At this point, the internet is crawling with butt-kicking homebrew 3D printers made with extruded profiles, but it’s easy to underestimate the difficulty in getting there. Sure, most vendors sell a suite of interlocking connectors, but how well do these structural framing systems actually fare when put to the task of handling a build with sub-millimeter tolerances?

I’ve been playing around with these parts for about two years. What I’ve found is that, yes, precise and accurate results are possible. Nevertheless, those results came to me after I failed and–dry, rinse, repeat–failed again! Only after I understood the limits of both the materials and assembly processes was I able to deliver square, dimensionally accurate gantries that could carry a laser beam around a half-square-meter workbed. That said, I wrote a quick guide to taming these beasts. Who are they? What flavors do they come in? How do we achieve those precision results? Dear reader, read on.

But First a Brief History Lesson

From college optics labs to factory floors, it’s too easy to let these extrusions fall into the cast of minor characters that were simply “always there.” After years of taking them for granted, I started to wonder: where did they come from?

Our story starts back in 1797, where [Joseph Bramah] coined the concept of squeezing a pre-heated hunk of metal through a precisely-shaped die. Behold — the first extrusion was born! Today, just like in the past, extruded profiles are made from a grown-up version of the process. (Nice job there, [Joe!] Your concept managed to stick around for the last two hundred years.) From here, though, the origin of our beloved X-shaped profile gets a lot murkier.

Digging through patent after patent from the last five decades until now, I still can’t find a clear single inventor of the most common structural framing systems that we see nowadays. (Fellow engineering historians, HALP!) Instead, what I’ve discovered is a slow, maturing system of interlocks evolving for years without standard until they finally morphed into the modern systems like Rexroth and 80/20.

Reading the patents, it’s actually rather exciting to see common tricks both reinvented and improved over the years. Starting in the mid 60s, we start to see basic extrusion elements that interlock at the corners. Their cross-sections are rather complicated, but the fundamental concept of interlocking elements is present. In the late 60s, we start to see slotted inserts and derivatives. In the 80s, we see mature ‘T-slots,’ albeit in an unfamiliar profile. In the 90s, we have something that’s far more mature: a simple, repeatable slot pattern reproduced on various geometric shapes.

Should we be surprised that no sole human can call “structural framing systems” their personal claim to fame? Nah. Competitors riff off of each other time and time again. Thomas Edison’s lightbulb was inspired by a previous inventor, Joseph Swan, who goes on the record for being the first person to put a filament inside a vacuum to prevent it from burning out so quickly. Alexander Graham Bell, the most memorable of the telephone’s many inventors, based much of his research on both experiments conducted by Hermann von Helmholtz and, with some debate,patents already filed by Elisha Gray. Like many inventions, even those credited to a single person, structural framing systems are most-likely the creative effort of many curious people over many decades. On that note, we can thank our predecessors for refining a once-crude system into the tested-and-tried elements that we know today.

The Competitive Budget Alternatives

 

Extruded profiles used to be pricey, but these days a few Kickstarter campaigns have knocked down the price with a few cheaper lines, namely OpenBeam, MakerBeam and VSlot. Each of these extrusions is more than capable of handling the day-to-day extrusion needs of the hobbyist. Let’s not get too smug, though. Admittedly, when I first discovered these cheaper alternatives, I remember thinking: “Those poor industry design engineers — all of them enslaved by corporate partnerships to pay truckloads to some overpriced big-name profiles; meanwhile I can pay half price on something that works just as well!” Truth be told, these maker-breeds are fundamentally different, so it’s best to understand what we’re paying for when we pick a budget alternative.

So what sets those shiny, industry-standard profiles apart from these budget alternatives? First off, most of our friendly commercial leaders have been in the business for well over a decade, so it’s pretty safe to assume they’ve been chewing on their nails long enough to get a few things right. As a result, they’ve ironed out dozens of use-cases for extrusions and their various interconnects, everything from clean-room grade profiles to “here, hold my coffee.” Rexroth alone has over a hundred various extrusion types and a host of brackets and connectors for just about every possible hooza-ma-whacha-doodle we could possibly throw down on a factory floor. What we get from inventory is convenience.

Second, these companies have had both the time and resources to characterize their material library. What this means is that we can quickly find material characteristics of these profile, such as their section moduli and moments of inertia, to help us better estimate the limits of our materials before committing to them. (GIANT PDF Warning: Rexroth Structural Framing Elements Catalog) Finally, most heavy-duty industry players make their extrusions from 6061-T6, aka: good-ol’ aircraft grade Aluminum. Each of hacker-friendly maker alternatives is formed with 6063-grade aluminum. As for the differences, 6061 has about twice the yield strength and hardness over 6063. Should these structural differences matter to you? For small gantry builds, probably not. On the other hand, 6063 is noticeably softer than its 6061 cousin, so keep those corners safe! Just to emphasize the soft part, here’s the effect of a recent unintentional ‘impact test’ that happened during transit around my garage.

Again, nothing too serious here, but if that same dent landed on one of the internal channels where roller wheels would slide, the entire profile would be hosed, at least, for the purpose of guiding around a rolling carriage.

Even though most small builds will never need to consider some of the benefits earned by the industry leaders, it’s to our benefit to know what the implications are when we trade pricetags for a budget alternative.

Designing in a Real World

For the untrained soul, CAD might sound like engineering wizardry, but the actual process of modeling parts is actually pretty straightforward. Sure, every software package has its kinks (**cough** Eagle), but, like any capable user, we can learn to work around them to get the result we want. As for getting the real-life model to precisely match the dimensions of the CAD model? Now that’s a bit trickier. In this next section, lets take a look at some of the principles behind these parts and how to work with them in a way that preserves the dimensions that we first put into the CAD model.

Structural Framing Systems Aren’t LEGOs®

Structrual framing systems arrive with a host of classy brackets and interconnects to help us build up large structures quickly. However, while parts might seem to snap into place, the reality is that they’re a bit sloppier, so nailing down the dimensions that matter takes a bit more effort.

In a perfect world, these extruded profiles and brackets would behave like the LEGO bricks of our hey-day. In that world, profiles and brackets would only fit together in a limited range of configurations–each of which guarantees some sort of constrained relationship between those parts. Put better, as MIT Media Labs’ founder [Prof Gershenfeld] would say, “the metrology comes from the parts“. [Gershenfeld’s] best example of this concept is with LEGO bricks. Since each brick is properly dimensioned and only accepts certain configurations by which other bricks can connect to it, we can predict the sizes our our Lego configurations just by keeping track of all parts we used and how each part connects to the other. (Of course, this concept assumes that bricks are manufactured “ideally,” which , for all intents and purposes of the folks out there playing with LEGO, we’ll assume is true.) In that world, we wouldn’t need to measure anything, or, as [Gershenfeld] would say, “you don’t need a ruler to play LEGO.” Why? Because simply by inherent nature of the part metrology, assembling parts together would preserve an ideal geometric relationship between them. As a result we’d get two benefits: first, our components would fit together only in a limited number of configurations; second, once we assembled our components, their dimensions would directly reflect the CAD model.

Well snap, we don’t actually live in that world, so each one of our brackets and fittings has some form of slop. In each of these cases, we need to understand where this slop comes from and how to handle it.

Understanding “The Wiggles”

Lets get started by looking at the slop in common interconnect techniques.

Heads up: I’m using VSlot components for these examples since the OpenBuilds folks have been kind enough to provide the community with their CAD models. If you’re not using OpenBuilds components, don’t fret. Most structural framing systems come with a choir of brackets and interconnects that look just like the ones in these examples.

Coplanar Slop

Corner brackets are one of the easiest ways to join two profiles at 90-degrees, but they’re notorious for their slop. On the left is a depiction of just how much of a range of motion we have before we fasten down those screws. Where does this range of motion come from? In this case, we’re assuming ideal components, so the slop comes from the four gaps between each T-Nut connector that slides into the extruded profile.

Once we tighten down those screws, though, those two beams will be joined together rigidly somewhere in that generous range of motion.

Angle Slop

Plates are another option for joining both corners and other right-angle extensions. Just like corner brackets though, plate connectors are also victim to slop from the spacing of the T-Nuts when seated in one of the profile’s four slots. Here, plates constrain the coplanar relationship between the two profiles, but they leave us some wiggle room trying to establish the correct angle between them.

Pitch Slop

Most common profiles have holes on the end that can be tapped to accommodate a corner cube. Simply screw the cube into the newly-formed threads and enjoy a rock-solid corner connection. Nevertheless, this method is also susceptible to slop in the final angle of the profiles, relative to the cube.

Guaranteeing Assembly Relationships

Earlier, I mentioned that, despite the range of nice connectors, structural framing systems aren’t Legos! Unlike Legos, these connectors have slop, or “wiggle-room,” that don’t guarantee the assembly relationship that we want. In a nutshell, we can’t guarantee that joining two connectors gives us the self-measuring characteristics of Legos.

Well, rats, now that we know that “the wiggles” are here to haunt us for all time, how to get rid of them? Fear not, dear reader, your predecessors have solved this problem for you! To preserve a tight tolerance on relationships that our parts have in the CAD model, we can do one of three things. We can fix the problem of “slop” in the CAD model with better design techniques, in the part fabrication process with better tooling, or in the assembly process with better measuring tools. I’ll take you on a tour of each.

1. Fully-Constraining Parts

By far, our best option to fixing “the wiggles” is in the design. In principle, we need to ensure that the relationships that matter are constrained with parts that enforce those constraints. From now on, we’ll call any connection point between two or more profiles a junction. For any junction, if we’ve figured out a way to lock down all possible options for “the wiggles,” we’ll say that our junction is “fully constrained.”

Let’s take a look at that first 90-degree connection from above. Here, we can say that our corner bracket handles the 90-degree constraint but not the coplanar constraint. Hence, when we tighten down those screws, our two profiles could fall anywhere within that range of up-down motion shown above. To fix this problem–easy–we just add a corner plate! Now, not only is our joint guaranteed to be 90-degrees because of the corner connector, it’s also now guaranteed to stem off with two coplanar profiles.

Presto! We can use this same technique to ensure that T-intersections are also fully-constrained.

Not too difficult, eh? Let’s try another one. In this next instance, we’re trying to prevent the profiles from rotating relative to the cube. Two additional corner brackets handle that issue. Once installed, they guarantee that all three extrusions stem out from the corner cube in a way that’s coplanar to the cube surfaces.

2. Measuring Tools to the Rescue

The redundant-connector method isn’t too difficult, right? Unfortunately, this method for constraining junction points will cost us an arm-and-a-leg in extra connectors. What’s more, those extra connectors are, in fact, redundant. In both cases above, the first connector, once tightened down, will hold the profiles in place. All additional connectors are redundant in that they don’ t prevent the junction from moving any more than the initial connector did. The benefit from using redundant connectors is that we don’t need to find that “sweet spot,” where the profiles exhibit all the geometric relationships we want, before we tighten them down. The redundant connectors force the profiles into the right spot with their geometric properties.

However, with the right measuring tools, we can toss out the need for redundant connectors by, instead, relying on our tools to do the job of finding the “sweet spots.”

Squaring-Up Corners

Let’s start by locking down our right angles. For most of what we’re building, odds are good that the majority of our junctions intersect at right angles. This is likely the case since most vendors already have a host of connectors setup for doing just that: holding our profiles at right angles to each other.

Let’s say we’re looking to connect our profiles together at a nice T-intersection plate. Naively, we can just screw them in, but we’re subjecting ourselves to the possibility of misalignment. The solution? We need a reference of some sort. For us, that reference is a machinist square.

These machinist squares are essentially glorified right angles held to a very tight tolerance, somewhere between 0.0001 to 0.0008 inches (0.00254 to 0.02032 mm) for a modest set that wont break the bank. There’s no magic to using these. Simply hold it flush to the corner we’re interested in keeping square; then, gently tighten down the fasteners. Without much effort, we can guarantee a very high degree of accuracy across corners. Squaring up corners is, by far, one of the easiest operations we can perform to keep our gantry frame in-line with the design. Provided that we can readjust our connector setpoints, a few minutes with a machinist square can save us hours of potential redesign time.

Closing the Loop on Rough Cuts

Back when I assembled my first laser cutter frame from these profiles, I noticed a problem with the cuts. Right angles weren’t right angles! Like some cruel joke, the reality of I-have-no-idea-what-I’m-doing started to creep over my shoulder as I realized that my measuring tools just weren’t good enough. My frame was skewed beyond what my eyes could detect as erroneous.

A couple head-scratching weeks later, I became the proud owner of a set of Ebay-ed 40-inch calipers. With this nifty tool I could quickly measure the 0.4-mm difference between two parallel gantry frame components that were supposed to be identical in length. Well snap; my gantry frame was a giant parallelogram; I just couldn’t see it!

Buying a set of 40-inch calipers isn’t necessarily the solution to a square gantry frame. In fact, a machinist square from the above section should be more-than-sufficient to find frame misalignment and then correct for it. Nevertheless, a set of giant calipers will enable us to close the loop on our rough cuts and actually measure our longest profiles to a high degree of accuracy. For me, finding and fixing this problem with a better measuring tool has been a telling lesson that our resulting machine is no more precise that the tools we used to measure and align it.

3. Upgrading our Fab Equipment

Before we leave this section there’s one last culprit that will threaten the accuracy of our design: it’s our fabrication tools. Now, a host of folks online will tell us that cutting down extruded profiles is a bit like chopping wood–and it’s true! Hence, with the right blade, it’s actually pretty common to see folks cutting these profiles down to size with a chop saw.
To my fellow companions armed with chop saws, I must offer a word of caution about the quality of the resulting profile edge. First off, with the right blade, the cuts are, admittedly, very clean. The risk, though, comes in assuming the squareness of the cut. If you’re chop saw wont let you adjust the tilt and yaw angles of the blade, odds are good that those profile ends aren’t square.

So our chop saw cuts our part at the wrong angles; what’s the big deal, right? The problem with this method arises when we treat the butt ends of our profiles as references before tightening down their fasteners. I strongly recommend avoiding this practice unless the profile ends are measurably square with a machinist square. If they aren’t square, neither will the result angle be square if the chopped end is fastened down flat against any other part of the profile.

For clarity’s sake, I did just that in the image below: I chopped down an extrusion with my chop saw, butted it flush against another profile, and fastened it down with a T-plate connector. Can you see any issues? (Hint: zoom in)

Here, this profile was chopped with a chop saw. Unfortunately, since I can’t control the tilt angle of my chop saw, all parts chopped with this tool will have a slight deviation from 90-degrees at their chopped end. If I fixture the profile end to the surface of the bottom profile, I’m effectively using the cut end as a reference. Since my reference is slightly off from the nominal 90-degrees, so too will my profile when it’s fixtured this way.

To fix this problem, if we must put our profiles together this way, our best bet is to finish off the ends with a milling machine.

Provided that the machine is aligned correctly and the profile is fixtured properly, we’ll get an angle that’s about as square as we can measure. I faced off the end of the same extrusion with a Taig micro mill, and as far as my measurement tools are concerned, it’s perfect.

 

Keep in mind that this solution only matters if we’re using the ends of our profiles as assembly references. At the end of the day, not everyone has access to fancy equipment like a fully blown milling machine–and we don’t necessarily need them. For most instances, we can handle slight imperfections by accommodating them in the design as we did in the first section. In that spirit, I could’ve easily fixed the problem by fully constraining the profile with a redundant corner bracket. Alternatively, we can rely on our measurement tools to provide us with the necessary references as we assemble parts together. Here, I could’ve easily loosened the T-plate connector and bumped the profiles flush against my square before tightening them back down.

I’ve done my best to make this guide a list of tips-and-tricks for getting accurate builds from these profiles, but it’s far from complete. If you’ve got stories from the trenches about precision builds, let us know what we’re missing in the comments. Happy hacking!