Join us in The Lobby to talk all about DFM—Design for Manufacture, a crucial but complex piece of the circuit design process that not all engineers have mastered.
In our eighth episode, host Dave Finch discusses how to get from prototype to market, zeroing in on the process of DFM or, more accurately DFX, which stands for "Design for" ...well, anything that comes after the design.
Why does cost go up after a product's been released for a while? When should you consider certification in the design process? How can young engineers hope to familiarize themselves with DFM if they never even designed a full board in school?
Learn the many good reasons engineers should know their board manufacturers. Consider how EMS (electronic manufacturing services) companies have evolved from mere contractors into design partners.
And hear some truly unfortunate stories about what happens when you fail to plan beyond the circuit design stage appropriately.
A Big Thank You to This Episode's Sponsors
The Voices You'll Hear
This episode, you'll hear from two experts in manfacturability who come from different backgrounds but share the same core message:
Dr. Dongkai Shangguan
IEEE Fellow, IMAPS Fellow
Mark Hughes
Research Director at Royal Circuit Solutions
"Design for Manufacturability" Annotated Transcript
Dave Finch:
One critical component that single-source is going line down. So we need to redesign that part of the circuit. And now you need to start all of those certifications over again.
Mark Hughes:
Yeah. That would suck. You know something? That happened to me.
Dave Finch:
From EETech Media, this is Moore's Lobby where engineers gather to talk all about circuits. I'm Dave Finch.
Today, two trusted industry experts layout—pun intended—the basic design milestones to prepare your circuit for production.
After our recent episode on prototyping, a number of you reached out asking what are the steps involved in preparing a design for manufacturing? And while I had the general idea—you know, testing, design for manufacturability, BOM optimization—as an applications engineer, it was nothing I'd ever had to worry about. So rather than guessing, I decided to invite to the lobby today two experts in electronics manufacturing.
Our friend Mark Hughes joins us from Royal Circuits to share his perspective from deep inside the trenches.
Mark Hughes:
A new engineer is going to be so overwhelmed by the amount of information coming at them, by the number of variables they have to balance, that they're not going to be able to consider DFM except as a separate step in the design process.
Dr. Dongkai Shangguan (1:42)
Dave Finch:
But first, Dr. Dongkai Shangguan is an IEEE Fellow and an IMAPS Fellow with 30 years of experience leading R & D in global electronics manufacturing. Throughout his career, Dr. Shangguan has accelerated the introduction of innovative technologies and products into high-volume manufacturing.
Dongkai, thank you so much for joining me in the lobby. So you've had this legendary career, which includes working at an automotive electronics OEM, then getting involved in IC packaging, and then bringing all these talents to one of the leading electronic manufacturing services companies or EMSs. You're decorated with all these industry awards. You've served on boards of industry associations, like IEEE EPS, iNEMI, IPC, and a few others. And you even shared your expertise through books and papers, keynote speeches, and, man, this list just goes on and on.
What's the First Step for DFM after Validating a Prototype? (2:40)
Dave Finch:
So with all of that in mind, this might seem like an oversimplified question, but let's assume that we have completed our prototype design and we verify that all the components are available. We think we're ready to manufacture the board. What is the first step after validating the prototype?
Dr. Dongkai Shangguan:
So, I would say, if you have selected the EMS as a partner, you'll start engaging with their team. Where you'll start, I want to start always like a DFM review.
Dave Finch:
Design for manufacture review. Okay.
Dr. Dongkai Shangguan:
Right. Because most of the time, designers, you're one company and the manufacturing is taking place in a different company. And the design team have never been to a manufacturing factory. And so manufacturability may not have been the number one priority for the design team.
So now we are going to bring the product to manufacturing, high volume manufacturing, manufacturability can become a very critical issue. So I will start with a DFM review and make sure that the manufacturer and your own team look at the design and make sure that the design meets the design rules, DFM rules, and such that when the product is introduced into high-volume manufacturing, they can get a high yield and a low cost and reliability and all of that.
Dave Finch:
Right. That's a good point, actually, because not every design team is sitting in the manufacturing location or working side by side with the manufacturing engineers.
So what you're saying is start with the DFM process to make sure the design is reliable, cost-optimized, and will be produced with the highest yield.
Dr. Dongkai Shangguan:
Also the EMS can make the recommendations about how you might be able to improve your design using alternative processes, alternative designs. And so optimize the design for manufacturability, for high yield, for cost, and for many other benefits.
DFX ("Design for X") Rules and Guidelines (4:54)
Dave Finch:
When you begin the DFM stage of the process, are there certain guidelines for the types of applications that you're working on or is there-?
Dr. Dongkai Shangguan:
Right. That's a very good question. By the way, when I say DFM, I really meant "DFX" because there's a design for test, as well. DFT: design for testability. And design for other attributes, design for cost, reliability. But just using DFM, manufacturing, as the example.
So many companies have DFM guidelines or design rules. What is the component spacing? What is the through hole size? And pad size? And many other things. Very, very detailed design rules for different components. BGA, different BGA, BGA pitch and the CSP. And normally, there is an EDA tool that helps automate the DFM review process that would drive the review using those specific design rules to check the design, each individual design, make sure that all of the design features meet the design rules.
Dave Finch:
And when you say, making sure that all of the different aspects of the design meet the design rules—yes, it's for things that you just mentioned: pad size, pad spacing, component footprints—but also the CAD software is checking for things like trace dimension, spacing, and so on.
Dr. Dongkai Shangguan:
Right. I mean, there are PCB fab design rules and there are PCB assembly and mechanical assembly rules for different process steps. There are specific design rules.
Dave Finch:
Oh wow. Okay. So even assembly, all of this is specified in design rules.
Dr. Dongkai Shangguan:
Right.
Dave Finch:
And then in the real world, you made a really good point about design for cost optimization. So maybe taking a look at—are we introducing more components on here that we need, or the more expensive components.
Dr. Dongkai Shangguan:
Right. And also maybe the process steps. If it involves a very complex mechanical assembly, for example, you want to look at assembly steps, tolerances. You want to look at accessibility. And then there are maybe some that are based on the experience of the engineers that they can leverage to give you recommendations, as well.
Dave Finch:
And like you said, there's a lot of human intervention in this as well. And there are the EDA tools, but I think the wisdom that comes with human experience is probably a major part in any of these "design for X" exercises.
Dr. Dongkai Shangguan:
Right. And it's human experience assisted with the EDA tool.
Partnering with EMS (Electronic Manufacturing Services) Teams (7:57)
Dave Finch:
So tell me about partnering with the EMS engineering teams, which you mentioned earlier.
Dr. Dongkai Shangguan:
So you start with the DFM review and that hopefully gives you a very good partnership between your design team and the manufacturing partner. And then you'll go from there—you'll get into NPI, new product introduction. And once you reach a certain volume, certain yield, then you ramp up and get into a higher volume.
Dave Finch:
Yes. And when we are talking ramping up into production volumes, that's where a longterm partnership is especially beneficial. And at production volumes, let's say it's one million units per year, even one 10th of a cent added to the manufacturing cost can make a real impact on your margin.
Dr. Dongkai Shangguan:
Exactly, and many products are made not only in the millions, but probably in the billions in terms of the number of units. And so that cost per unit really can multiply very fast.
The Learning Curve: Design for Functionality to Design for Manufacturability (9:34)
Dave Finch:
In your experience, do a lot of engineers... are they good about designing forward knowing that they have to make this a high yield, highly profitable circuit design? Or do they look at the functionality more than anything, and are surprised later by things like cost optimization?
Dr. Dongkai Shangguan:
I think it depends on the individual designer, his or her experience. It also depends on the maturity of the product. Because I think most of the time, we are talking about startups, and at the early stage of a startup, your main objective is to design something that works. And so, functionality is a high priority.
Over time, as you have proven the design, the functionality, the priority is shifted more towards manufacturability, yield, cost, and reliability. And then of course there the things we just talked about become more important.
It's a learning curve. It's a learning curve for each individual as a professional, and it's a learning curve for a new company. It's a learning curve for each program.
Dave Finch:
Totally, yes, totally. A learning curve for each person, for each company, and for each program. That's exactly right.
Dr. Dongkai Shangguan:
And hopefully with experience, with collective knowledge, the learning curve can be accelerated.
Dave Finch:
Yes, and accelerating that learning curve means, eventually you will bump into the limitations of your capabilities, and there will be an unsuccessful outcome. But you take away, and you understand, okay, next time we design this we are definitely going to look out for these factors. You keep accumulating that wisdom throughout your engineering team.
Dr. Dongkai Shangguan:
Right. And also very importantly, you want to have a long-time partner, an EMS partner that can work with you, not only just through your manufacturing, but hopefully that can engage with your team early in the product development cycle and give you recommendations, and DFM reviews early in the design cycle.
Dave Finch:
Exactly. Engaging your manufacturing partner early enough in the design process that you can design these things efficiently.
Dr. Dongkai Shangguan:
Right, exactly, because it's a lot easier, it's a lot more cost-effective if you make design changes early in the design cycle.
Dave Finch:
Exactly.
Dr. Dongkai Shangguan:
Versus finish all the design, you build the prototype, you test it, then you give to somebody else to do the DFM review. Then they come up with a lot of very good recommendations, but you say, "Oh gosh, it's too late for me to change," because it's going to cost money, it's going to cost time—so it's too late.
Dave Finch:
A key mistake, a fatal mistake that an engineer could make is waiting too long and making assumptions about the manufacturing worthiness of their design before they get in touch with a manufacturing partner.
Dr. Dongkai Shangguan:
Right, so that's why you need the partnership. You need the trust between the design team and the manufacturing partner.
The Evolving Role of Electronic Manufacturing Services Companies (13:06)
Dave Finch:
Are a lot of the engineers at the manufacturing service companies—whether it's Jabil, Flex, Foxconn, whoever—are a lot of those engineers also familiar with the design engineering, and are equally experts in the manufacturability? Or do they have more expertise in taking an existing design and making it manufacturable?
Dr. Dongkai Shangguan:
Certainly there's a lot of expertise in the EMS companies in high-volume manufacturing, DFM, and optimizing the design for manufacturing. But more and more, I think over the past decade at least there has been very, very strong design engineering expertise in the EMS industry, as well.
You mentioned ODM. Many of the EMS companies are doing design as a service, either a complete design or pieces of a design, depending on the relationship. Many EMS companies have strong design engineering experience on their teams that can be leveraged.
Dave Finch:
Yes. Yes, I've noticed that. When I first started in the industry a little over 20 years ago, for the most part, the EMS companies that I was dealing with more in the industrial electronics space, really wanted to take a complete design, and then take it from test and into production. They didn't offer too much front-end design service.
But over the years, I've seen a lot more EMS and ODMs offer very strong design engineering capabilities, as well.
Dr. Dongkai Shangguan:
Yes, so the industry certainly has evolved and made a lot of progress. And, exactly as you said, 20 years ago, EMS was just an outsourcing supplier for capacity overflow. In those early days, many OEM companies were doing manufacturing themselves. But when they would run into capacity issues, they go to an EMS.
But that was 20 years ago. And since then, so many things have evolved, and a lot of progress has been made, and EMS companies have taken on more and more technology capabilities, and design engineering expertise. They have a lot more to offer today in technology, in engineering, in design support and services, and they can really be a very, very strong innovation partner for the design community.
Dave Finch:
Yes, absolutely. Absolutely agree with that.
Considering Certifications, Cost, and Manufacturability in the Right Order (16:00)
Dave Finch:
Let's talk for a minute about agency compliance and agency approvals, because at some point, this will either need to be a product that passes UL or CE certification, FCC certification, whatever the case may be.
Is this something that occurs after you've done the DFM and verified that, yes, I'm certain that this is the most reliable, cost-optimized circuit, and now we go into agency approvals? Or is that done well beforehand?
Dr. Dongkai Shangguan:
I would say, if you look at the certification, then on the one hand there is certification for the EMS partner, for their operations. There's automotive certification, there's medical certification, and others. So, many of the factories of an EMS company have some of these certifications for their manufacturing operations.
Then you also look at the product certification—the ones that you just mentioned, Dave, like FCC, and the other CE, and UL, these are more product specifications for each specific product. And there, once you build your prototype and you finalize your design, I think that's where you can engage with an agency or certified lab to perform that product certification.
Dave Finch:
Right.
Dr. Dongkai Shangguan:
Because you need that before you ship products.
Dave Finch:
Exactly. Get the design right upfront. Get it manufacturable and then get the agency approvals as sort of a later step.
Dr. Dongkai Shangguan:
Right. You don't want to do it too early, because if you're going to change the design, then you have to do this all over again, right? So you want to do it as soon as you have finalized the design. You'll have done your DFM review and your reliability test. You'll say, "Well, that's the design I'm going to take for the market."
You want to do it as early as possible because it takes time—some will take quite a few months. So you want to get a start early but no too early. You want to do this after you'll have finalized the designs.
Dave Finch:
Also again, another reason why it pays to have the manufacturing partner engaged as early in the process as possible to avoid these lengthy delays in the design process.
Dr. Dongkai Shangguan:
Right. Early engagement is a key word here for a good relationship. If you really want to leverage the value of an EMS partner.
Dave Finch:
Yep. There's also cost out. You mentioned earlier in the conversation, the first generation of anything will probably not be the most cost-optimized. It will probably be the most electrically functional for that given set of features, but I think maybe in a company's efforts to get a first-generation product out the door quickly and into the hands of consumers, they have opportunities they discover to maybe further cost reduce in the second generation and beyond.
Dr. Dongkai Shangguan:
For sure. For every product, there is a cost curve.
At the beginning, there is this novelty factor. You can afford to sell it at a higher price so your cost pressure is not so high.
But over time, as a product becomes more mature, that novelty factor dissipates and so you'll have price pressure, then that translates into cost pressure. And so those cost optimization efforts become very critical. And there the EMS can come in and offer value engineering. You know, work with you. They can see how product design can be optimized. The components can be replaced with some alternative that's lower cost, but the same functionality and reliability parts that are optimized for cost, as well.
Dave Finch:
Right. Have we left anything out? Is there anything else that an engineer should know if they're going into a role where they're going to be designing products that will be manufactured?
Dr. Dongkai Shangguan:
I think we covered the most important areas, Dave, and I would just encourage designing engineers to engage with EMS partners and see what they can offer so your design team can focus on their most value-add, and then leverage the partner for everything else.
Mark Hughes (21:50)
Dave Finch:
Our next guest is a good friend of All About Circuits and even joined us on our very first episode of Moore's Lobby earlier this year. Mark Hughes is Research Director at Royal Circuit Solutions in Hollister, California. Mark, welcome back to The Lobby.
DFM vs. Young Engineers: Go Talk to Your Board Manufacturer (22:05)
Dave Finch:
My experience—I never had to design anything where I was the one responsible for its manufacturability. As an application engineer, I could provide pseudo code, I could provide reference schematics that were so not optimized, and it was just assumed that there would be some other design team who would take that and say, "Thank you very much. This is how a class D amplifier works." And then they would get it ready and make it productized.
So I can speak from personal experience, I never even considered what would go into manufacturability.
Mark Hughes:
Somebody who's got years of experience in this—it's ingrained in them. It's part of the design process from the very, very beginning.
A new engineer, though, doesn't have that capability. A new engineer is going to be so overwhelmed by the amount of information coming at them, by the number of variables they have to balance, that they're not going to be able to consider DFM except as a separate step in the design process. There's just too much.
Dave Finch:
What would your advice be to that inexperienced engineer who's listening to this thinking, "Oh, [bleep]. I'm going to have to... This thing's going to have to go into production and I don't even know what that looks like."
Mark Hughes:
That's a great question. So, some shops, like mine at Royal Circuits, we offer free engineering. It's a value-added service. You've got a question and we'll put you in touch with the line engineers that are going to have to make this thing, and they will walk you through how to make things better, cheaper, faster, whatever.
To give you an example, one of my favorite Royal Circuits employees is Elijah Gracia. He got a board in and he saw that the designers had included a single buried via. He went and he talked to them and said, "You know, guys, this is going to really increase the cost of your board. Is there any way we could maybe do a different arrangement and create a blind via at least or something along those lines?"
And he went back and forth with him a couple times, and then he finally found out enough about their design, he said, "Just make it a through hole via." Now we were making a thousand panels or something like that, but that simple removal of that one buried via from the design saved that company $50,000 on a thousand-panel order.
Dave Finch:
And that's exactly the kind of thing I'm talking about where I was probably the first person to just think, "Well, this is a cool way to solve this problem." Not even thinking about the cost that adds, or even maybe reliability issues, maybe not in this case. There are so many factors.
Mark Hughes:
Yeah. So get out there and see how boards are built. Talk to the engineering. If they offer engineering services, make use of them.
One of the things that we've been doing lately in the time of COVID is offering free webinars to companies. So they just call us and say, "Hey, my skill set's a little weak in designing for manufacture. Can I put my engineering team on with your engineers?" And yet we set up the web call and they ask whatever questions they want to ask, and we answer them.
So find out if your company has that availability. In my experience, most of them will talk to you, some more than others. It just depends on how needy you are. But talk to the company that's going to build your board.
Spectacular Failures (Or, Learning via Schadenfreude) (25:36)
Dave Finch:
Can you think of any spectacular failures that have arisen because somebody said "No. My design is perfect. Just go build it."?
Mark Hughes:
Oh yes I have.
There's another podcast that you might not be aware of and it's Mike Konrad, Aqueous Technologies. His podcast is Reliability Matters. He's got this great example of a lawsuit that happened between a contract manufacturer—this is an assembly house—and a company. And the reason was the company had their products fail. And they were, let's say obnoxious, about the way they were treating the assembly house saying, "No, Build it the way I told you to."
The assembly house went back and said, "You know we really don't think this is a good idea, but you're the customer." And the assembly house was very much trying to help the customer out and the customer just told them to stuff it, so they built it.
The product failed in spectacular fashion, and the company obviously wanted their money back, and the assembly house said, "Listen, we just did what you told us to." And the lawsuit was, well, you should have known better. You should've not listened to us. You should have pushed back harder.
And I would say that every time that I am walking through the shop and I see artwork on a screen, layer artwork, or I see a design, I can oftentimes look at it and say, "I wouldn't do it that way." And sometimes it matters and sometimes it doesn't. But the times that it matters, it's usually where we can save the customer, depending on how many boards they order—four, five, six figures of cost if they would just be willing to listen to our guys.
Dave Finch:
I suppose an even worse case would be you finish your design, or so you think, you get it through all the certifications that you need, and all the compliance and regulatory stuff. It comes back approved, and then they say, "Oh, but you know what? We just found out that this one critical component that single source is going line down, so we need to redesign that part of the circuit." And now you need to start all of those certifications over again, right?
Mark Hughes:
Yeah. That would suck.
Do you know something? That happened to me. Not even that long ago.
March of last year, I had a design and the manufacturer's like, "Oh, you have to use this chip. It's great." We went through all the specs. It really is a fantastic integrated circuit for what it did, and I got through the design and everything relatively quickly. I think it was three months from even just starting to talk about the project, to having product in hand.
And as I was about to have some more built to start selling them, I talked to the manufacturer and was like, "Hey, this is great. You were right about this chip. It outperformed in every metric we have." He's like, "Yeah. I've got some bad information for you." Like, "What's that?" "Well, the ASIC manufacturer is no longer going to produce that chip."
Dave Finch:
Oh no.
Mark Hughes:
I had to get a new chip.
Dave Finch:
The ASIC, by the way.
Mark Hughes:
Yeah. Application-specific integrated circuit. Yeah. It's done. It's... You're done, and I might–
Dave Finch:
You're not going to find a lot of those in the catalog.
Mark Hughes:
No. I'm like, "Really? Well, that's unfortunate." And I'm like, "Remind me again. You told me to use this chip, right?" He's like, "Yeah."
Dave Finch:
See, that's why I was always just as pleased as could be, to be in the application lab thinking, "Glad this isn't my headache." Oh man. That's the nightmare scenario right there.
Mark Hughes:
Yeah. Well...
Dave Finch:
You know, but it is kind of bringing to life a really good point you just made, which is that I think a lot of us have EMS or assembly houses, a picture in our head of, "You send us the design and we just print it." Right?
But it's not that. You do have engineering resources on hand, and these should be thought of as an extension of your design team—if for no other reason, then to get yourself into production faster.
Mark Hughes:
By all means. These people, all they do all day are make boards. Maybe a good engineer, somebody whose pretty busy, maybe they put out six to 12 designs a month while our engineers see six to 12 designs a day, or more. Absolutely use the line engineering and the engineer, the CAD/CAM engineers, as an extension of your resources.
In my experience, they're often happy to help. They might be busy. Might take a little while for them to get back to you, but get them on the phone. They love to share. They love to save money.
Don't Forget Development Environments from CAD/EDA Companies (30:41)
Dave Finch:
Hey, what would you say are the top three, if you could identify the top two or three CAD or EDA packages that you work with or that you see your clients working with?
Mark Hughes:
Oh gosh. Certainly Altium, Cadence... The Cadence tools, the Altium tools. Now for a third, I would probably say just because we get so many orders from a certain conglomerate that you're probably looking at Eagle or KiCad or one of the lower, I would say, entry-level tools just based on volume alone. But I would say the professional engineers are on mostly professional tools—Altium Designer, which is a fantastic tool. I can't afford it myself, but it's great, and then the Cadence Allegro Ecosystem is up and coming as well. That's pretty popular.
So if you talk to Altium and they're... You really should because they've got great people. Instead of looking at design for manufacture, which is... as we talked about earlier, a lot of people just think about that as a pass you do at the end of the design process. Really what they're looking at with Altium is design with manufacture, right?
We can have constant back and forths with the customer through every step of the design. As they're going through, they have a question, they send it over to us, we can answer it for them, send it back to them, and we can have back and forths. That's really the direction that Altium's aiming. They want you to be intimately involved with your fabricators from the very beginning.
Dave Finch:
So it is worth a worthwhile for obviously any engineer, but certainly the younger engineers to get acquainted with the sort of enterprise-grade design tools and ecosystems?
Mark Hughes:
Yeah, and it also makes them a more valuable employee. Remember that board design is often not taught in our major colleges and universities. You can get a degree in electrical engineering, a bachelor's degree, having never built a board.
Dave Finch:
That's right. That's right.
Mark Hughes:
Which I think is criminal. I mean, what else is electrical engineering, if not design?
So, if you're not training these kids how to do it, I mean, what are you doing?
But that's what we're doing. So, I would say if you're a young engineer, find a way by hook or crook to get your hands on that stuff. Form a club, build a robot, contact Altium, and they will find a way to work with you. I promise you that.
You have to develop an intimate understanding of how we're going to make your boards. Are we going to use cap or foil construction? How are we going to make your vias?
All of these things, you have to slowly develop that understanding, and you're not going to be able to catch it all the first time you go into a board shop and look around. You're not going to be able to pick up everything. If you read articles, you're not going to be able to pick up everything, but grow that knowledge little by little, bit by bit, over time until you could theoretically walk in the shop and build your own board.
You can develop this mental picture of how somebody would physically construct your board. Then, you're probably making a board that's manufacturable.
Dave Finch:
So there you have it. Big thanks to our guests, Dr. Dongkai Shangguan and Mark Hughes, who joined us from their offices this week.
Certainly cleared things up for me, but we'd love to hear from you.
Have you developed your own workflows to prepare your designs for manufacturing? How early do you engage with EMS? Do you engage with an EMS at all, or is it all internal at your company? What sorts of things have you learned in this process?
Drop your comments onto this episode's page on AllAboutCircuits.com, or connect with me on LinkedIn, and if you enjoy Moore's Lobby, please leave us a review on Apple, Google, Spotify, or wherever you listen to us. This is how we can build a meaningful community of listeners and contributors. Thanks for listening.
