Reading time ( words)
I recently spoke with Max Clark, business unit manager with Mentor, a Siemens Business, and Z-zero founder Bill Hargin about the newly formed partnership that resulted in a new stackup tool that Mentor is now selling worldwide. Fun fact: Hargin used to work for Mentor as part of the HyperLynx team, which now has an interface with Z-planner Enterprise. Talk about coming full circle.
Andy Shaughnessy: Max, can you explain Mentor’s objective here?
Max Clark: Originally, when we started working on what we wanted to do with the next release of ODB++Design, there were two recurring themes that we wanted to try to address: stackup and impedance. Those were the two themes that our customers using ODB++Design were looking for us to convey.
In parallel, I was looking to find a stackup solution that I felt was best-in-class. I wanted to tackle shortcomings that I had seen in just about every other stackup tool. When Bill worked at Mentor, we didn’t really know each other. I found Z-zero by looking at other tools that were out there. I partnered with Bill and explained our vision of stackup and the challenge that was being presented to us repeatedly over the years. We began by simply working together in defining what the stackup content should be inside ODB++Design. Bill was a key partner in developing that part of the structure and content.
It only seemed natural at the time that we began to work together more closely, and that led us to start talking about how we could partner up. The rest is history. We are now going to be representing Bill’s product and using our sales channel for a tool that has significant differentiations between the closest competitors; they are going to be straggling behind once we get this ODB++Design connection to Z-planner Enterprise completed.
Bill Hargin: Max and I were both going toward the same thing: a cool stackup tool that connects the OEM designers to their fabricators bidirectionally and works within a design flow—not just as a standalone impedance/loss calculation tool, but as something that has the overarching PCB design flow in mind. It’s no surprise that we ended up in proximity to one another and working together since we were working toward the same thing. It all came together.
Around 2005, when I was part of the HyperLynx team and head of high-speed marketing within the board systems division at Mentor, I wanted to do a stackup tool, and I was approached by Isola. We began to talk to them about developing a stackup solution within Mentor under the HyperLynx umbrella. But we really didn’t have time to take that any further than some of those initial meetings and discussions because the HyperLynx development team was busy on other things like power integrity, DDRx, and SerDes—all those advances in the high-speed world.
When I left Mentor, I was interested in developing what I had talked to Isola about. I worked for five years for the laminate manufacturer Nan Ya and continued to gestate what became Z-zero and our product, Z-planner Enterprise.
Shaughnessy: How does this work? If someone gets HyperLynx, for instance, is Z-zero an option?
Clark: Z-planner Enterprise will be a separate solution. It will come with the HyperLynx field solver, but it will be a separate product that we sell and market as a stackup solution. The software drives the communication between the designer and the fabricator.
This describes the problem that we’re trying to solve. I’ve been with Valor longer than ODB++Design has been around. The purpose of the format is to tame the confusion that is out there when it comes to what it takes to manufacture a product. You could put as much content as you want in a format, but unless you have the infrastructure or the software to support it at each end, it doesn’t benefit anyone.
Bill developed a solution that sits on both sides. We have the designer using Z-planner Enterprise, as well as a solution for the fabricators. Then, we use ODB++Design to communicate this information back and forth between the two solutions, eliminating any confusion that ultimately can transpire during this transfer.
You take the OEM-design information, and it goes into the fabricators, and then we’re going to use the stackup portion of ODB++Design—which is a standalone file and only a fraction of the ODB++ software container—to send that into a Z-planner solution on the fabricator side. There, they will have not only the library that Bill provides, but they will have their own libraries. Experience just shows us that fabricators run simulations, tweak the numbers, and define a material library based on their manufacturing experience. But in the format itself, the trick was to send that portion of the library information to the designer. When the designer sees the stackup, they see it as simulated and created based on what the fabricator is going to construct.
What you really want to be able to do is send this out, potentially, to multiple fabricators, or to enable your fabricator to send you back more than one viable stackup. Then, you want the software to be able to manipulate that and compare and contrast stackups. Something that was very important in the ODB++Design stackup was the structure stores—not only the original EDA data that maintained through the life of the transfer, but it also contains sections of manufacturers, each supplying viable stackup options.
In Bill’s product, you can see what was originally sent to the fabricators, as well as parade through the viable stackup solutions received from the various fabricators and make informed decisions from that point forward. The idea is if I’m a designer, and I received three stackups from three different fabricators, which of the three do I want to approve and move to volume production with?
Hargin: Let’s say that each of those fabricators is going to come back to you with a different stackup, and the OEM has to manage that process. Twenty years ago, I used to think that there was just one stackup for a design. In fact, there are as many stackups as there are people touching the stackup.
A common scenario is that you’re going to choose all of them because, as the OEM, you’re going to distribute the PCB spend so that you’re diversified. And you have price and geographic competition. You may end up qualifying all three stackups rather than just choosing one. More often than not, they’re not going to be identical.
Clark: You could end up with a situation where you go to three different manufacturers, and if you’re having a signal-integrity problem with one, it could be that that stackup failed for some reason. Even though all the software said everything should have lined up, maybe they have a quality issue in their manufacturing process that skewed the numbers. Here, you’d be able to look at the design again and say, “Empirically, was something wrong here?” as a means to rule that out.
Hargin: The other interesting thing that you learn as you get deeper into this world is the stackup is a contract. It’s an agreement between the OEM and the fabricator. When I’m the OEM and give the fabricator a PCB layout to fabricate a board, I own that. As the OEM, I own the layout. But the stackup is an agreement. It becomes a contract. And if I’m on the OEM side, I need to carefully examine that proposal back from the fabricator and make sure that I agree with it. It’s a unique aspect of PCB fabrication and design.
Clark: There’s a lot of change there, which is why it’s important for the process to be well-defined. When I send an inner layer to be etched, there’s not a whole lot of wiggle room there. I’m not arguing about who makes the copper. But the dielectric material that’s being used with that copper rides on can vary drastically from manufacturer to manufacturer. That’s the hardest part of the whole equation. There’s really no software out there today that manages these issues well. Z-zero is working very hard to fill that void, and our visions aligned very well on how to go about doing that.
Shaughnessy: It sounds like the Z-zero software is kind of optimized with ODB++Design and vice versa. How does that work?
Clark: Bill and I worked on defining the format and definition, and we published that. And contrary to popular belief, ODB++ is free of charge. You can go to odbplusplus.com, register, and download it. There’s no cover charge. And even though we’re working with Bill on this, we welcome others to embrace that ODB++ format, including the stackup. Bill will commit to working with ODB++Design stackups that come from other solutions. It should be transparent to the customer. Bill supports stackup output in too many flavors already. It’s crazy with all of the disparate formats out there. We’re eliminating the need for all of those formats. There’s no ambiguity. It’s crystal clear.
Hargin: I received a customer’s stackup that was a Word document recently, and I had to use Google Translate to figure out what the fields meant so that I knew which column was the thickness and which was the Dk because the numbers were similar. I couldn’t understand any of it unless I translated it first.
Clark: It could even be in an Excel or PDF format. If it’s in another language, you’re dead. Are the units documented? We’ve seen it all and have developed a data structure that enforces best practices.
Shaughnessy: It seems like the one is driving the other. ODB++Design is a big part of Z-zero and vice versa at this point.
Hargin: Z-zero is the horse, and ODB++ is the cart, but that’s just how I view it (laughs). I’m just kidding.
Clark: It’s almost like Z-zero is the car, and ODB++ is the track because Z-zero is running on top of the ODB++ format, so to speak.
Hargin: One other thing, Andy, that we should be clear about that we might have skipped over earlier is when I started Z-zero and our products, I knew I needed a field solver. Because I worked on the HyperLynx team going back to 1995, I went to them, and I came into an OEM relationship with Mentor back in 2016 for me to use to license the HyperLynx field solver. We’ve used the same field solver as HyperLynx, the signal integrity products, from the get-go.
There has always been some level of alignment. I come to the stackup and materials world from both ends: the signal integrity end, based on that history, and the laminate end at the far end of the spectrum. That’s part of where Max and I kind of resonate. It’s true that we’re a small company, but I cover most, if not all, of the playing field, from signal integrity all the way to fabrication. It goes into the software.
Shaughnessy: Every fabricator we talk to says that 90% of the data that comes in for new customers is either wrong or incomplete or both. It sounds like this will help.
Clark: Right now, we’re tripping over the stuff that’s very important but also silly or trivial. In 2020, I still can’t communicate a stackup effectively from design into fabrication and back? Let’s vision ourselves forward, where communication is no longer the challenge. We’re not there yet. Now, you have to ask, “What’s the next step? What’s the cost?” As the train leaves the station, you begin to resolve items, and then you get the next item in series; there’s a lot of back and forth.
The low-hanging fruit right now is to solve this documentation issue that you just mentioned because stackup is part of that. The stackup comes to me in five different flavors; it’s wrong, and there’s an inconsistency between file formats. I have a lot of hope in this area because we can address it very nicely through products in the industry that want to join us in trying to resolve this challenge. It’s a real opportunity.
Shaughnessy: Max, you said that Z-zero would be the first company to support the exchange of stackups using ODB++.
Clark: Bill is doing it on both sides. It will be the first that I know of because I haven’t been addressed by any other software suppliers yet to participate. We just released it, so it’s still new. I’ve found that through the years of working on this, you first have to put the format out, and then you have to create a compelling event. And so what will happen here—and Bill will be the first out the gate, which is a tremendous advantage to us—is he will be able to go out and tout that. Then, there are a lot of competitors out there that are going to say, “What did this company do?” They will come along too. We’re committed, so even if a competitor comes in generating an ODB++Design stackup file, he will take it into Z-planner Enterprise just as though it came from his own tool. All the content will come in, and we’re not going to play a game of favorite software tools.
Shaughnessy: Does Z-zero also work with Gerber and IPC-2581?
Clark: It works with IPC-2581 because Gerber doesn’t convey stackups. But the big difference between the IPC-2581 stackup and ODB++Design is we have this concept of the OEM stackup and all the manufacturer stackups. IPC-2581 doesn’t have that concept; it just has stackups. There’s no differentiation between them.
We tie the stackup to the ODB++ matrix information. In the future, we will be able to view that information side by side, simultaneously, inside our products. Our competitors could do the same. What will be nice about this is once an individual using Z-zero sends their ODB++Design design into manufacturing, the manufacturer will not just get the data to build the board, but they will also get the stackup that is intended to be used. It will all be part of that one data set or digital twin. There’s no ambiguity anymore.
Shaughnessy: That’s good because every time there’s ambiguity, there are three different ways of looking at it. This seems like a really good fit for both sides, and congratulations, Bill. I know you’ve been working on this for a while and spent a lot of money, time, and effort. Is it available right now?
Clark: It’s available now through either of the two companies. Mentor has exclusivity on the Z-planner Enterprise product that will be sold through our channels, but Bill has his own sets of products that he continues to market in other spaces, too, including PCB fabricators.
Shaughnessy: Is there anything else you’d like to mention that we haven’t talked about?
Hargin: There’s this concept in telecommunications and transportation of the last mile. A stackup is like the last mile. It’s the most costly and problematic mile. We’re trying to solve the problem and get that last mile of PCB implementation correct.
Shaughnessy: Thank you both for your time.
Clark: Thanks, Andy.
Hargin: Live long and prosper.