Alright, time to dive into discussing the PrintNC build.
I got started on the PrintNC project after a long period of deliberation on exactly which CNC to go with, a little before the end of 2022.
It all originated from a need I identified when making soap – the need for custom soap molds. While it was easy enough to order them, they were rather expensive, and I knew that with Smooth On silicone or something similar, I could make my own and be assured they were exactly what I wanted. I didn’t want to have to place an order every time I wanted something new, be subject to die costs and minimums. I wanted that control in house.
The best way to do this was going to be to use a CNC to cut out mold masters. I had tried 3D printing, but the layer lines were unacceptable, and I couldn’t find any smoothing compounds that worked – so a CNC it was.
Initially the plan was just to get something small, but as I looked into CNCs and saw all I could do with them from a creative perspective, I decided to get something larger to make it worth my time and use it for other things as well.
I looked at a number of different options and felt that there was no perfect machine. Everyone had some compromise. You could get all of this, but it would cost that. You could get this, this, and this, but not that.
Obviously, if you paid enough you could get just about anything and everything you could want, but I had a budget to stick to, and ideally that really needed to be somewhere $3000 or less. Weighing the differing compromises between what was in my range became dizzying.
While I didn’t have control over what parts cost, I did have control over the labor and profit aspect of it. Building something myself would present a compelling way to get all or at least most of what I wanted and still stay in budget.
zNC Beta
In early 2022 I started designing a CNC and several months later finished a design for one with rotating ball nuts and double X gantries, so the Z assembly would be supported from in front and behind. Some of the parts needed for this build were beyond what my Ender 5 Pro could print in terms of size, which launched me into going through the same comparison process with 3D printers I had with CNCs, before eventually designing and building the massive cross gantry printer I’ve nearly got finished now.
That whole process of designing and building a 3D printer from scratch was eye-opening. There was a ton of work and time invested in that, and while it was worthwhile, I wasn’t sure I was ready just yet to spend that much time (or more) all over again on getting the CNC built and all the kinks worked out. I looked at a number of different ready-to-run options – Shapeoko, XCarve, Onefinity – but again they all had compromises I wasn’t happy with.
After a period of deliberation, I went with the PrintNC project. I would have to build it, so it would take some time, but it appeared to be the best overall solution between time and cost. It wasn’t perfect, but it was something to get started with at the very least.
PrintNC Kit
The PrintNC is a project you can either source entirely piece by piece yourself, or get a kit that has some of the parts all together. I went with the kit option, as it presented a pretty economically-priced bundling of all the different parts.
I ordered it ____. Probably the worst timing possible. The factories in China were going into Chinese Lunar New Year, a phenomenon unlike anything I’ve ever seen in America. Most of China basically shuts down for anywhere from a week up to a month, and there’s no official date for when any of it begins or ends, or how long it goes. That’s great for them, just something I had no real understanding of prior to this.
Communication with the factory, understandably, was very sparse. It was ordered on _____. It was shipped on or around _____ via UPS ____. The large package with all the long heavy parts in it arrived ______. The other smaller package with all the small parts sat in China at UPS for quite some time after that, and finally showed up _____. It was frustrating, but in their defense, it was horrible timing – so I’ll own that. Many others who ordered near the same time experienced the same delays.
Overall, the kit arrived in great condition. The packaging was phenomenal. I have not done a complete inventory on every last bit, but so far the only piece I’m missing is one of the collars that goes in the BK bearing block. It would probably take an act of parliament to affect such a minute change, but it would make much more sense to package those collars in the little sealed bag the ball screw end nut goes in, instead of inserting them in the BK block itself to fall out and get lost.
Sourcing the Steel
After checking with a few local places, I found a supplier that had the steel I needed that was super easy to work with. I took my machine dimensions, created a cut list, and then determined based on the lengths they sold it in – in their case it was 20’ lengths.
Before you order, I would advise checking to see what the lengths are, as some places I checked with sold it in 20 or 24’ lengths. They were able to cut things down in more manageable lengths, so I got 3x 12’ pieces and 2x 8’ pieces.
After getting it home, I measured everything out and cut it with the metal bandsaw. I’ve cut wood and aluminum before, but have never done steel. Luckily, the blade that I had was the right TPI (tooth per inch) and did a decent job cutting it.
While the recommendation is to use cutting fluid, I found over time that as it accumulated on the blade it started popping off. We cleaned all the fluid off and things improved, however it would still pop off from time to time. I would imagine the belt is probably stretched a bit, but it worked well enough to get through all the cuts I needed to do.
While cutting wasn’t that bad, drilling was an exercise in patience and a lot of frustration. In the PrintNC wiki, it’s stated that all you need is a cordless drill. I found out the hard way, likely through poor technique and possibly inferior bits, that drilling took quite a lot of work. In rather short order, I had a broken bit.
After a while of research and going back and forth on Discord, I discovered that it could’ve been a number of things:
- Too high a speed (opposite of many materials)
- Not enough pressure (opposite of many materials)
- Rocking back and forth, even a little (digs part of the bit in but not all of it)
- Starting and stopping and starting again (creates shards that the bit gets caught on)
It was amazing how long it took to get through a couple holes, even with cutting fluid, and what I thought were good bits. I got very discouraged, put everything up, and didn’t come back to it for a couple days.
Alternatives
In that interim period, I really grew to have a distaste for the steel.
The pros were that it was reasonably easy to source, affordable, and insanely rigid. The cons were that it was very heavy, not as versatile as extrusion, and much more difficult to cut, drill, and tap.
While the decision to use steel instead of aluminum was understandable – the concern of rigidity probably being at the top of the list – there wasn’t a lot of information available on the testing that had been done, which made me skeptical of the decision to seemingly throw the baby out with the bathwater.
What type of aluminum was used for testing?
Was steel chosen because it truly was the better option all around, or just because it was easy enough to get and far stiffer than an equal piece of aluminum?
I’m certain there could’ve been some rather involved analysis, but it was nowhere to be seen. Given all that aluminum had going for it, I had to get to the bottom of it.
I had this inkling feeling that there could be a solution with aluminum that would work. While aluminum is a little more difficult to source and not quite as stiff, if it meant it was far easier to work with and more versatile, I felt those could be reasonable compromises.
At any rate, the curiosity got to me. Most breakthroughs, or even improvements are discovered through AND tested by skepticism.
Analysis
I decided to run some calculations. To cut to the short, I went to ChatGPT to learn rather expeditiously what the figures were that were involved in performing analysis on rigidity. While there were programs like Fusion 360 that could do this, I wasn’t interested in learning all that; I just needed some quick answers to give me an idea on whether to continue down this road or not.
The primary concern was with the X gantry, where there’s a suspended weight – the Z axis and spindle assembly – that works on and stresses the gantry. What I was searching for was the maximum deflection I could encounter with a given weight or force. I felt as though that was one of the most important measurements of whether or not something else could work, as it’s one of the most stressed components of the machine.
Solving for Deflection
When you calculate the rigidity and the deflection of a long beam under weight, In these types of calculations on a simple long beam or piece of metal, there are a few variables:
- Modulus of Elasticity
- Moment of Inertia
- Length
- How the beam is supported (simply supported, ie just sitting on something, fixed support, attached to something, etc)
The Modulus of Elasticity is a figure that, for the purposes here, tells you what you need to know about a given material and its resistance to deformation. That’s a horrible simplification and not very textbook, but that’s what it is. The MoE is specific to a material; for any given type of aluminum, any single alloy or version, the MoE is the same, no matter what size or shape of object you have.
The Moment of Inertia is the figure that, for the purposes here, tells you what you need to know about the shape itself, the cross-sectional profile, and its characteristics as it pertains to deformation. Another horrible simplification, but finding out the differences between these two figures that one sees thrown around so often helped me a lot in understanding things.
The length takes the cross-sectional profile and creates a three-dimensional object from a two-dimensional shape. The properties of the shape (MoI), along with the properties of the material (expressed in MoE), and the length give you a unique set of variables that, when a specific type of support is determined, an equation can be used to determine the deflection.
The first formula I started with was the one that, in my case, would show me the worst possible deflection I could encounter, given a particular load. That’s the formula that describes a load of a given weight, dead in the center, with the beam supported on both sides. There are two formulas, the first for one that is simply supported, ie just resting on two supports, and the second for one that has fixed supports, ie is bolted to something that is entirely rigid.
- Simply Supported: δ = FL3∕ 48EI
- Fixed Supports: δ = FL3∕ 192EI
The variables are as follows:
- δ = maximum deflection
- F = force or load
- L = length
- E = Modulus of Elasticity
- I = Moment of Inertia
As you can see, the two formulae are pretty close to one another. One important thing is that all numbers must match in terms of units.
The length and MoI will both be measures of length or area and must match, be it inches, millimeters, centimeters, etc. The load and MoE will both be measures of weight and/or force, and also must match, be it GPa, Newtons, PSI, etc. Google can do conversions on these very easily.
To calculate all of these quite easily, I created a Numbers (or Excel, if you prefer) spreadsheet with formulae that I wrote out, that I’m happy to share here.
I created five columns: Load, Length, MoE, MoI, and Deflection. The first four columns you fill out. The last column contains the equation below. Obviously, you would replace Load, Length, MoE, and MoI with references to cells that you input those figures into.
- Simply Supported: Load * Length^3 / (48 * MoE * MoI)
- Fixed Supports: Load * Length^3 / (192 * MoE * MoI)
In the case of an X gantry, the actual figure will be somewhere between the two. While it is certainly not sitting freely on two supports, and it’s likely closer to fixed supports, there may still be some give or stretch in the system.
Results
What I found, based on data I’ve supplied for reference at the end of this post, is that the difference in maximum deflection between the 50mm x 75mm 11-gauge structural steel, and a similarly sized 80mm x 80mm section of 40 Series aluminum extrusion – both simply supported – was negligible.
In fact, based on 20lbs of weight hanging from the center, the deflection on the aluminum was 0.01mm less than the steel, a 16% improvement. It wasn’t until you went to a 50mm x 100mm 11-gauge steel profile that you saw an improvement in the steel over the aluminum, and even then it was only 0.02mm less.
You could increase the force on both and the deflection increased in a fairly linear fashion.
When you looked at all of those in a fixed support scenario, understandably, they all delivered significantly better numbers.
Going Forward
Prior to testing out these figures I’d been working on a much simplified design with aluminum extrusion. I took inspiration from the simplicity of the PrintNC, and designed an original machine built from aluminum.
My goal was to reduce the number of parts, the number of joints, the wasteful use of excess.
I took a number of the pain points, and I’ve been working to solve them to present something that is just as or even nearly as good, while being far simpler to assemble.
The question at this point was whether to order a bunch of aluminum, another $400 or so, or give the steel another go.
I decided to switch out the metric bits for some SAE / imperial bits that were much easier to get close by, and were what I felt would be a higher quality. I found out that of the 5mm and 6.5mm bits I needed, a 3/16” (4.8mm) and 1/4” (6.4mm) bit would give me nearly what I needed. I also decided to set up a rig to support the beams so I could use my drill press.
Luckily, this second attempt fared much better. The combination of the better bits and the more stable approach with the drill press proved to be a winning combo. The holes came out much better.
Unfortunately, even with the drilling working better now, there’s still a ton of holes to drill. As I’m drilling out eighty holes in five bottom crossmembers which will interface to the two side rails, which will soon have another forty holes drilled into them, and tapped mind you, the satisfaction of being able to drill better is dulled (no pun intended) by the monotony and the feeling of excessiveness in this set up.
I will pick back up on this comparison and chronicling in another post soon, to discuss a torsion box, rigidity, and more alternative ideas.
Overbuilding is common in engineering, and while it works,
The sheer weight of it, which seemed excessive, and the difficulty of working with it between cutting and drilling and tapping really turned me off.
The impression I got was it was overbuilt. While steel was definitely stiffer, I wondered if the better solution might not be a better aluminum option, than this steel one.
I did some exploratory work on aluminum. While it’s not as stiff, it is much easier to work with. But it’s just not popular, and I really wanted to understand why.
There’s a lot of talk about the inferiority of aluminum for something like this, but no real data showing numbers and evidence. While there’s been a lot of testing done, I haven’t seen evidence of a lot of engineering in a more mathematical, absolute sense. Aluminum didn’t work, so we switched to steel. I wondered, is it