Koh Young's Joel Scutchfield on Smart Factory Connectivity
Joel Scutchfield of Koh Young speaks about the struggles system providers currently face in implementing the machine-to-machine connectivity needed for a smart factory with no common platform or protocol in place yet. Joel also emphasizes the importance of process control and having machine learning and data collection technology powerful enough to make use of the massive amount of data made available through machines like Koh Young's 3D inspection systems.
Barry Matties: Why don't we just start with an overview of Koh Young and what you guys do?
Joel Scutchfield: Sure, Barry. Thank you for the opportunity to talk with you today. As most people know, Koh Young has been the market leader in 3D inspection technology for many years. From the beginning, we've focused on 3D technology, which is a unique approach from everyone else that's playing in the SPI and AOI arenas. Since the beginning, we have used quantitative parametric-based measurement data for true 3D results, which is the mechanism behind our leadership position. It is very different than anything that has been done before and since Koh Young entered the marketplace.
Going forward, we're continuing in our leadership role by using true-3D parametric data to fuel our AI engine, which now provides the ability to create tools that allow manufacturers to realize a real smart factory. By using machine learning concepts and automating programming, for example, our systems will become much more productive, intuitive, and easier to use. Also, data analysis becomes easier and more automated, thereby creating a situation and environment where actionable results can be extracted and used very quickly to make a positive change in real time.
That's where we've come from, where we’re at, and where we're heading. We are in a unique position to be able to do this because we have a specialized approach that uses a parametric, quantitative dataset, which is extracted the right way. Having the relevant data needed to fuel this AI engine, we can now move forward faster on the only real avenue to realize a true smart factory.
Matties: When we talk about the smart factory, you mentioned the information coming from your technology. How does that integrate into the lines, like making adjustments in the manufacturing process in real time?
Scutchfield: There are a couple of pieces to that puzzle. We began this process by taking care of our own backyard first and being mindful of how we could share information with other systems in the future. The first goal was to use the measurement-based data our systems are generating and make our own systems as efficient, productive, and useful as possible. We accomplished this by creating the ability to communicate between our AOI and SPI machines across multiple locations in the line. For example, we can link post-print SPI with pre-reflow and post-reflow AOI machines to quickly identify anomalies in the process and help the customer understand what's going on in real time while providing the data and information needed to correct the issue. This also includes using our dataset and machine learning to create true AOI auto-programming tools. We want to focus on using AI and machine learning to make our Koh Young users as efficient as possible.
The second piece is what we're all striving to reach. By using the various developed M2M protocols like CFX/Hermes, we are working to allow our machine to communicate with everyone in the factory. We have projects on multiple fronts and are moving to the point where we can effectively communicate with all our partners involved with printing, mounting, and more. Additionally, we have undertaken our own projects to create a level of connectivity. The issue we all face though is the fact that there is simply no common protocol at this point. There's no common platform for all system providers to communicate beyond a very basic level. Everything has to be done specifically with that partner, and the results achieved apply to that partner. So, again, what we're striving for here is to be able to communicate on a broad scale in an all-inclusive fashion.
When I say communicate, I'm talking in terms of real machine-to-machine communication. When we find an anomaly or an issue with one of our inspection tools, we can give immediate feedback to the mounter and tell if there's an issue with a feeder or nozzle, for example. Or we can direct information back to the screen printer and provide detail far beyond just an offset adjustment to trending information like volume. In addition, we want to provide information that tells the operator what went wrong, and then expand to tell them the changes needed to address the problem quickly. Eventually, we will do all of this automatically without any human intervention. This is where we get to the point where we really treat the entire line as a single machine.
The goal in my mind behind the whole Industry 4.0 initiative is to reduce manufacturing costs; that's really what this is all about. We want to provide the customer with a solution that truly improves first-pass yield and reduces the need for human intervention. Again, that can be multiple machines or an entire line working in conjunction with all the required software, so the system is the driving factor here.
Everything else is encompassed in achieving those two results. If we can deliver a system to our customers that allows them to lower their manufacturing costs and maximize quality, they will certainly become more profitable, which provides them profits to reinvest and grow their business. In addition, if the efficiencies are evident, legacy products can be converted and upgraded, which helps the whole industry expand. That's the objective here. We're not doing this just for fun; we're trying to leverage the available and planned technology to improve the industry for everyone.
Matties: When you look at the smart factory, there are a lot of benefits that you're describing. The machines adjust as you need to increase yield without human intervention. That's extremely desirable. What sort of obstacles or challenges are the fabricators facing in implementing this into their manufacturing facilities?
Scutchfield: I think it all starts with the ability for us as systems providers to be able to communicate with each other, and we are clearly moving in that direction.
Matties: Is the challenge that the equipment is not universally linkable yet? Do they have to go out and buy a new line to take advantage of it, or is it something that they can retrofit?
Scutchfield: Obviously, the end goal here is that we will have this communication protocol in place so that you can achieve what we're talking about without having to be overly specific in the equipment choices you make. Now, the choices are a whole other subject, but I think you're absolutely right, Barry. At this point, we are still not quite ready with the complete machine-to-machine initiative. Again, it's one thing to be able to access our inspection results and post that in an MES tool created by a pick-and-place manufacturer, but do we really call that full connectivity? It is connectivity, but it's only a part of it; it’s not giving the customer the ability to allow for any automatic or even manual adjustments. Getting the communication protocol established is what's going to drive this forward quickly. That's the limiting factor now.
Everybody's doing their own thing. Everybody's picking their partners to develop some level of connectivity, but those solutions are very specific and only potentially applicable with certain combinations of manufacturers’ equipment. We want to move beyond that as quickly as we can, which is why we are supporting the IPC Hermes and CFX connectivity standards.
Matties: It looks like there's a lot of energy as we see the CFX participation level increasing daily. There are other languages out there but, in the end, it all must talk together.
Scutchfield: That's right, much like the effort to develop the common connectivity protocol years ago. I would equate this to something very similar to that effort, but on a much larger scale. I don't know if we want to call it a race, but there's parallel path development occurring. Aside from that, you have all the initiatives between suppliers like us and other industry leaders for mounters and printers doing the same. We are incrementally moving to the point where we can each talk with one another on a certain level. At some point, though, we're going to be able to communicate broadly once the common protocol is in place.
Matties: Let's talk a little bit about the SPI. I know this is an area that you’ve been in. There are a lot of different approaches to SPI, but I think your latest announcement was talking about capabilities that are pretty far advanced. Can you just talk a little bit about that?
Scutchfield: In general, we've seen a shift in the application of SPI as a tool and technology. We are moving it from simply understanding the data and results to a system that will tell you specifically what needs to be adjusted to eliminate a bad result and how to make that adjustment. Manufacturers do not want to analyze reams of data; they want to quickly identify the issue and then quickly know what is needed to fix it. Better yet, they want to have the system make the adjustment automatically for them.
As we go forward, we want to use AI and machine learning to make those changes automatically, but it all starts with accurate, measurement-based data. Koh Young’s process optimization tools that we're developing hit that mark. It begins with giving the engineer the ability to perform an automated DOE quickly by extracting a sample of printed boards before production starts. This DOE would take place in the new product set-up phase and delivers the optimum printer settings for speed, pressure, and release. Then, once that proper set is in place, it provides a tool to monitor results and trends, and then a trend analysis well beyond the standard X-shift, Y-shift, and rotation anomalies. It goes to the point where we're providing very detailed visual information for where trends are occurring with regard to volume, inefficiencies, and more—what potentially is causing that issue and how you can correct it very quickly. The final piece is utilizing the machine learning and the AI engine to automatically assess the information and make changes as needed without human intervention.
Matties: The more data you capture, the smarter the system becomes.
Scutchfield: No doubt about it. That's where I go back to the whole concept of “true 3D.” The system must accurately capture a statistically relevant dataset of parametric, quantitative information the right way. Just like anything else, the more data you have, the better the result—assuming you have developed the tools for quickly analyze. We hit this mark every time.
Matties: I think we're in the age of data overload, so it's critically important for manufacturers to cherry-pick the data that they need and disregard the rest as noise.
Scutchfield: Exactly. There's so much data out there now, and we know this better than anybody. It’s just the nature of the beast. True 3D measurement creates much more information than pseudo 3D and definitely more than 2D. There are many more important parameters in 3D measurement, and we know both sides of big data. We must use the systems and our AI engine to help our customers extract meaningful information as quickly as possible, and then help them implement the result. Alternatively, we must implement the necessary changes automatically.
Matties: To shift gears a little, your founder and CEO, Dr. Koh, says your entire staff and executives are pursuing the single goal to develop the world's best product and to contribute to the evolution of society and humanity through the challenges and innovation without the fear of failure. That's a pretty lofty statement.
Scutchfield: I think you have to take that approach. The fear of failure is what has the potential to stifle anyone who is on this path. Eliminating that fear from the process really frees you up to do a lot of different things that you otherwise would not be able to do. It certainly helps that Dr. Koh provides the financial stability for innovations to be investigated, evaluated, and pursued. It's a good place to be when the leadership stands behind the pursuit of our goals, and then gives you the bandwidth to achieve them. He continues to raise the bar higher and makes sure we are supported to carry out those goals. Even if we fall short, we're still further ahead than we were before we started. We learn, regroup, and continue to pursue that stretch goal.
This is very much the Koh Young way. The voice of the customer plays a huge role in our development process. We visit Korea several times a year and meet with R&D teams to communicate our customers’ needs face to face. All the while, we are being mindful of the big picture and how our efforts can better mankind. That comes back to what I said earlier; if we can collectively help our customers build a high-quality product at the absolute lowest cost possible, then they win. From there, they can create other good things that help the rest of society win as well.
Matties: You keyed into the one area I did want to end this with. You talked about taking data that you've learned from the marketplace—the trends and such—back to your R&D department. What sort of trends or requests are the manufacturers putting on you right now?
Scutchfield: There's multiple. Obviously, when you are the inspection technology leader, the underlying premise is about how to adapt the technology in other ways. To that end, there are various conversations ongoing about multiple projects that are driven by the voice of the customer. At the forefront is, “How do we better use our data? How do we fuel an AI engine to realize a smart factory?” Many elements have already been completed, but there is so much more we will do going forward.
Beyond that, there are some very specific technologies within PCBA that we have progressed. For example, pin or terminal inspection. There's a real need for inspecting pins on the board and component level. Our automotive customers ask to inspect and measure staked pins and press fit pins. We also receive requests to measure the pins inside connectors for height and perpendicularity to ensure proper mating to ensure the connection won’t be compromised.
We've developed specific technologies for inspecting various pin styles like straight pins, board terminals, forked terminals, and a lot of critical distance measurement capability as well. That's just one example. We have other applications being evaluated and pursued, some of which are not specific to PCB applications. For instance, we are pursuing 3D inspection for machined parts, which requires a dedicated solution—things like metal cases where we inspect for blemishes, scratches, chips, and more. In a nutshell, we use our 3D technology in multiple applications outside of electronics manufacturing to provide solutions that customers, quite frankly, are seeking. They bring these challenges to us, and we've been very successful in providing solutions.
Matties: Joel, is there anything that we haven't talked about today that you feel we should include?
Scutchfield: I guess the one key thing I'll leave you with here is there's a lot of discussion in the marketplace about how to evaluate an inspection tool properly. Whether it be AOI or SPI, the key is you want to understand the company and what drives them. You certainly want to understand the technology under the hood and the system completely. It is important to look at how these three variables align with everything we talked about. Manufacturers need a company, technology, and system that can provide measurement-based, quantitative, parametric data. Everything else related to achieving the goal I talked about earlier begins with highly accurate, reliable data.
You must peel the onion because not all 3D systems are equal. Without a large dataset of reliable parametric information, you're not able to achieve everything we talked about today. Again, it’s the dataset that allows for the development of the AI engine, machine learning, and even industry 4.0 connectivity. As customers are moving forward, they should think about it in those terms. Look under the hood and make sure there is a logical method to the approach and ensure everything aligns to allow they want to achieve their goals.
Matties: Joel, we certainly appreciate you taking time out today for the interview, and we thank you very much.
Scutchfield: Thank you.