This guide is legendary. It helped me to learn and follow the process, and I made some pretty successful parts. For example, my own keycaps for all keyboards that I use: much larger than usual (for my large fingers), with a pleasant gently matte texture, in beautiful colors.
Polyurethane resins are an amazing achievement, and very much underrated. So are platinum-cure silicones. With care, you can get design-to-parts precision of ±25μm, which is spectacular (and a bit surprising, too). The fact that modern polyurethane resins (Sika Biresin F50) have essentially zero shrink helps quite a bit, too.
Incidentally, there is a whole bunch of youtubers doing casting using epoxy resins, or cheap silicones, there is a large following, but this is not representative of what the techniques really allow.
If you want to step up from 3d-printing, this is the way to go! Especially given the proliferation of inexpensive desktop CNCs with really good precision (Makera and others).
Those resins are absolutely fantastic but do read the MSDS and be very careful, it doesn't take much to get yourself in the emergency ward with that stuff. Another risk to be acutely aware of is that these reactions usually are exothermic and can go runaway faster than you can blink of the conditions are right.
Of course, one should always read the MSDS. I use a 3M respirator with VOC inserts while working on these things. However, one should mention that a) polyurethane resins and platinum-cure silicones are much safer than many other compounds, and b) polyurethane resins are different from the more common epoxy resins and reactions are only very slightly exothermic. It's not a problem like when you're building a river table.
As a rough estimate, using a resin 3d printer is more problematic than these compounds.
A CNC router is on my list of tools to figure out and own. Routing aluminium, wood, and things like HDPE and being able to make moulds for silicones and resin? Yes please. 3D printing on the other hand never appealed to me.
I would strongly recommend NOT following the general advice of buying a cheap "3018" or something similar. Makera Z1 should be the baseline. Otherwise you're stepping into a world of frustration where you will spend most of your time trying to get your tool to work, rather than getting parts produced.
Unfortunately, reasonably precise and rigid mechanical assemblies do have to cost a certain amount of money.
Agreed. If you want to just make parts and not tinker with a CNC machine, get a Z1.
I had near-zero experience with CNC and got a Cavera Air last year and it mostly "just works" from the hardware side. I just fixture stuff and run my gcode, zero issues with the hardware. The Z1 seems to be even more streamlined w/r/t things like chip evacuation.
But, my god, Makera's firmware/software is fucking garbage. Especially the CAM workbench.
The community firmware and controller software (https://github.com/Carvera-Community) is so much better and feature-filled that it's kind of sad. They also have a tool library and post-processor for the FreeCAD CAM workbench in that repo which will let you make a clean break from Makera's terrible software.
On the upside: Makera apparently won't invalidate your warranty for using the community firmware/controller software, which is nice.
Start with one of the cheap kits on Amazon. A good chunk of the learning curve is software/design/workflows. On the machine side, learning how to properly secure your work pieces, and find the right bits, speeds, and feeds is another art. You can do all of that on a ~$300 3018 CNC kit. Your work output is limited in size, and precision, but that doesn't matter as much when you're just trying to get the hang of things.
I have both, and a manual lathe and mill, and a laser cutter. 95% of everything I do is with the 3D printers. There is no indexing, no work holding, no dealing with shavings or smoke or dust or cutting oil that gets everywhere, no accidentally breaking your last end mill, no screwing up the only one of the thing you're cutting into, no cutting down stock so it fits in your machine, etc. You just press print. Setting up another machine is a right hassle by comparison.
3D printing and plastic parts isn't good for everything, but it is good enough (and easier) for a lot of things.
Yeah same. I’ve done a lot of CNCing but 3D printing isn’t appealing because I don’t care much for making plastic parts. When metal 3D printing becomes hobby tier I’ll be all over it.
Today 3D printing makes a lot of thing possible. Now that multi-toolhead printers are coming, some already available, it's possible to make composite parts. Like hard frame in soft wrapper, conductive lines (resistance still high), etc. I'm still learning, but it's exiting.
As for CNC, some cheap tabletop are available. FreeCAD is useful for design and g-code generation. The problem with cheap they are imprecise and shaky. I'm thinking about using 3d printed frame with metal everything else. Should be light enough to lift with one hand. For precision it'll need calibration from time to time as plastic moves. The goal is to have 3 axis mini CNC mill able to cut soft metals with precision better than 0.1mm.
which a couple of projects have copied (with attribution) and at least one company copied (w/o acknowledgement).
3D printing seems to have taken up a lot of this interest due to how much easier is to source and load filament and start a print and how good opensource slicers are as opposed to traditional CAM software.
One thing which has helped it be somewhat timeless is only covering mechanical and geometric considerations in detail --- rather than cover software specifics it off-loads things to the Wikipedia article on Computer Aided Manufacturing, which has seen a lot of edits since this page was first put up --- my own usage has gone from my first project:
- requiring that I make a font with a character which had all the text and markings of a caliper dial using FontForge
That said, there are still some folks making new software: https://news.ycombinator.com/item?id=47638498 and FreeCAD's 1.1 release seems a big update (despite the new book _FreeCAD Beginner's Handbook_ being branded version 1.1 on the cover but telling one to download 1.0 or later and providing screengrabs and instructions for 1.0).
This guide is legendary. It helped me to learn and follow the process, and I made some pretty successful parts. For example, my own keycaps for all keyboards that I use: much larger than usual (for my large fingers), with a pleasant gently matte texture, in beautiful colors.
Polyurethane resins are an amazing achievement, and very much underrated. So are platinum-cure silicones. With care, you can get design-to-parts precision of ±25μm, which is spectacular (and a bit surprising, too). The fact that modern polyurethane resins (Sika Biresin F50) have essentially zero shrink helps quite a bit, too.
Incidentally, there is a whole bunch of youtubers doing casting using epoxy resins, or cheap silicones, there is a large following, but this is not representative of what the techniques really allow.
If you want to step up from 3d-printing, this is the way to go! Especially given the proliferation of inexpensive desktop CNCs with really good precision (Makera and others).
Those resins are absolutely fantastic but do read the MSDS and be very careful, it doesn't take much to get yourself in the emergency ward with that stuff. Another risk to be acutely aware of is that these reactions usually are exothermic and can go runaway faster than you can blink of the conditions are right.
Of course, one should always read the MSDS. I use a 3M respirator with VOC inserts while working on these things. However, one should mention that a) polyurethane resins and platinum-cure silicones are much safer than many other compounds, and b) polyurethane resins are different from the more common epoxy resins and reactions are only very slightly exothermic. It's not a problem like when you're building a river table.
As a rough estimate, using a resin 3d printer is more problematic than these compounds.
A CNC router is on my list of tools to figure out and own. Routing aluminium, wood, and things like HDPE and being able to make moulds for silicones and resin? Yes please. 3D printing on the other hand never appealed to me.
I would strongly recommend NOT following the general advice of buying a cheap "3018" or something similar. Makera Z1 should be the baseline. Otherwise you're stepping into a world of frustration where you will spend most of your time trying to get your tool to work, rather than getting parts produced.
Unfortunately, reasonably precise and rigid mechanical assemblies do have to cost a certain amount of money.
Agreed. If you want to just make parts and not tinker with a CNC machine, get a Z1.
I had near-zero experience with CNC and got a Cavera Air last year and it mostly "just works" from the hardware side. I just fixture stuff and run my gcode, zero issues with the hardware. The Z1 seems to be even more streamlined w/r/t things like chip evacuation.
But, my god, Makera's firmware/software is fucking garbage. Especially the CAM workbench.
The community firmware and controller software (https://github.com/Carvera-Community) is so much better and feature-filled that it's kind of sad. They also have a tool library and post-processor for the FreeCAD CAM workbench in that repo which will let you make a clean break from Makera's terrible software.
On the upside: Makera apparently won't invalidate your warranty for using the community firmware/controller software, which is nice.
Start with one of the cheap kits on Amazon. A good chunk of the learning curve is software/design/workflows. On the machine side, learning how to properly secure your work pieces, and find the right bits, speeds, and feeds is another art. You can do all of that on a ~$300 3018 CNC kit. Your work output is limited in size, and precision, but that doesn't matter as much when you're just trying to get the hang of things.
I have both, and a manual lathe and mill, and a laser cutter. 95% of everything I do is with the 3D printers. There is no indexing, no work holding, no dealing with shavings or smoke or dust or cutting oil that gets everywhere, no accidentally breaking your last end mill, no screwing up the only one of the thing you're cutting into, no cutting down stock so it fits in your machine, etc. You just press print. Setting up another machine is a right hassle by comparison.
3D printing and plastic parts isn't good for everything, but it is good enough (and easier) for a lot of things.
Yeah same. I’ve done a lot of CNCing but 3D printing isn’t appealing because I don’t care much for making plastic parts. When metal 3D printing becomes hobby tier I’ll be all over it.
Today 3D printing makes a lot of thing possible. Now that multi-toolhead printers are coming, some already available, it's possible to make composite parts. Like hard frame in soft wrapper, conductive lines (resistance still high), etc. I'm still learning, but it's exiting.
As for CNC, some cheap tabletop are available. FreeCAD is useful for design and g-code generation. The problem with cheap they are imprecise and shaky. I'm thinking about using 3d printed frame with metal everything else. Should be light enough to lift with one hand. For precision it'll need calibration from time to time as plastic moves. The goal is to have 3 axis mini CNC mill able to cut soft metals with precision better than 0.1mm.
Previous discussions:
https://news.ycombinator.com/item?id=34339459 (157 comments)
https://news.ycombinator.com/item?id=27645605 (74 comments)
https://news.ycombinator.com/item?id=41989322 (106 comments)
Note that there's a version of it as a single page: https://lcamtuf.coredump.cx/gcnc/full/ which is a bit easier to reference.
Used it a lot as a guide when working on the Shapeoko wiki (which is now off-line), so maybe some of it survives in:
https://shapeokoenthusiasts.gitbook.io/shapeoko-cnc-a-to-z
which a couple of projects have copied (with attribution) and at least one company copied (w/o acknowledgement).
3D printing seems to have taken up a lot of this interest due to how much easier is to source and load filament and start a print and how good opensource slicers are as opposed to traditional CAM software.
One thing which has helped it be somewhat timeless is only covering mechanical and geometric considerations in detail --- rather than cover software specifics it off-loads things to the Wikipedia article on Computer Aided Manufacturing, which has seen a lot of edits since this page was first put up --- my own usage has gone from my first project:
- requiring that I make a font with a character which had all the text and markings of a caliper dial using FontForge
- Making G-code for that w/ F-Engrave: https://www.scorchworks.com/Fengrave/fengrave.html
- Making the G-code for a fixture and the balance of the design using a Flash file: https://github.com/Jack000/PartKAM
- sending the design using a Python file, bCNC: https://github.com/vlachoudis/bCNC
(ob. discl., I work for Carbide 3D)
These days, I'd just draw it up in Carbide Create: https://willadams.gitbook.io/design-into-3d/2d-drawing make toolpaths in that application: https://willadams.gitbook.io/design-into-3d/toolpaths and then send the G-code using Carbide Motion: https://willadams.gitbook.io/design-into-3d/machine-motion (other manufacturers have similar workflows or bundle commercial software such as CarveCo or Vectric, and of course, there's Autodesk Fusion 360 which has taken up just about all the oxygen for free usage) and I've moved on to making my own project using Open(Python)SCAD:
https://github.com/WillAdams/gcodepreview
That said, there are still some folks making new software: https://news.ycombinator.com/item?id=47638498 and FreeCAD's 1.1 release seems a big update (despite the new book _FreeCAD Beginner's Handbook_ being branded version 1.1 on the cover but telling one to download 1.0 or later and providing screengrabs and instructions for 1.0).