I think you would have to live under a rock to have not been exposed to the term “additive manufacturing.” 3D printing, quite honestly, there was somewhat of a hype bubble that has burst, but the good news is the market is transitioning from that hype hangover and really getting down to business and getting this to be a technology that’s going to help companies impact their innovation, impact their overall bottom line and results.
To watch this free webinar, please register here.
There’s a few metrics that are on this particular chart. This is extracted out of some of the pundits that really focus on the industry. I’m not going to read these bullets to you, but the punchline of all of it is, additive manufacturing, also known as 3D printing, is becoming quite omnipresent. I’d like to put some notes on this as it pertains to mechanical engineering and mechatronics design. Some of the things that have changed are, as an example, it used to be print areas were comparatively small. There are now printers that literally are printing things the size of a car, of a home, so when you talk about scale and dimension that’s a change.
It used to be that 3D printing focused somewhat narrowly on the topic of prototyping, but it has diversified dramatically. In fact, the single largest growth segment is, in fact, production, and production support, whether that be end use parts, or whether it be things to go and support manufacturing, jigs, fixtures, job aides, 5S tools and the like; having evolved beyond just polymer, metal printing, and some other variants with very innovative materials.
Then lastly, going back to the theme of this session, which is mechatronics, it used to be that 3D printing literally everything looked like an XYZ cartesian robot, but there are many new formats that are becoming adopted and expanded upon, including delta robot configurations, articulated robots, SCARA robots. For those of you who’ve lived in the world of factory automation, things that you’re used to seeing are actually coming into the world of additive manufacturing in a significant way. Now the takeaway from this slide is, additives growing up and things that we’ve come to expect and see in the manufacturing world are showing up in this domain as well.
What does this mean relative to mechatronics? First of all, I’d like to go ahead and call attention to the bar chart on the right. You go ahead and take a look at this, this is a survey of approximately 800 customers. It was actually generated by Stratasys Direct and they went ahead and surveyed their demographic of customers, small, large customers, engineers, operators, management, and what you see there is the response in a Pareto Format that highlights what are people looking for that would allow some of the constraints of additive to be broken and take things to the next level?
If you look through that list, starting with cost of equipment, and then going down, lowest on the list being enhanced design accuracy, or output accuracy, of the seven items that you see reflected on the right, mechatronics has significant impact on four of the seven. In particular, some of the ones that are higher on the list, and we’ll talk to that. In order to go ahead and make this of substance, as opposed to just being abstract in nature, I have elected to go ahead and utilize a format of reviewing somewhat of a case study of what I will call a real world application involved with a large format 3D printer from a company named 3D Platform.
That image you see in the right hand side of this slide is in fact a 3D printer with an overall work volume of one meter by one meter by a half a meter, or roughly 36 inches by 36 inches by 18 inches. That particular platform incorporates what I will call that cross section of mechatronics functionality, spanning using the icon on the left, the ellipse, mechanical enhancements, electrical enhancements, and control related enhancements that ultimately address the use case of being an effective 3D printer. We’re going to go ahead and work through that ellipse, kind of using it as a visual itinerary, if you will, or a visual agenda for aspects of the machine design that were impacted by mechatronics. We’re going to start mechanical, progressing to electrical, and ultimately control.
Let’s talk about mechanical. When you look at linear motion and cartesian robotics, there are a lot of things that we may take for granted that ultimately show up as impacting print quality. As an example, in order to effectively print something in 3D space, you can think of it as having elements that parallel any three axis contouring application, or a two axis contouring application with a step to go ahead and achieve a change in height. You can imagine a motion control system where the mechanical axis, what is it you’re looking for? You’re looking for axes that are straight, axes that are precise, axes that are orthogonal to one another, or at right angles.
If in fact we have a system that is not, in essence, a good cartesian coordinate system, any of the errors that are associated with rails not being straight in and of themselves, axes not being parallel to one another or perpendicular to one another, result in objects that won’t be printed accurately. The punchline of this: rectangles become parallelograms if you don’t have that straightness, flatness and orthogonality, circles become ellipses, and that’s problematic.
If the printhead, independent of what printhead technology, whether it be fused deposition modeling, whether it be some of the other techniques, such as stereolithography, SLS, if the printhead is not moving in the same plane as the print bed, you will have issues with the first layer essentially not adhering to the bed. There’s a saying in the world of 3D print, “The first layer is the only layer,” meaning that if you don’t get that one right, it’s kind of like foundation, I should say, building a house on a lousy foundation.
If we don’t have squareness, flatness, there’s other problems. It causes binding that results in extra drag that could in its most extreme case cause stalling of motors, and therefore ultimately causing a print failure or at minimum some, what I’ll call, print defects, anomalies, and we’ll talk about that more in just a few slides. How do we go ahead and design the system to go ahead and achieve what I’ve just described in a one meter by one meter by a half area work area without turning it into a nanoscience project? There has to be good design and so we’ll highlight some of those characteristics. Excuse me. Just trying to go ahead and wait for the slide to be advanced here.
Okay, so now let’s talk about linear rail alignment and structure. What’s an ideal system? An ideal system would be lightweight, high stiffness. We don’t want a diving board that’s going to go ahead and result in waves or inaccuracies in our printed part. High precision. When we tell something to be a one inch square, you want it to be one inch, not 1.1 or 1.2. Then ultimately, that linear rail assembly needs to be easy to assemble. Everything that I talk about here, although the use case is mapped against 3D printing or additive manufacturing, in reality it’s relevant to just about any XYZ coordinate motion application. It could include something like dispensing, pick and place robotics, could include something along the lines of non-contact cutting.
We’ll continue to use 3D printing as the backdrop for this dialogue. If I want something to be lightweight, as an engineer, I’m inclined to go towards something like aluminum extrusion. Aluminum is one-third the weight of steel, so I like that. Kinematically, moving less mass allows me to go ahead and accelerate things, decelerate thing, so that’s all good. But I want high stiffness. I can’t have my system wobbling like the Tacoma Narrows Bridge, if any of you remember that from Engineering classes. I’ve got a challenge. Aluminum is light, but it’s one-third the strength of steel. How do we mitigate that?
Contemporary linear guides, there are versions that are available on Steel, but there’s an increasing offering of aluminum extrusion based linear guides, that are available from companies such as PBC Linear. In order to address the issue of yes, I want lightweight, but I want it to be stiff, the ability to go ahead and design different cross-sectional configurations in order to maximize moment of inertia. Just like that golf club where, for those of us who don’t hit the sweet spot, we want that bigger moment of inertia. The reality is by having cross sections designed to the application, we can minimize bow and flex, which is important for our use case.
Let’s keep going with this. High precision. One of the things about aluminium is, if you were to look at typical dimensional tolerances that are achievable with aluminium, the reality is, standard aluminum extrusions, like those that you might be using to go ahead and build a machine safety guard or the like, the reality is, standard aluminum extrusions bend, bow, and warp like a bad two by four, the kind that I buy, anyway, all right? That’s unacceptable on a linear motion application. Companies such as PBC Linear have developed processes, in the case of PBC Linear, a patented process that delivers class H linear rail performance over a 20 foot length while still maintaining some of the cost structure that people are looking for in linear guides.
Next is this idea of easy to assemble. If we’re trying to go ahead and achieve an affordable application system, anything that can be done to reduce part counts, reduce labor, end up having the significant benefit to error proofing system, as well as, quite honestly, making it cost effective and easy to assemble. One of the things we’ll speak to is this idea of reducing components and labor in order to go ahead and simply assembling that square, flat, or diagonal system I spoke to in the prior slide.
Let’s move beyond linear rails and let’s talk about the transmission in our motion control system. Once again, if you look in the upper right hand corner, I’m kind of walking around that visual of the ellipse. We’ve got choices as design engineers. When we look at transmission for linear motion there’s some very proven technology options out there, including ball screw, lead screw, belt drives. There are other options that are available, including linear motor, able drives, but these are some of the more prominent ones and I’m going to go ahead and narrow discussion to these three.
When we talk about choosing optimal transmission, whether it be for your sports car or whether it be for your motion control system, the only way you can have objective dialog is to map it against an application. The application requirements that I’m going to use for this dialog will be an XYZ system that maps into that X axis travel of a meter, Y axis travel of a meter, and the Z axis travel of half a meter. We’re going to concentrate on the X and Y, which is the planer move that is the most prominent move in current state FDM and FFF printing.
If I’m going to have an X axis, and let’s say that that X axis has a gantry, between the mass of the gantry, the mass the cable management, which is a very significant portion of the mass, extruders, extruder filament, let’s just say an order of magnitude, we’re talking about moving a mass that is approximately 40 pounds. Not only are we moving that mass, but we’ve got to go ahead and start and stop. It’s kind of like, if you’ve observed 3D printing, there’s very little highway driving. It’s all start, stop, move left, right. It is, what I will call, short moves, high fidelity. To build on the metaphor of transmission in cars, it is like city driving, Formula One, hairpin curves, start, stop, accel, as opposed to highway driving, where you’ve got a lot of, what I will call, a runway, if you will.
If I’ve got an application that has these type of demands, where am I inclined to go? In this application I still need precision. I’m printing parts with 70 micron layer resolution and to go ahead and have errors that in some cases might even be larger than the layer resolution is just a nonstarter, so I’ve got to be geared, if you will, for this rapid start, stop paradigm that I’ve talked about. Quite honestly I’ve got to be able to achieve my print velocity. It is not uncommon for FDM style printers, that print velocities in the 100 millimeters per second, some are starting to creep into the 200 millimeters per second, some even as high as 300 millimeters per second, but, quite honestly, this is also paced by the printer extruder, what’s melting the filament.
In this case, although it’d be desirable to have print velocities that might be a meter per second, in reality it’s somewhat moot because the gating factor, in this case, is the ability to go ahead and extrude the plastic at the type of rates that I described above. With that prequel, what might a good transmission choice be? Let’s look at some options here. We’ll just wait for the screen to be up here shortly.
Okay, it’s grayed out. I’m going to go through some clicks. I’m, first, going to go ahead and take a look at belts. Belts at face value look pretty attractive. If you look, belts are used very frequently on desktop class printers. Why? Because that belt’s not very expensive in terms of a cost per millimeter. You’ve got potential for very high max speeds. It’s not uncommon with belt style actuators to be able to achieve two to five meters per second linear travels, but the reality is, going back to our earlier discussion, this application doesn’t have enough runway to get up to high velocities. We need to have good city driving. We need to be like that Formula One vehicle.
Although it’s a pro, it’s not applicable in this use case. Some of the cons that we have to go ahead and reconcile is that in reality belt drives are comparatively poor accuracy. Even with very high performance belts from world-class companies, it is not uncommon to go ahead and see backlashes in the 100 micron range as you reverse direction. That 100 microns, once again, when compared to a layer resolution of 70 microns, you can see why it’s problematic. Also, accuracy is problematic, not just with the belt and the belt’s interaction with the pulley.
The reality is unless you spend significant money on high accuracy pulleys, you can get into pulleys, which are not round, so they’re acting like aides or cams, and they may not even center lines, which are concentric, and it further creates an issue where you command one revolution, but across that one revolution there’s going to be high points, low points and therefore velocity errors that show up as print quality errors. Then the final comment I have down below is with belts, especially smaller belts, maintenance is an issue with tensioning and the belt acting like a rubber band.
What other alternatives should we be looking at? The next alternative I want to talk about would be ball screws. Ball screws are unquestionably high precision, they’re also low friction, but unfortunately they’re the highest cost solution when we compare belt, ball screw and lead screw. Max speed can be limited, but in reality it’s not a factor for this style of application. It’s not uncommon to get into 25 millimeter or one inch leads and it’s very easy to go and ahead achieve linear travels that are in that 200, 400 millimeters per second.
There is another comment, which is balls can be noisy, balls require a lubrication over life. These are likely loaded applications, so that tends not to be an issues, but many times these type of cartesian systems are operating in an office environment right next to the bullpen of engineers who are doing prototyping and in reality it can be quite a distraction to have those recirculated balls.
Let’s look at the third option for consideration, which is lead screw and anti-backlash nut. With anti-backlash nut technologies we’re able to achieve very high precision. We’re talking, in terms of repeatabilities, into the low tenths. It’s economical. Something that’s very nice for start and stop is that the lead of the screw, in many ways, acts like a gearbox, it gives you a mechanical advantage and you’re able to go ahead and have very good accel, as well as decel, because the flanks on the screw can actually help to behave as a brake to go ahead and slow the system down.
Although I’ve put the emphasis on XY, there’s a clear benefit for the Z axis. If selected appropriately, from the design standpoint, as long as your lead is roughly one-third of the diameter a lead screw will not back drive, and that’s in contrast to a ball screw, where a Z axis will back drive. That can be an issue, when people are starting to push prints that are in 4 and 500 hour range people want risk mitigation in the event of power loss. Inherently appropriate selection of a Z axis lead can serve as a back drive brake without the expense of supplemental braking.
The same comment about max speed limited. In reality with 25 millimeters, 10 millimeter leads, the linear travel is not a gating factor when looking at ball screw or lead screw. As you go through all these criteria, in the case of 3D platform, the ultimate decision was that for this application the lead screw with anti-backlash nut was in fact the preferred and, in this case, selected choice for the application.
Let’s keep walking through our ellipse. We’ve now moved from motor related, excuse me, from mechanical linear guide transmission. Now we’re going to go ahead and talk about the motor. We’ve got some choices relative to our motor and amplifier pair. In making the choice, what are the acceptance criteria that are going to go ahead and buy us the solution. In this style of 3D printing what matters is high torque for quick accel/decel, positional accuracy, repeatability. There’s a lot of demand and ask for more and more affordable systems. You saw that in one of the early slides. It was the single, most cited comment from customers who either have printers or are looking to buy them.
When you look at affordability, maintainability of systems, anything that can be done to lower complexity around wiring is a big plus. It helps with reliability, it helps with troubleshooting, and then finally it’s desirable. In many cases customers do not have staffs of control engineers who are familiar with the most complex feedback devices, whether it be linear encoders, in extreme cases, laser interferometers, so they need systems that are just going to work with the class of skill trades that they have available to operate them.
That being the case, imagine, I want to frame cost for a second, imagine in the example we’re using the belt price of that meter by meter by half meter 3D printer is $27,000 US, so we’re talking about some pretty complex motion control, we’re talking about some demands in terms of speed and quality. Yet the marketplace is requiring price points that are challenging. With that being the backdrop, what are the options and, in this case, what was the ultimate solution that was adopted?
I’m going to go ahead and talk about classic stepper motor and standalone drives. Pro? They’re low cost. You can get into a NEMA 23 single stack stepper motor and you can do that for under $30. It’s low cost. The control is quite simple. A stepper motor, you tell it how many pulses to rotate. A typical stepper motor will have 200 steps per revolution. You give it 200 steps and it’s going to turn 360°. That pulse and direction paradigm is really quite simple.
The con is this is an open loop configuration, there’s potential for the motor to stall or lose steps. If you’re trying to accelerate very quickly or, let’s say, you’ve got an obstruction, quite honestly the motor may think you’re at point A when you’re really at point B and the result is your print ends up having, what I will call, offsets in it that may in fact cause it to be unacceptable. With all those things being said, stepper motors with drives, whether they be chip level drives or whether they be box amplifiers, as depicted in the image, they are the design strategy of choice for desktop. Okay?
You might imagine at a price point for a pretty complex piece of equipment, that will print something roughly one foot by one foot by a foot, those price points are under $2,000. In some cases they’re down to $500. You can see why, quite honestly, it is the preeminent choice for that use case, but we talked before that this is industrial grade, larger masses, 40 pound gantries, work areas that are meter by meter by half meter, so the open loop stepper has some limits that we have to go beyond.
Let’s talk about the servo, servo motors and drives. In reality, it’s the highest precision option. Inherently servo motors are closed loop, typically with rotary encoders, and the nature of servo motors, very high dynamic performance. I like all those attributes, but there’s no free lunch. It’s the highest cost solution. The complexity and the sophistication of control system, not only when we look at cost, you have to go beyond just looking at what is the procurement of the motor, what is the procurement cost of the amplifier, there is an overhead that gets placed on the control system with the type of I/O for dealing with the, whether it be encoders or resolvers.
What other options are out there? There is a, I’ll call it, a tweener technology, that I will describe as step servo. In essence, as the pictures connotate, imagine integrating the motor, the amplifier and a portion of the closed loop control right into the, what I will call, smarter motor that I’ve got depicted there. The ability to achieve high precision. Imagine, at a cost structure in the low hundreds of dollars, closed loop performance inclusive of the motor, amplifier, 20,000 line encoders, because of the ability to apply more sophistication for managing the magnetics and the commutation, imagine being able to get 85% more torque versus the open loop stepper I depicted in the far left hand corner.
Comparatively, economical and simplified wiring since the motor and the drive are combined. Simple control, because this motor accepts pulse and direction as its input and does all of the closed loop functionality resonate to the smarter motor. In this case, it actually ended up being the optimal solution for this 3D Platform large format industrial printer. Let’s go beyond that. There’s actually … Oops, it looks like I’m going to have to go to a video. This video is about two minutes long.
And our closed loop, upgraded premium SurePrint Servo Technology motors. The differences are in speed, quality, and reliability. Let’s start with speed. What we have here is a demo setup with an open loop stepper side-by-side with a SurePrint Servo. They’re both set up to run the same cycle, so let’s start the test here and see which one finishes first. Right away you can see the SurePrint Servo, with 85% greater torque, allows it to finish 45% faster than the open loop stepper. This results in faster print times.
Here we have another demo with a standard open loop stepper and the SurePrint technology servo. Now what happens when too much torque is applied here to the open loop stepper, I can actually make this lose location and move the shaft of the motor here. You can see the red mark that indicates where it was. I can actually make it lose steps and lose position. With the SurePrint Servo Motor, that has that encoder built it, it’s sending a signal constantly fighting to hold position, so when I grab ahold of this dial, I’m trying to make it lose steps and I cannot move it and cannot make it lose position. The benefit for this in 3D printing is that you’ll have better quality prints, more consistent, because the motor is actually fighting to hold position much more accurately than an open loop system.
John Good: Okay, I hope that video clip helped to go ahead and visually convey this whole metaphor I was using with the difference between a sports car that has, imagine, instead of a 200 horsepower sports car, having a 370 horsepower motor, that’s that 85% torque difference, imagine having a transmission that was geared for start and stop, and then one last thing, maybe some of you have teenagers, God bless you, if they’re starting drive the reality is in the world of 3D printing you want to go ahead and drive like a teenager.
You want to race into four way stop, you want to be able to slam on the brakes, okay, without fishtailing and ended up in a ditch, and then you want to be able to step on the, push the accelerator to the floorboard without spinning on. Basically you want to drive like a teenager, but have anti-lock brakes and traction control, and that’s, in essence, what the step servo technology, with its closed loop algorithms, brings to this application.
The final thing, we talked about affordability, and we’re all challenged to go ahead and lower cost, higher performance in any of the systems that we are out designing or collaborating on. An overall paradigm that is accelerating. Historically we would see design engineers’ procuring multiple components. I use the analogy of it’s kind of like going to NAPA and buying your car one piece at a time and then assembling it. You know what, there are people who have the skills to go out and buy best in class components and then integrate them together, but it takes experience, knowhow, and, in reality, we all tend to be thin on resources.
In some cases some people just don’t have the domain expertise to do it well. Increasingly people are looking toward pre-engineered solutions that achieve the price and performance without the overhead of the build it mentality. Going beyond that, looking at things as, not just components, but as subsystems and, ultimately, systems. One of the other reasons why aluminum excursion based linear guidance is growing at very substantial rates is because there are attributes, in terms of, combining structure, as well as linear guidance, into a single assembly, as opposed to buying components, drilling, and tapping, putting them on a machine surface.
Imagine having pre-engineered alignment, where you get that literally out of the box, but with a structure that has mounting features, whether that be T slots, or whether it be, what I will call, holes, lag bolt, mounting features, and so on and so forth, that make it easier for people to go ahead and, utilizing some simple jigs and fixtures, literally bolt together the erector set, as depicted in the upper right hand picture, that is square, flat, orthogonal.
All of these things come together and ultimately the proof is in the pudding, which is come the end of the day do you end up with a motion control system, optimized around the use case, that achieves the design goals: speed, quality, work area? Then, ultimately, it’s got to hit the market price point that is the threshold for customers to go ahead and vote with their dollars. Hopefully some of what I’ve covered today serves to show that, as hardcore motion control junkies, okay, we can help and we have a role in taking these transformative technologies, in this case featuring additive manufacturing, and making them more accessible to the masses and making them better than are historically and even today.