Rick Almeida Discusses DownStream's Latest News
At the 2018 electronica exhibition in Munich, Rick Almeida, founder of DownStream Technologies, sat down with me to discuss the company’s latest news.
Pete Starkey: Rick, I think almost exactly two years ago we sat down and talked about building the interface between the designer and fabricator. I've seen news recently about your 2018 release including BluePrint-PCB and CAM350/DFMStream. Could you bring me up to speed with those?
Rick Almeida: Sure. So, this is the largest release in DownStream's history that we've worked on. The major theme around the release is bringing 3D technology into the back end of a design chain. It's used widely in mechanical CAD and is being heavily adopted now in ECAD. Our thoughts with 3D were twofold. If the designers were already doing a 3D design, we didn't want them to always have to go to 2D for post-processing, so we wanted to maintain the continuity of 3D throughout the design flow. For us, 3D was also a little bit of a different purpose where instead of modeling the design data in 3D, what we're doing is modeling manufacturing outputs in 3D. A designer can export Gerber and drill files, read them into CAM350 or DFMStream, add dielectric materials, assign thicknesses, and model the board as it would appear coming back from the fabricator.
Many years ago when I was a designer, you would do your designs, go out and get photoplots, and you would always have to imagine what the PCB was going to look like; you'd never see it until it arrived back from the fabricators.
Starkey: And the fabricator had to imagine what the designer was thinking. There just wasn't that interface.
Almeida: Right, and with a 2D system, you're very limited as to how you can look at it. You can look at it top-down or bottom-up, but it's very difficult to discern any type of via technologies that are being used in there. It's harder to see the relationship between the artwork and the drill files, and 3D solves all that. You can now rotate it at any angle and spin it around. We put some specific PCB features in our 3D technology; for example, we can take negative data such as a solder mask or a plane, which is great for design purposes, and automatically convert that into positive data, so it looks like it would appear in the actual printed circuit board.
Another key thing that we do is allow coloring not just by layers and objects, but by layer types. So, all the copper layers can be copper, and the solder mask screen or the silkscreen could be any color so that it appears like the actual PCB. What's nice about it as well is when we're running DFM analysis on it, we can probe the errors with both 2D and 3D, and the 3D view gives you a different perspective of what that DFM error actually would do to the copper clad itself in the copper, so there are some unique advantages in that.
We also bring 3D technology to the PCB stackup; we bring it into the one-up design, and I think we're probably the first and only company that can model the entire stepped-up panel in 3D. Once we have all of these 3D models, we also use them in BluePrint-PCB to create 3D model-based documentation—not necessarily to replace 2D documentation, but to augment the types of drawing data typically delivered by engineering. In a drill or a fabrication drawing, you can add a bare-board model to that drawing as a reference of what you expect the design to look like when it comes back from fabrication, and more importantly, the fabricator can gain a better understanding of what the designer had in mind before they start fabricating.
It's the same thing with assembly. We can add these 3D assemblies into documentation and also bring in step information. If a designer wants to document not just the PCB but also the enclosure that it goes into, that information can come in STEP, and they can use that on their drawings as well. Then, we output it to Adobe using their 3D PDF output, so it becomes interactive documentation.
That's the key part of the release content altogether. We've also made a number of productivity enhancements to CAM350 as far as changing the user interface to bring it up to more current standards and made it a little more dynamic and easier for people to work around the product. We've redesigned over a 100 different dialogues, consolidated a lot of information, and flattened out the user interface that users can be more productive with our products.
Starkey: You have a very large user base. How is this new release been adopted? Is it something that people have been waiting for?
Almeida: What's interesting is it just started shipping in early November, so users are still in the process of downloading and installing it. Our beta customers seem pretty happy with the direction that we were going in, so the feedback is very positive.
Starkey: That's great. So, how many users do you have worldwide?
Almeida: We have about 7,000 users worldwide who use the product.
Starkey: That's pretty impressive.
Almeida: And that's between both CAM350 and Blueprint, but CAM350 has been around for a long time; it's probably the most widely used CAM tool in the market.
Starkey: Forgive my ignorance, but in my day, there was effectively a big brick wall between design and manufacturing. I know you've been progressively knocking bricks out of that wall. Is there a barrier at all now, or is it all one continuous process?
Almeida: I think there still is a barrier there. One of the problems is that when you get into things like DFM, adding DFM into a design process typically means you lengthen the design process, which no one is ever keen to do unless they absolutely must. There's still a reluctance to do the analysis in the design side and pass it off to the fabricator to build because they’re going to build it either way. Things like ODB++ and IPC-2581 are going a long way to take out the bricks. We support both those interfaces because now instead of sending out a dozen different artwork files and drill files, it's one single file that has all the data in it, and it's all connected.
Starkey: It’s good to be able to communicate in an intelligent language instead of a dumb one. I was brought up in an era where we would be given taped artworks in the first instance.
Almeida: That's how I started too.
Starkey: And then master photoplots. Nobody would give us the data that produced those photoplots in case we interfered with it. They said, "Right, that's where you start," which is fine, but that's just dumb picture information.
Almeida: It's ironic because when you think about the design database, every designer looks at it as the holy grail. However, before you can go to manufacturing, you have all of this intelligence and interrelationships in there, but you strip it all out into just a number of disconnected files when you go to manufacturing.
Starkey: When we first started manufacturing from data put in a CAM front-end and a laser plotter, we were on the sharp edge of technology. But we were being fed with dumb information and spent an awful lot of time, particularly when we wanted to do automated optical inspection and electrical test. If you're on quick-turn, short-run stuff, you don’t have an established golden board to compare with; you have to try and put the intelligence back into that dumb information.
Almeida: Right, and the designers take a lot of pride in making sure that design is correct, and then they rely on a third party to put all of this unconnected data back together and deliver their little green card.
Starkey: And this is what you've been achieving.
Almeida: Yes, that's where 3D helped us sort it out. We get a visual model of what the PCB will look like depending on how you design it before you go to manufacturing and before manufacturing engineering. You get a better understanding of what this manufacturing data will look like before you actually start using up copper clad.
Starkey: Even the bare-board PCB fabricator can see what they’re aiming at in 3D and look at it from all sorts of angles. And at the next stage, the person who is going to assemble it can see not just in a schematic but also where those components physically are and how they interact with each other—the tall ones, short ones, etc.
Almeida: It's a realistic view before you have to spend the money to get the design back. It's an interesting paradox that we have because ODB++ and IPC-2581 go a long way to solving that because it keeps all that data together in one file. I think there's a little bit of a trust issue on the design side when adopting these intelligent formats.
Starkey: There has to be.
Almeida: Because of the intellectual property, which is understandable too—IP theft is becoming a big issue that we have to address now. But at the same time, as I said, you disconnect all of this design data, rely on somebody else reading the documentation to put it all back together correctly, and you never know if they did it correctly until arrives it in your shipping department.
Starkey: But if you can step back and effectively see a seamless chunk of the supply chain all talking to each other in a uniform language and accessing meaningful data, that's going to benefit everybody in that chain.
Almeida: And that is one of the other things we've introduced in the CAM350 with this release—it’s a stackup visualizer. The idea of the stackup visualizer is to orientate the stackup for manufacturing and understand the construction intent, and also extract that stackup out of CAM350 and send it to your fabricator to get feedback. What materials are they using? Can they build the board as you're basically specifying it? Then, they can send a revised stackup back to CAM350 and read that back in based on the construction materials they your supplier uses. So, we can try to work with some of the inherent barriers that are still there.
Starkey: Do you get involved at all with the signal integrity people? Because they look on a stackup from the point of view of controlling impedance values.
Almeida: We don't because it tends to be done further up in the design chain than where we are. We look at it from the construction intent of it, and using things like Gerber and drill information, which is not also the best to do signal integrity on.
Starkey: Sounds good. Rick, I really appreciate your time. It’s good to see you again, learn about what is state of the art now, and compare it with what was state of the art then.
Almeida: Yes, it's come a long way. My first boards were done on a light table as well, and God forbid if you stepped away from the light table and had a little piece of tape on your wrist: "Where did this come from?"
Starkey: The big unknown is where it's going to go from here because it’s accelerating exponentially fast.
Almeida: I think the cloud will eventually catch on and cloud-based EDA will let you take the design with you wherever you go—on your tablet or phone, whenever you need it.
Starkey: It’s been great to talk to you, Rick. Thanks again. I’ll catch up with you next time.
Almeida: Thanks, Pete. I look forward to it.