Enhancing Your Machine Tool with Data

Wade Anderson:

Hey manufacturing world, welcome to another episode of Shop Matters sponsored by Okuma America. I'm your host, Wade Anderson. Joining me in the studio here in Charlotte, North Carolina today, I've got Rob Caron from Caron Engineering and John Joseph from Datanomix. Welcome, guys.

Rob Caron:

Thank you, Wade.

John Joseph:

Thanks for having us.

Rob Caron:

Glad to be here.

Wade Anderson:

Yeah. Take a moment, just give a quick introduction about yourself. Rob, I know a lot of us at Okuma, we know Rob Caron and Caron Engineering really well, but people outside that don't give us a little bit about your background and your company. Okay.

Rob Caron:

Well, I started the company 35 years ago. Caron Engineering's always been a company that's developed unique solutions that are not available normally on machine tools. We develop sensing technology for tool monitoring. We work in the RFID field for automatic tool offset setting from tool presetters. We do automatic gauge feedback from measuring devices. So, all these products are not automatically coming on machine tools. So basically, everything we do allows the machine to run better, more efficient, and allow easier transition into unattended operation. That's the basis of our product line.

Wade Anderson:

Okay. Where are you headquartered out of?

Rob Caron:

We're in Maine, the state of Maine.

Wade Anderson:

Yeah.

Rob Caron:

Up in the Northeast, where it's snowing right now.

Wade Anderson:

Yeah. Always famous for your lobsters.

Rob Caron:

Absolutely.

Wade Anderson:

For open houses and events.

Rob Caron:

Exactly. We do lobster rolls at our booth at IMTS every two years.

Wade Anderson:

There you go. So, anybody going to IMTS, make sure you check them out for a lobster roll.

Rob Caron:

Absolutely.

Wade Anderson:

All right. Excellent. John, tell us a little bit about yourself.

John Joseph:

So, I've got two vouchers for lobster rolls at Rob's booth now that I know he is having lobster. I'm one of the co-founders of Datanomix, currently the CEO of the company. Datanomix is a manufacturing analytics technology. We connect very simply to CNC machines, just about any machine that is of the right vintage and is able to broadcast a protocol called MTConnect. Once we connect to that machine, we are extracting the data off that machine and being able to look at your production cycle time, utilization part production, and we calculate a production score from A+ to C-. That production score is broadcast up on a TV screen for the entire production staff to look at and react to and respond to. Our objective is not to do machine monitoring; our objective is to be a production management framework.

John Joseph:

We are trying to get people focused on jobs that need improvement and getting them moving towards jobs that need to be debugged and brought back on track. So far, things have been very successful, people love our product. I think the one differentiation in our product from the rest of the market is that all of the data that we derive for our broadcasting system for our production score is derived directly from the machine. We don't require any input from operators whatsoever.

Wade Anderson:

Oh, excellent. So, you guys are really good companies that benefit one another, right?

John Joseph:

Yes.

Wade Anderson:

So, you guys work together in conjunction. Tell me a little bit about that tie in. How does Caron Engineering and Datanomix really marry up well and provide benefit to the customers?

Rob Caron:

Well, Caron Engineering's products provide really high-fidelity data, very accurate data, very high-precision, and high-resolution data about the whole process of the machine, whether it be the measurement of tools going into the machine, to the actual measurement of the cutting process in the cutting environment. And then also the part quality as part leaves the machine. So, we've got all this really amazing data that we know has got more value to it than just what we do with it on each machine. We look at Datanomix because their data science capability and analytics now allows them to take our data and do a much more in-depth look and advanced analysis of what's going on. They are looking across an entire plant, so they can summarize all that data together and come up with anomalies and problems and maybe suggestions of better operating conditions based on analytics using machine learning and artificial intelligence.

Wade Anderson:

Okay.

John Joseph:

Yeah. I would say that Caron Engineering was a natural fit for us, natural fit because the journey that we are on is a journey, again, as I said, less about machine monitoring and more about a horizon, a spectrum of monitoring technology that starts with the cutting tool itself and works its way all the way up to business impact. So, with the Caron sensor technology, as Rob said, we're able to intercept a high fidelity data stream, process that data stream, embed the Caron outputs into our product so that our customers can see every instance of Caron within our product, in their production facility, on the TV screens that they've mounted to create this visual factory experience. It's very important that we go all the way from the tool to the machine itself, to the factory production system, and to the business implications of what all that means so that people see everything from the mechanical engineering to the business process.

Wade Anderson:

Okay. Can you step me through? Give me a real-world example of a part, you don't have to actually name a company or part, but real-world example of how a shop owner - the biggest question I wind up getting hit with a lot of times when we talk about data monitoring or data collection, things like that is where's the customer base? Is it large shops? Is it small shops? And then, as they get this data, what do they do with it? How do they see and realize a return on that investment?

John Joseph:

Yeah. Well, I'll start off. I think what's really interesting here is that the people we sell our technology to are people that are experts at subtractive and additive machining, right? So, they've spent their entire careers figuring out how to optimize a process of removing material from raw stock. The Caron Engineering solution helps in that optimization process. What we learned on our journey was that the ability to transform data into usable, actionable, contextualized information, to get people moving towards problem-solving wasn't as apparent as their ability to machine parts well and at high quality and high rates of speed. So, we focus on the data transformation aspect of it, and we rely upon the Caron Engineering subtractive machining process, the cutting process, the tool life, tool management, we take that data and convert it into actionable insights that informs people when they're on track and when they're off track.

Wade Anderson:

Okay.

John Joseph:

So that's the kinds of customers we sell into range from 20-machine shops to 2,000-machine shops. They want to be able to see these cutting tools in operation, in action, real-time. Right? So, the focus is to give them that information as it's happening, not after it's happening.

Wade Anderson:

Okay.

Rob Caron:

Yeah. I don't think the shop size is really the determining factor, more of a shop that's just doing one-off parts may not be as applicable to it, but anybody that's doing any numbers of parts, the more numbers you have for any machine learning type calculations, the easier it is to see anomalies that are happening in a process. I think the shop size is less relevant than that.

Wade Anderson:

Okay. So, if you take a manufacturing process that somebody's been doing for a while, and they have it dialed in, they have it running the way they want to. You have a company X comes in with a new cutting tool, new piece of technology, they want to try that out and realize whether or not this actually does what it says it will do. Is that an area where you guys fit in?

Rob Caron:

Absolutely.

Wade Anderson:

Okay.

Rob Caron:

We're looking at all parts of the data. We're looking at the geometry of tools going into the machine. That's part of our system as well, what were the measurements of the tool going in? Then we're looking at the cutting data as the tool cuts the part. And then we even have our external component, which is in our AutoComp product where we're looking at the measure data of the part, looking at part quality, how much adjustment was required to keep the part in tolerance. So, using those parameters and having Datanomix collect it from a machine learning standpoint, they can actually do an analysis of did this tool cut better? Looking at all those parameters. So, it actually lends itself very easily to, did this different type of tool improve the process? Or is it actually doing a few things better, but there're other things that's doing worse?

Wade Anderson:

Okay.

Rob Caron:

And so, by looking at that entire flow of process, then Datanomix side can do that.

Wade Anderson:

Okay.

John Joseph:

I think there are cost implications to new tools, right. People have to measure their complete cost of goods sold, it's raw material, it's labor, it's machining time, it's preventive maintenance costs, and things like that all wrapped into it. We look at that because we're extracting that data from sensors like the Caron sensors in the machine itself, wrap that up and turn that into a cost implications. And it gives you that spectrum of tool to business impact all in one user interface.

Wade Anderson:

Right. That's an important aspect. I know from the machine tool side, there's trade-offs to everything. Right. I think it's basic economics, always trade-offs to everything that you do. So, you can take a part, you can put a tool in, and I can crank up, and I can get a lot more material removal rates, but then what does that affect down the road? It's really cool to talk about pulling 40 cubes of titanium, but if I'm tearing the machine apart in the process at the end of the year, am I actually making more money by pulling that much load? Versus if I had it back down to a more reasonable rate, that's easier on the spindle bearing life and the ball screws and everything else.

Rob Caron:

Yeah. That's why we also have a machine health component in all of our systems so that we're looking at bearing health, we're looking at the load of a spindle just to rotate all the time. We're feeding that data to Datanomix, and that's the entire goal is to be able to give a customer an actual report of saying, well, when you run this fast, it's going to cost you this much in machine maintenance down the road and the machine may only last this long. So that's really where the ultimate goal is to have that as a reporting tool for a customer. Because those are things that no customer can do today, even with the maintenance department, they just don't have enough data and information to be able to determine those parameters and those factors.

Wade Anderson:

It really feeds into your decision-making process then, right? So, from a maintenance perspective, the learning side of it, you can get predictive on when your maintenance needs to come up, what kind of maintenance needs to happen, things of that nature.

John Joseph:

That's really important because if we're looking at the current situation with service, response time in an industry that's really growing rapidly at the moment. And if you don't have predictive technology, like the technology that Caron produces, you are waiting for something to fail in order to respond to it. And that response time could be weeks.

Wade Anderson:

Right.

John Joseph:

Multiply three to four weeks of response time times the shop rate, that's tremendous lost revenue, lost capacity, customers that are disappointed because they're not getting parts. It doesn't have to be that way is the point. The point is that you can be predictive, you can determine when a tool needs to get maintenanced, and you can schedule that maintenance in a proactive way. So, I think the combined technologies of the two companies allows people to have forethought instead of being responsive and reactionary to everything that's happening in a firefighting mode, which is very difficult for people and long term it's exhausting.

John Joseph:

You're trying to grow top line at your company while you're trying to fight fires inside the company. It's really hard for these guys. And to give them the tools that they need to look ahead, right? So, look ahead, look up, look at TV monitors that are giving you information about the health and status of your production floor in real-time, it's a massive benefit for folks. What I find interesting is the profile of some of these owner/operators, the executive teams of these companies that are utilizing this technology are forward thinkers, right? They invest in technology, they see the benefits of it. They can measure the benefits of it. I think what's really important is purchasing technology, using it, and if you can't measure the benefits of it, you really don't know what the impact of it is on production, on output of the factory.

John Joseph:

These people all have a common thread around thinking ahead, thinking about how to drive efficiencies, leverage digital technology to make them better. And they're always thinking about innovating on that dimension every single day.

Wade Anderson:

To me, it's a paradigm shift in how you think about your manufacturing process and how you look at machine downtime and things of that nature. Are you seeing that mental shift in shop owners? Is that growing at an exponential curve as people are talking?

Rob Caron:

I don't think it's exponential, but it's definitely growing. I think the other piece that's coming to play, too, is that obviously, everybody's trying to automate now robots. Robots are going on machines. People are trying to automate every part of the process. So, the eyes and ears that used to maybe hear something that was wrong before aren't even there anymore. So, I think that's even driving it harder to having some sort of automated way of measuring the machine health and just general capabilities that are going on and problems.

Wade Anderson:

That's an excellent point. Automation, one of our themes for our IMTS this year's automation. Virtually almost every machine we bring is going to have some form of automation, whether it's software-related or physical unattended robot loading, things of that nature. That's been a drastic shift, especially since the pandemic. Shops are realizing, finding skilled people is harder and harder to do. If you can automate it, that's great, but you lose some of that eyes on target when you automate.

Rob Caron:

Yeah. Something as simple as the coolant flow on a machine, the coolant pump gets clogged. Well, a human walking by is going to be able to see that there's hardly any coolant coming out of the nozzle, right? But when you put a robot on the machine, and there's nobody there anymore, that problem still happens, but you need sensing technology then to be able to tell you there's a problem.

Wade Anderson:

Yep. So, what about, I guess, operator mistakes or process mistakes? I believe some of the technologies that you offer, Rob - I'm going to back up to when I was running machines, and it's been a long time ago now. I haven't run machines in many, many years, but when I worked this prior to my life at Okuma, I worked for a grinder manufacturer, and we did a project with a medical company and we were doing, it was a cobalt chrome material. We had four plated wheels, it was basically a double spindle grinder. We had four plated wheels. We were making four parts at a time. I don't want to talk too much and out the company, but we were doing the runoff at our facility. And being as it was a runoff, what could possibly go wrong? Those are my famous last words.

Wade Anderson:

So, we didn't have the fire suppression system hooked up. We were grinding with oil, it was a very lightweight oil. It was practically like trying to grind with kerosene. Looking back, there was a comedy of errors stacking up. This part had a hardened steel dowel pin, basically, that was a center locator for the part. So, as it's grinding on one side of the part, you get to a point where the wheel just lifts up. It does a very small little circular interpolation move over the top of that. And as we were making adjustments, somewhere along the way, that radial move got flipped and it went down instead of up.

Rob Caron:

Oh gosh.

Wade Anderson:

So, I've got the spindle backed out. Of course, doors are shut and you're seeing everything from a distance, but I start half-inch away from the part, everything looked good, brought in a quarter-inch a part, 100,000 off the part, everything looked great. Finally, I came down, sparked off on the part, turned cooling on, let it rip. It's rolling along, and this is about a 100-horsepower spindle with these nickel-plated wheels. Me and the operator from the company, we’re standing near the door and about that time, it hit that little move and those plated wheels started digging into those hardened steel dowel pins, and that thing went up like a Roman rocket ship.

Wade Anderson:

I had flames that shot about 40 feet through the air, spread across the ceiling of the building, everything that was plastic inside the machine got melted, all my way covers, all the air lines. Didn't do any physical damage to the machine, scared me half to death though. It took me a while to get the nerve back up once we got the machine fixed. But that spike in horsepower and all that, your technology would've prevented my mistake from being able to do that, correct?

Rob Caron:

Yeah. I'm assuming that you got employee of the month that month?

Wade Anderson:

Yeah. The CEO was not very happy with me.

Rob Caron:

Yeah. So, our technology absolutely would've prevented that. We're looking at the power of driven tools, grinding wheels, whatever it is all the time. We're looking for abnormal increases in those power levels. We also look at vibration, potentially strain of certain members of the machine. We can react to those very quickly. We can react to them automatically and absolutely prevent things like that from happening. We also can adaptively control the feed rate of cutting tools or grinding wheels so that we're optimizing the power that it's cutting at. If something happens like that, then it's automatically going to reduce the feed rate and even stop the machine if it deems that necessary.

Wade Anderson:

Okay.

John Joseph:

Yeah. I think you're also pointing at a bigger issue. And if you take a look at the employees that are joining these companies today, they're very different from the employees that joined these companies 30, 40 years ago. They don't have the fine machining skills, in some cases, they're not hiring people with fine machining skills, deep experience in machining technology, they're hiring operators, right? Operators to run machines that are computer controlled and automated. Automation is the big thing, and the reason for automation is because the workforce has changed, right? And so, with this changing workforce, you've got to deploy technology that supplements them, that augments them. And the beauty of our products is that we produce a data stream, or we're reading a data stream, and when you start to read a data stream, you look for trends and patterns.

John Joseph:

You look for the frequency of alarms that get triggered. You look at those alarms now in the context of a production shift and what time the preponderance of alarms occurs. Is there a pattern there? And so, you look at these patterns and you say, "Geez, that's a problem." Or an alarm triggers because there's a problem with the machine and the only way you sense the problem with the machine 20 years ago was to listen for it. But today, those skills are not there. So how are you listening for something that you would've heard 20 years ago? But today, you pick up out of a data sensor located within the machine that tells you something's wrong. And so, the software that we're building together needs to be so sophisticated that you're looking at things coming through the window, the real-time window of time, and sense that there's either something going very right or something going very wrong.

John Joseph:

And so, we need to be smart enough as software people to predict that data trend that we're seeing in the moment is an indicator of either a great thing or a not-so-great thing. And that's the power of what the two companies are building together.

Wade Anderson:

I may be going down a different rabbit hole on this question, but how do you alert somebody? So, when you're sensing, "Hey, wait a minute. Things are not on a good path here." How do you make it to the point so that a guy like me could look at it and go, "Ah, I'm sure it's okay," and turn around and walk off versus saying, "No, wait a minute. This is something you really need to pay attention to."

John Joseph:

Yeah. In our case, I think in Caron's case, there's adaptive control, so they're feeding back to the control system to affect change. We're not feeding back to the control system yet. It might be something we do together in the future, but we're not doing that today. There are two ways to notify people that something's headed in the wrong direction. One is visually, and I think visually is the fastest way to do it, is to visually give people cues that it's headed in the wrong direction. That's the main focus of our product, when we mount 65-inch TVs on a production floor, and you see a flashing indicator that an Okuma machine has had a problem and someone needs to attend to it. That's important, that's visually important, and a person's going to respond onto that.

John Joseph:

The second methodology is audible, right? There might be an audible indicator that something's wrong. We don't do that today, but it could be something we do in the future because I see control systems getting more and more sophisticated around facial recognition and adding audio to control systems and things like that. The third way is to notify people using text, email, phone alerts because everyone's carrying a smartphone on them on the production floor, that's the next wave of notification for people who need to know. Just contrast that to, I'm saying this tongue in cheek, but some of the old-timers that we interviewed in building our company said, "I walk around the floor trying to smell smoke, look for fire, or feel vibration in my feet. That's the way I figured out what was broken on the production floor." Well, those days are gone.

Wade Anderson:

Right.

John Joseph:

Those days are gone. When you have a 100 machines, you're asking one operator to run two to three machines in a shift. And there's no way that he can sit at one machine and watch it because the profit's gone, if you're dedicating a one-to-one relationship on an operator for a specific machine and a specific part. There are parts that are hard enough and sophisticated enough that warrant a one-to-one relationship, but the owners I talk to say, "I need people running three machines at a time." If the cycle times are right, if the parts are right, three is a great number, I think the average is about 1.5. But you now need to use technology as a lever to help you apply your span of attention to a broader set of problems in real-time.

Wade Anderson:

Okay.

John Joseph:

If that makes any sense to you.

Wade Anderson:

Yeah. Absolutely.

Rob Caron:

I also want to clear up. TMAC and our on-machine products are going to react instantly to anything that's happening on the machine. The problem that you had, any type of changing, cutting conditions, we're going to react immediately. I think where Datanomix is going to come into play is they're going to look at that data and see things that aren't immediate. In other words, they're not instantaneous problems. They're developing problems. That's why we're not seeing them because we're looking at everything in real-time. But they're going to look at the results of all of that data and predict that something is trending to failure, which gives somebody a better chance, a more advanced notice to maybe try to look into the situation and see what can be corrected there. Or we potentially make suggestions, or they make suggestions of what could be corrected.

Wade Anderson:

So, if I could try to package up everything that I'm hearing you guys talk about, I think of it in terms of a high-performance organization. How do you take a shop that has been successful, what they've done, got them where they're at, they've been successful doing what they're doing, but how do you raise them up to another level? And to do that, everything's got to be functioning at a high-performance level and controlled, and everything moving in a controlled atmosphere. That's basically what you guys combine, your two technologies are doing. It's giving that tool to shop owners, who's been very successful, it's got their company to where they're at, but gives them the leverage to move that company, elevate it to another level.

John Joseph:

Absolutely.

Rob Caron:

Yeah. Actually, one of the things that even these top shops can't really correct is they have vendors that are providing tooling and material and coolant and all types of different products that have some level of inconsistency. So, the guys that are the best out there still have to deal with that every day. And things that are caused by the material being a little bit different today. That's where I think even the best shops, efficient shops that are out there, these are some things that we can really help by looking at the real-time data and make corrections with these standard anomalies and changes that are coming into them every day.

Wade Anderson:

Okay. Excellent.

John Joseph:

The technology tests the assumptions. I think you said it-

Wade Anderson:

Say that again.

John Joseph:

The technology tests the assumptions.

Wade Anderson:

That's great. I like that.

John Joseph:

The assumption is that we'll run production today, and tomorrow we'll measure the output of production. We'll either gauge block it, we'll gauge measure it, we'll apply metrology to it to figure out if what we made yesterday was the right quality, met customer requirements. I talked to a customer yesterday in South Carolina that told me that his milling machine was booked out for 30 weeks. That's based on a set of assumptions. The set of assumptions are that machine is going to produce a quality component, a precision component, for the next 30 weeks and meet the forecast customer demand.

Wade Anderson:

Right.

John Joseph:

If anything happens to that process, the 30 weeks is out the window.

Wade Anderson:

Right. Yep.

John Joseph:

So how do you apply technology to a methodology of machining parts to guarantee accuracy, to guarantee that you're going to hit the target? When you pull your parachute, you're going to hit target on the football field, in the middle of the Notre Dame game, for example, right?

Wade Anderson:

Right.

John Joseph:

You've got to hit that target every single day. The assumptions that the owners that we meet with have is that the capacity of their factory is fully utilized, to the maximum ability possible. When they see our product come in and we test their assumptions on cycle time. We test their assumptions on utilization. We test their assumptions on part production. We see over the course of 90 to 120 days, their assumptions were wrong.

Wade Anderson:

Right.

John Joseph:

That the utilization wasn't where they thought it was, it was lower, in some cases higher. That the cycle time wasn't what they quoted, it was something else, and that the quote is wrong. We are applying technology to make the process of estimating what production is going to produce as accurate as we possibly can. Because these folks are not interested in making parts at a loss.

Wade Anderson:

Right.

John Joseph:

They're interested in making parts at a profit. How do we ensure their profit? How does a Caron Engineering TMAC solution ensure that this part's going to meet conformance, quality, field testing, life cycle, et cetera, at a profit for that company? That's important. Rob comes to work every day at his company because he's trying to guarantee that they can deliver that consistently and repeatedly. We come to work every day, trying to make sure that the impact of what Rob's doing ultimately works in the machine and works to give their business the profit objectives that they're trying to reach. It's really important, really important. These guys are all trying to grow top line. They're trying to grow it at a profit. That's the point, right? Get your business growing at a profit, increase profit, increase customer satisfaction. That's what we're here to do every day.

Wade Anderson:

So how do you implement? So, if I'm a shop owner, I've got ten machines and five operators and making production, and I want to implement this technology, what are the steps? What do I have to do? What kind of skillset do I need to have on staff to be able to implement this?

Rob Caron:

Well, from our technology side, basically, the systems need to be installed on each machine. The best shops that implement our technology usually take someone and they become the expert in the technology. Then they get it all up and running and implemented, now the data's available immediately. It's reacting to whatever's going on in the machine. It's helping improve cutting and tool life and all the different parameters that we're doing. Automate the process of bringing tools into the machine, measuring parts outside of the machine, and correcting automatically. That all needs to get implemented. With our help, the customer implements it and brings it online. From there, the data is immediately available to Datanomix, and then they can take over from the analytic side.

John Joseph:

Yeah. We plug into the controller, and we're the wingman to the data stream that's being produced by their product. We're picking up on the draft of that data stream. We're picking up that data, processing that data instantly, and providing them a real-time production score for the directional health of that job. It's very different than what they've done in the past, as I said, tomorrow they'll measure to see if yesterday they produced the right output. You don't want to be solving yesterday's problems tomorrow. You want to be solving today's problems now. Closed-loop control and real-time production scoring, that's the way to go after it.

Wade Anderson:

Excellent. Well, guys, this has been a fantastic conversation. I really appreciate your time and joining us here today.

John Joseph:

Thanks, Wade.

Rob Caron:

Thank you, Wade.

Wade Anderson:

So, if the guys, or... I said guys, I should say people in the field, listening, they want to learn more about your products, Rob, how do they find you?

Rob Caron:

We're on the internet, on the web at www.caroneng.com.

Wade Anderson:

Okay.

John Joseph:

We're at www.datanomix.io.

Wade Anderson:

Perfect. Guys, thank you so much today. Thank you all for joining us. If you have any thoughts, questions, ideas for future podcasts you'd like to hear, please reach out to us and let us know. Otherwise, till next time, we'll see you then.