Julie Ellis: TTM’s Interface Between Designer and Fabricator
As a field application engineer for TTM, Julie Ellis sees the problems that can occur between circuit board designers and manufacturers. I spoke with Julie at the AltiumLive event in Munich about the age-old problem of throwing designs “over the wall,” the trend towards HDI, and what advice she would give new designers.
Barry Matties: Julie, as a field applications engineer, you act as the liaison between the fabrication team and the design team primarily, correct?
Julie Ellis: Yes. And sometimes contract manufacturers when they're buying and building and assembling boards for my customers.
Matties: You're in a unique position to see the challenges and the successes and the frustrations and those sorts of things. Can you tell me about your background?
Ellis: I started in engineering at the University of Iowa and followed a boyfriend to California when I was 20. As I was finishing up school at California State University, Fullerton, I was lucky enough to be awarded a student engineering position at Hughes Aircraft. After graduation, I stayed on with Hughes as a design engineer for another five years. Then I left Hughes and went into technical sales and support.
Matties: Then you landed at TTM?
Ellis: Actually, I did a lot in between, because I was a manufacturer's rep for semiconductors, and then we picked up a board line and I really liked the circuit board side. From there, I moved to represent Nan Ya Printed Circuit Board Corporation in the U.S. for the nephew of the founder of Nan Ya out of Taiwan. When Nan Ya started pulling out of the rigid printed board business in the United States and began supporting substrates for the chips instead, I went into contract manufacturing for a small company, moved around a little bit and then literally begged for this FAE position at TTM when it opened up.
Matties: How long have you been with TTM?
Ellis: Three and a half years.
Matties: Tell us about that role. What's your typical day like?
Ellis: I check my emails at 6 am, to see what came in from China. I try to send everything out to customers, like technical queries after we've done DFMs on new parts coming in, early. Then I aspire to go outside to feed our horses and start my official work day by 8 am, but my boyfriend gets stuck doing it most of the time because work is so demanding. Between supporting new standard designs, investigating science project designs, supporting assembly issues and being “PCB problem solver,” and then starting up with China after dinner, it's typical for me to work very long hours.
Matties: What kind of horses do you have?
Ellis: Thoroughbreds and Appaloosas.
Ellis: Ex-racing. One was a lousy racer. She was so slow that her breeders never raced her. And another one started getting injured tendons at the racetrack and the owners needed to dump him. I’ve rescued and rehabilitated several injured ex-racehorses, as well as a couple others in need.
Matties: So, from horses to circuit boards, you've got it covered.
Ellis: I do have it covered, yes.
Matties: Let's talk about circuit boards. What are some of the most common problems that you see?
Ellis: First, fab drawing dimensions that don’t match Gerber dimensions and other data conflicts. Second, tight mechanical tolerances that are not supported by standard material and equipment. And third, thermal via-in-pad designs that are not properly protected to prevent solder from going down the holes. Plugging them with solder mask is the low-cost solution, but the plug depth is variable. Non-conductive epoxy fill completely fills the holes and can be copper-plated to solder to. Optical routing and via filling require special equipment and take more time, so they add cost.
Matties: Are the designers that cost-sensitive?
Ellis: Customers have to be competitive, so yes, absolutely.
Matties: When they're not willing to pay the extra that causes extra work on your end, how do you mitigate that?
Ellis: Sometimes I have to draw pictures of min and max conditions, so everyone understands what the final product they’re committing to will look like to their receiving inspector. That way, they can approve it as-is.
Matties: Here at the AltiumLive conference, after the designers speak, and it comes to the Q&A session, people have a lot of really good questions, but they're challenging, in some regard. What is it like to deal with designers in this arena and challenge the existing paradigms?
Ellis: Usually it's good. If I explain what happens in our process, or to the material, so they can understand why paradigms are challenged, they can often work with me.
Matties: I've done a lot of interviews, and one of the common themes that I frequently hear is, "Designers need to understand the manufacturing process." And then on the other side of that argument people are saying, "No, they don't need to understand the manufacturing process with deep knowledge; what they need is better feedback from the manufacturers." I've heard that as recently as one hour ago. How many times do designers request feedback from a completed project?
Ellis: I estimate designers request a preliminary DFM on 5% of PCBs purchased by my customers before they release their final documentation. Designers normally end up receiving feedback on their completed project before fabrication begins whether they want it or not, because design violations that can’t be processed, and data conflicts found during our CAM process, will be returned to them in a technical queries (TQ) form. "We request approval to remove the solder mask dams between pads on components U22 and U25 because there is not enough distance between them" is a typical example.
Matties: There's a large window of acceptability and it may not be a manufacturing issue, but there may be a strategy, or if they changed an approach that they could really improve their product.
Ellis: Strategy planning is most effective at initiation of a new design. That's why I always try to ask my customers to engage with TTM at the beginning of a design, so I can get an idea of what they're looking for and what their requirements are. Then I can propose the most cost-effective, reliable solution, including stack-up, pad stack, and design guidelines before they begin routing.
Matties: It’s great that you provide that service, but often times I hear that the designers throw the design over the wall. "Here you go; build what I asked for." And at that point it's too late, but if they're in that window of acceptability, how do they improve if there's not a proactive feedback loop and if they don't take the time, which could be a lot of time, to really understand the PCB manufacturing process?
Ellis: Busy designers can still request and use suppliers’ DFM guidelines for lines/spaces, aspect ratio, min. drill size, min drill-to-Cu, min Cu to fab line, dimensional tolerances, etc. I’ve been very fortunate with my on-going TTM customers, who generally appreciate and utilize up-front design guidelines and stack-up support.
Matties: Maybe you are the exception, because I hear this so often.
Ellis: I run into this more on designs that were previously prototyped in North America and then get thrown over the pond to Asia. TTM provides “Global Seamless Transfer” from site to site through our global front-end engineering programs and procedures. But a lot of work goes on in the background to make that transition as seamless as possible. So if I'm working on a new stackup, I qualify the product. "Is this something that's only going to be a few prototypes here, or is this something that you want to plan for migration to Asia?" And if they say, "Migration to Asia," then I will propose a preliminary stackup just to see if that's about what they're looking for. If the answer is yes, then I'll ask Asia to create a similar production stackup using their common material and copper thicknesses. Then I will request that the North America site stack matches the Asian production stackup as closely as possible.
So that's the background work that we do to achieve that global seamless transfer. It's more difficult if somebody throws a design over the wall and says, "Build it here and then I want the same thing over in China.” I’ve worked on a few of transfers, where TTM wasn’t involved in the prototype, and they can be really painful to the point of having to completely re-route multiple large, complicated multilayer boards that couldn’t be transferred to Asia as designed.
Matties: How many customers engage in the “throw it over the wall” mentality?
Ellis: Usually it only happens once or twice with a new customer, and then when they see the pain of not working with us first and they see the value and all the work that I do to assure the best fit over in Asia, they learn to come to me in the beginning.
The “Throw it over the wall mentality” happens more from people at contract manufacturing, whose kitting and assembly responsibilities begin with already-completed design documentation.
Matties: TTM has a lot of resources, and a lot of the fabricators in North America do not have a position like yours. You're a gift to your customers.
Ellis: Free engineering.
Matties: And if you can get the spin right the first time, it's not just free engineering; they're gaining a lot of advantages in time and cost.
Ellis: Also, we have so many resources within TTM. We're nearly a $3 billion company and the engineers, even all the way up to the top guys, the business unit presidents, are all circuit board technologists. There is so much expertise in this company that if I don't know something or if I think, "I'm a little bit worried about that design because of the high voltage that the customer needs," I have expert resources to go to. That's extra engineering and knowledge that the customer wouldn't even have access to if they weren't working with a company like TTM.
Matties: What sort of trends are you seeing in board design right now?
Ellis: High voltage for electric vehicles, HDI stacked microvias for Autopilot, and a global need for stable, high speed, low loss materials to support automotive radars and 5G.
Matties: That's always a challenge and will be the continuing challenge, right?
Ellis: A huge challenge, yes.
Matties: I'm hearing a lot of talk about HDI at this conference; are you seeing a trend in that?
Ellis: Definitely. The semiconductor manufacturers keep reducing die sizes, which shrinks the component pad diameters and spaces between them. We have to use laser microvias to route I/O down to internal layers whenever there is a multi-row BGA with pitch of less than 0.65mm. The smaller microvias (4-5mil) have smaller pads (10mil) than mechanical through-holes (8-12mil drills with 16-22mil pads, Class 2), leaving more distance between pads to run traces. Annular rings and antipads chew up a lot of space, and they leave controlled impedance reference and ground planes looking like Swiss cheese. So reducing the ring diameters is better.
Matties: We have to get rid of those rings.
Ellis: Yes. In fact, at the last Altium conference Happy Holden showed us micro-sections of padless vias from last century.
Matties: There were mixed reactions to what Happy was talking about.
Ellis: I’m not aware of any high volume process to support it, or reliability data. But I believe there is a consortium forming to investigate this.
Matties: Because some people were saying, "You're not going to lower the cost.” In any case, I just keep hearing comments like in Lee Ritchey's talk, if you're not doing HDI, you will be doing HDI. What was your takeaway from all of that?
Ellis: HDI is a necessity once high I/O BGA pitches get down to less than .65 mm pitch. But electric vehicles and other high power devices will still require thicker copper and larger linewidths supported by conventional circuit boards until a new conduction method is invented. Plus, there are other registration and material reliability issues that need to be considered when drilled (especially mechanical) holes get too close to each other and/or adjacent copper features.
For simplicity, I separate rigid circuit boards into six categories (see chart) and then confirm the stack-up and DFM guidelines from there.
Matties: What's the best part of your job?
Ellis: I love it when I’ve worked with designers on a complex design, we build the prototypes quick-turn, send them to an assembler, and I get that call from the designers informing me the assembly works as designed!
Matties: You're saying not every customer uses your service until they suffer some pain.
Ellis: Well, Sales Rule #1 is “Find a customer with a problem and provide the solution.” But not all customers have painful crises. So, most of my work is sharing information and DFM guidelines that help designers design circuit boards that can be fabricated.
Matties: What advice would you give a designer coming into the industry?
Ellis: Learn about the fabrication processes so you can be sympathetic and aware of what tolerances are really involved, and how much trade-off analysis we fabricators have to review just to plan a job. Talk to your manufacturing and test engineers, so you understand clearances needed for your board to run down an assembly line and how test points need to be designed and accessed for good coverage. Lastly, pick your experts wisely. Don't always go with the guy who's saying, "You're doing a great job," or "Yeah, I can build this." Listen to who is saying, "Sorry, I can't do this because of this...but I have a different option." Pick your experts wisely.
Matties: We were listening to Lee Ritchey's keynote yesterday and during the Q&A question part, a question was regarding a PCB fabricator, and Lee’s response was, “get a different fabricator.” I don't know if that was good or bad advice. What do you think?
Ellis: It was somebody who had controlled impedance, and normal tolerance should be plus or minus 10%. But these boards came out at plus or minus 15%, and the fabricator was asking for an exception. The customer was asking, "Should I accept it?" And Lee said, "Find another fabricator." As a fabricator, my blood kind of boiled, because we didn’t have enough information. Small lines and spaces are more sensitive to processing than large lines and spaces, but process control and tolerances for a given Cu thickness are fixed, regardless of the size of the features. So, requesting additional tolerance on impedances of fine lines less than 2.8 mil is reasonable.
Matties: Is there anything that we haven't talked about that you feel like the industry designers or fabricators should know?
Ellis: YES! Keep up the good work! Every day, I engage with sharp, respectful humans, who strive to solve incredible problems together. You are amazing in the creativity you bring to the table, and your products are changing and improving on a daily basis. Remember to be in awe of the vastly complex concert of chemistry, physics, mathematics, design, materials, processes, and manufacturing that compose physical electronic products, and that we are so lucky to be contributing members of this new evolution of technology!
Matties: Julie, thanks for speaking with me today.
Ellis: Thank you for the opportunity.