flows on the bottom and sides of a trace—is using nickel really causing the unwanted losses, or is copper roughness the culprit? Many fear using ENIG-plated traces and go to great measures to avoid and resort to using costly mixed plating, etc. Below is measured data of a RF trace with standard ENIG plating. The overall loss may be surprising.
This is just scratching the surface (no pun intended!) in regards to the topic of conductor losses and copper profile. When dealing with coupled trace structures, the effects of nickel could have a big impact due to having adjacent trace walls interacting. But I will leave this topic for Michael to cover at another time!
Spectrum Integrity has been very successful in high-speed/high performance designs, and has done so focusing more on design technique and paying great attention to all the properties of materials such as the smoothness of the copper rather than focusing on surface finishes or restricting to very low dielectric loss materials. They have enjoyed very good success with avoiding costly mixed plating and with the use of smooth copper laminates like the Taconic RT/ED material.
When it comes to the smoothness of outer layer copper traces, a board fabricator can go a long way to hurt or help your desired results! The higher the speed/frequency and more critical the application, the more you need to be working with a company like Transline Technology who understands high performance board fabrication. For instance, when we clean the outer surface of the boards, we do it with chemical cleaners that are non-abrasive and maintain the smooth copper outer surface. Additionally, there are many points throughout the manufacturing process where standard practices of PCB handling can also compromise the otherwise smooth copper outer surface that can create havoc for your design performance. High performance boards can appear deceptively simple in their design, but there are many intricate details that must be considered when making a sound high performance board. Many a board fabricator does not possess the knowledge or the trained eye for the subtleties of high performance PCB fabrication.
Conclusion. Skilled and well-informed partners are the key to success when it comes to choosing surface finishes and materials for high performance designs. These issues are complex and have many nuances that must be considered to create successful products. As such, it is critical to forge strong working relationships with both your advanced material suppliers and your board suppliers. By doing so, you will save much time and money and avoid a host of needless headaches. I have listed some additional resources below. Many thanks to Michael Ingham of Spectrum Integrity, who is always teaching me something! I highly recommend Spectrum Integrity for RF/MW and high performance design. Their website link can be found below. As always, I welcome your input and comments! [email protected]
This development out of London holds interesting implications for printed circuit board manufacturers. Given the vast size of the fuel cell market, to potentially insert a PCB in every one (in addition to the boards going in inverters and other related equipment), offers a massive new market for fabricators.
Productronica, the trade show that’s so big, they can only hold it every other year, enters its final day today.
The PCB fabrication exhibits have shrunk over the years and are now down to about one hall (although exhibitors were spread over two, intermingled with large lounge areas and contract assemblers). Like the (much bigger) assembly sections, the exhibitors felt Tuesday’s traffic was slow, but Wednesday and Thursday were strong. While China may have spirited away most of the production, this is still the event outside of the Pacific Rim.
Ours is a boom/bust industry, characterized by periods of substantial growth followed by years of sheer gut-wrenching pain. I’ve been through four of these cycles in my 20 years in PCBs, and what’s clear to me is that, overall, there is ample capacity worldwide to suit the PCB industry’s needs.
Granted, a large amount of the funds are earmarked for machine upgrades, and Viasystems has not publicly stated how much of the investment will be geared toward additional space. Not to pick on Viasystems, because this problem clearly is industry-wide, but what I’d like to see is more than three quarters of growth coupled with the possibility of more to come before the industry starts ramping capacity. Pricing has long been a problem for PCB fabricators. Adding capacity won’t fix that.