How to Use Panelization Planning to Save Money and Resources
In this interview, Siemens’ Patrick McGoff speaks with Nolan Johnson about strategies to utilize panelization planning to save money and resources.
Nolan Johnson: Patrick, please tell me about your role at Siemens.
Patrick McGoff: I am the market development manager for the Valor division within Siemens. I have the responsibility to market and do sales enablement for the worldwide account teams for the promotion of the Valor DFM products.
Johnson: You co-authored a white paper titled Lowering PCB Costs with Material Utilization. Since we’re going to be talking about how a design team can influence the panel for being more efficiency in cost savings, would you start by defining what a panel is?
McGoff: There are two types of panels used in the PCB industry. Sometimes people gravitate first to a fabrication panel because that’s the first thing they see for a product. But there’s actually two panels—the fabrication panel, which is used to manufacture the bare board and then there’s the assembly panel. The assembly panels are the units that go through the assembly lines, the SMT lines to populate the PCBs. I make a distinction in the paper to differentiate between the two so that as we’re discussing panels, people don’t misunderstand which of the panels I’m speaking about in that part of the paper.
Johnson: The assembly panel is a subset of the fabrication panel, in other words.
McGoff: That’s correct. The assembly panel is what the design organization or OEM expects back from the fabricator as a deliverable.
Johnson: The finished good from the fabricator.
Johnson: Then that assembly panel moves down onto assembly. When it comes to optimization, then, designers have both panels to optimize.
McGoff: That’s correct. It’s important to note that it’s a two-step process to truly optimize your product for material utilization. The OEM is responsible for its product costs. And we all tend to kick the can down the street if somebody else is willing to pick it up. We’re guilty of that in all aspects of the electronics manufacturing flow. Panelization is one of those topics. Every fabricator out there does an excellent job of panelizing the design that they receive for their purposes. But their view of panels is different than what the OEM’s view is, especially as you get into the assembly panel optimization. You can punt to the fabricators, they’ll panelize it, and they’ll quote based upon what you’ve provided them. However, they aren’t able to break down what you’re asking from them, the assembly panel configuration. They’re stuck with, or handcuffed with, a rectangular box that they’ve called the assembly panel, and it behooves the OEM, the design organization, to look at optimizing that assembly panel first, and then see how it fits also in your fabricator’s panel sheet sizes.
Johnson: Why should a design team care about the panel?
McGoff: Because we’ve proven, and it’s documented in the paper, how there are substantial material cost savings if OEMs can have that view early in the process and understand the ramifications of what they’re sending to the fabricators.
Johnson: Okay. So first optimize for assembly panel then see how you can optimize that structure for fabrication. Does this process iterate during the prep for manufacture?
McGoff: Well, it certainly does. The goal is to stop it from iterating. The back and forth, we want to eliminate the “back and.” We just want them to do the “forth.” And just to be clear, when we talk about optimizing for assembly, I want to make sure that we’re clear on talking about optimizing the material utilization for the assembly panel, so that people don’t mistake my comment for optimizing the pick and place time itself.
Johnson: A key distinction to make. Where and how do the savings start? Walk us through the implications of paying attention to this. I happen to have a board fabrication background, and there were plenty of times where a little bit of work from the designer could have made a huge difference.
McGoff: Very much so. In the paper I have an illustration of a U-shaped rigid PCB. And it’s got an assembly panel presented in each block of two. That assembly panel has been stepped and repeated into the fabrication panel to see how it fits. Well, if an OEM sent that over to the fabricator and the fabricator is going to say they’ve got great material utilization on an 18 by 24. It’s going to show as I state in the paper, 58.9% utilization. You’re wanting to use as much material of a panel as you can, because you’re paying for the material costs there.
From the fabricator’s perspective they’re doing pretty good, but when you look at it in terms of the actual total surface area of that panel, the OEM is only using 18.7% of the material. They’re paying for 100% of the panel, but only using 18.7%. And if you look at the next illustration, it shows how you can use the optimization utilities to nest that in an efficient manner. And not just to get a smaller footprint for a two-up array, but also allows for orientation changes as well, where you can have them all oriented the same or allow rotation as you see in this example, 90 degrees. They get even more material utilization on the same panel sheet size.
Johnson: That starts to make sense; utilization does depend upon who’s doing the counting.
McGoff: A designer can punt it to the fabricator. Like I said, they’ll take that rectangular box you give them, and step and repeat it as efficiently as they can, so they can do a competitive quote. But after that point, they can’t touch it. They can’t break up that assembly panel. Now you’re stuck with the cost. One of the sayings that I have is that designers think they’re buying boards, but actually, they’re paying for panels. Does that make sense?
Johnson: You’re buying real estate, whether you use it or not.
McGoff: Exactly. And you mentioned a moment ago about how this can have potential cost savings. You can look at it in several different ways. A lot of people think I don’t need to optimize because I’m just building small lot sizes, and so it doesn’t affect me. It’s absolutely true that the higher volume you build, the more impact this is going to have for you, the more savings it’s going to have. I’ll even say that we have some of our automotive and mobile phone customers that do this panel optimization before they do layout. They’re wanting to see when they go to high volume production whether this is going to give them the maximum material utilization possible.
Johnson: Is the outline coming from mechanical dimensions to which they’re laying out the panel?
McGoff: Exactly. Think about automotive. You’re building a million boards a year. If they can look at a tab, for example, and then ask, “Can I shave this by a quarter inch or 100 mil? Is the dimension of that tab critical? If I can rearrange something else, and I can therefore fit 20 up on this panel instead of 16, then it’s going to be worth its weight in gold.” We have customers doing high volume, and clearly this is a no brainer to them. They’re the ones that gravitate to it first.
What I also try to illustrate through the examples in case studies we’ve done in the paper is show you that even in as few as a quantity of four panels it can produce a cost savings. We looked at this and the different PCBs from our customers and said, “Here’s the volume that they’re building in, and the cost per PCB, using a common number for real estate here. Even in as few as four panels you can save by having this material optimization.” It doesn’t have to be a high-volume application only. It can apply to even those with small and medium volume production runs.
Johnson: Is there a point in the design cycle going from design to prototype to prep for production, where it makes sense to really start to pay attention to the optimization?
McGoff: I think that would be in a direct proportion to the volumes you’re going to build. Like I said, these high-volume people will do it even before layout, and you can have anything built in prototype. However, even prototype builds must be cognizant of how it’s going to be built. I touch on a related subject in the paper about doing DFM on the panel configuration. Even with a prototype, you could have two boards back-to-back and the connectors are butting each other, and it would not be a feasible construction. So even in prototype you want to look at the panelization.
Even at prototype, you might be having 20–25 boards built, and that could represent a few panels. So even then there is an opportunity for cost savings.
Johnson: So many variables are in play here. Is there even a possibility to quantify this? What kind of savings are we talking about?
McGoff: I can show you on a spreadsheet (Figure 3)
Johnson: Thank you.
McGoff: We took four different customers’ designs, and we looked at the volume that they’re building in and then spread this all out. We extracted it for each of them. We looked at the panel sizes that had been used. We used a price of 28 cents per square inch for the fabrication material costs. We looked at how many were going to fit in that panel. The number of panels needed. And one of the things with your fabrication experience, you can hopefully appreciate this comment, fabricators build overage into virtually every job that they take on. The amount of overage it depends on the difficulty the technology, and the urgency of the delivery.
When you look at the overage, how many more are there actually in the build? Of course, the fabricators factor that into the quote that they’re giving the OEM. They know that if they need to deliver a certain number of assembly panels, within a given amount of time, they do their assessment of the difficulty and challenges. And then they say, “We’re going to need X amount of overage for this.” We reflected that and said, “Based on that, here’s how many real panels they’re going to put into production.” Their fabrication costs are going to be defined, costs per board. Then, the non-optimized costs show that as current state. The optimized costs are after we ran it through our software. It shows a lower cost per board. Instead of $10 and 89 cents, a board is $6 and 36 cents, a 41.6% savings for that build and a $453 savings on the job. Does that make sense so far?
Johnson: It does. And all of that of course falls through to the bottom line for the project.
McGoff: Exactly. And we can talk about revision spins, and how many times you have to go through proto or small volume ramp up before you get to volume. Each step makes a material difference in cost savings. We did this for each of the different technologies, extrapolated the volume, and then we came up with a summary tab here. First section is for the first PCB, the summary data. And the graph here is cost per PCB. The obvious is that the higher volume you get, the lower cost your PCB is going to be, because you don’t need as high percentage of overage to meet deliveries and so forth.
What some people are surprised to see is that it’s not a linear or smooth curve, but the variations depend upon the actual configuration of the panel, and at some volume points you’re building extra boards just because you’re using a panel and you don’t need all of boards on that last panel. So, that’s reflected in the numbers here. Regarding the cost per PCB, we have a savings percentage that we’re showing, a total fabrication cost savings, and then the actual number of fabrication panels needed.
That’s our data behind our savings. This is the summary that we gave in the article. You see the potential here, and you can see the savings can be substantial. If you’re doing 100 boards per quarter, this says you’re going to save roughly $42,000 per year. The next one says you’re going to save close to $300,000 per year. It gets quite substantial, and that’s not even the high-volume runners like automotive and mobile phones.
Johnson: Right. You’re talking here about 1,500 a quarter in this example number two.
McGoff: Right, less than 5,000 a year. Appliances could be an easy one there. One other thing I want to emphasize is we ask the OEMs, “Would you change your order quantities and intervals if you knew this information?” If you take somebody that only needs 1,000 boards per quarter and you looked at the difference on it at a thousand, it’s $4 and 70 cents, but at 600 it’s $4 and 66 cents. That’s not a big savings, but you get the concept. Would you change your order quantity and intervals if you could have a lower cost per PCB?
Johnson: What seems to be the customer response to that idea?
McGoff: It gets them thinking. These are typically the engineers, not procurement. But when they understand that the assembly panel affects the utilization on the fabrication panel, then they do start putting focus on it. It’s worth them spending some time doing the effort during the layout process, as well as making a pitch for the software with the right tools. The panelization is a fast process; in 15–20 minutes, you can have an optimized data file for the assembly panel that’s going to meet your objectives.
Johnson: At one point I was thinking that this is just an extra step in the design flow for a design team; yet another thing that they need to do that traditionally they haven’t done. Now they can take this optimization step on and be much more effective and much more efficient overall, which is great for their margin in their profitability much later. But design engineers often don’t think that far down the line. You’re saying that this is something that now becomes just a few minutes?
McGoff: Yes. Actually, if you think about it, the designers are already doing some form of panelization. When they send their dataset, their deliverable package to the fabricators, it’s not just the data for the design, but it’s also drawings. And they send a drawing file of how they want that assembly panel to look. Now, it’s usually just that, a drawing file, a PDF, for example. Then the fabricator takes that PDF, recreates that assembly panel in their own CAM software, sends it back via email to the OEM, and asks if they got it right. The designer says yea or nay, and then they’re off to the races. The designer is using a mechanical CAD package, like AutoCAD, to create this assembly panel. Now it’s not optimized. It’s just basic 2D drawing with annotation and dimensions, but they’re going through an effort outside of their layout tool, and outside of their DFM tool to create this assembly panel drawing today.
What we’re saying is that in less time you can do that because you also have to put in the other features, which I believe I mentioned in the paper. You have to put in other features, like where do you want to break tabs? And are you going to have tooling holes or fiducials that are necessary for this assembly process? The OEM is responsible for defining those features in the drawing, and by doing it in the same flow as your DFM tool and taking, in essence, the same amount of time that it would have taken to put it into drawing on AutoCAD or another mechanical CAD system, you can do it now and get an optimized result with no manufacturability issues related to DFM at the same time.
Johnson: What’s in it for the fabricators and the assembly houses when their customers use this tool?
McGoff: For this assembly house, there’s not much in it for them. We’re not touching their process. We’re affecting the material that’s used on the fabrication panel. It happens as a next step to be assembled, but we’re not really affecting the assembly side. The fabricators have value-add in terms of their expertise of how to process and manufacturer a PCB or panel. The tasks of recreating a drawing are non-value added. They’re inefficient, and they would all rather be given golden data to begin with than go back and forth using humans to interpret a drawing that oftentimes has misleading dimensions or notes with conflicting data because it’s a boilerplate that was sent. Instead of that inefficiency, they prefer to say, great, you gave me good content to begin with. I can take it from here and do my value-added engineering to prepare it for manufacturing.
McGoff: You never want to speak for somebody else, but over time if a fabricator sees that a given customer has given them efficient data, it’s possible that they could have lower NRE costs in future quotes.
Johnson: Care to tease the conclusions from the paper, Patrick?
McGoff: It’s the idea that the OEMs are responsible for the cost of their products, and everybody is looking at how to do that. I also make the point that we’ve all heard the phrase of shift-left, right? Moving things earlier in the process, the challenge is what’s a reasonable burden to shift left. We’re not asking the designers to do the fabrication panel layout, with the venting, thieving, coupons, and such, the nomenclature they need. What we’re saying is, look, your fabricators work with a fixed number of panel sizes especially here in the U.S., you know what you’re working with here. You’ve got the borders, all that’s well-defined, you can have those in your libraries, and it checks all of them all at one time. We’re not asking the designers to do the tooling—that is the fabrication panel generation. We’re asking them to be more efficient in what they’re already doing by having a tool that facilitates that efficiency.
Johnson: I think that’s a key distinction for the casual observer. This may look like it’s yet another step; It really is not.
McGoff: Right. We’re keeping it in the same flow. And now the assembly panel is delivered as a data file, not as a drawing. So, the fabricator doesn’t have to recreate that in his CAM system with his panelization software, he merely imports it, and then begins tooling.
Johnson: Patrick, thank you.