The great thing about CNC is that given a correctly drawn file, with correct endmill selection, appropriate feeds and speeds, and matching toolpaths, a file will be cut to make a correct part so long as nothing goes wrong during the cut.
The awful thing about CNC is that a part will only come out correctly if the file is drawn correctly, the right endmill is selected, feeds and speeds match the material and endmill, toolpaths are properly assigned, and nothing goes wrong while the cut is being made.
Thank you for the comments. This is valuable to me, I’m not a machinist but am working to tackle these common issues that are acting as impediments to those just wanting to make parts. The machining market as a whole is at least 30 years behind in process, horribly inefficient, which has the unfortunate effect of damping the interest in new ones wanting to enter the field and making it more difficult for those in production environments to become more profitable.
I don’t think this set of problems can be solved with raw process alone, It’ll need some software and hardware mixed in but I do think it scales quite well, at the very least it will allow smaller operators to punch above their weight on the output quality and speed side of things.
Yeah, things which I think ought to be automated / standardized:
endmill entry — there really should be standard files for manufacturers to make which software can import — Autodesk Fusion 360 accepts XML definitions but that’s not widely accepted unfortunately
integrating those twain should be easier — some sort of machinist notebook might work well, I’ve tried to use Jupyter Notebook for this: Jupyter Notebook as a CNC machinist notebook but it requires more discipline than I’ve been able to summon as-of-yet
There was a neat bit of discussion about what a “Shapeoko.app” might look like on the old Shapeoko forums ages ago — it slightly influenced the wiki, and I think informed some aspects of:
and see the nascent ebook at:
I’ve been trying to work all of this out in a Kickstarter/website/project/book as discussed at:
For commercial efforts, Design and Make is notable, and of course, Autodesk Fusion 360 overshadows pretty much everything else for folks who are willing to sign up with Autodesk.
Keeping things as open source / universal would be the best bet at streamlining the flow.
Dragging and dropping files or paths into open source type programs and applying straightforward settings or guidelines to the toolpath would be something.
Even having the software open and have a novice/intermediate/advanced interface would be a unique twist.
Create or drop a CAD model
Drag and drop a readily available tool file into the workspace
Drop down window to specify the machine and perhaps a few different parameters
Select or specify your workholding via model or some generic setups
Via few different clicks have some easy to understand toolpaths applied
A very intensive simulation with crash detection etc
Walk through the work and tooling setup
Unfortunately stuff is proprietary, theres lots of options/machines and some dont speak the same code, tons of mods/configurations, processing power, etc etc.
There has been a lot of focus on breaking down the feeds and speeds lately which is a big hurdle once you have everything else figured. However, even just CAD modeling for manufacture can be critical/helpful as you can create the same part but one is inefficent and takes 2x or more time to machine than the other and the function the same. It’s really just a lot to take in and understand to get really good results that you expect every time (more/less).
The problem is tying everything together for a nice, neat end-to-end workflow — that’s what I’ve been working on with https://designinto3d.com/ — current effort is to work up the toolpaths using https://tplang.org/ which would preview in CAMotics).
I don’t think standardization is possible at this point, too many languages and dialects. Tool manufacturers bear some responsibility for this, but understandable given the need to extend platforms, although it’s also a way for vendors to lock in customers forever.
Open source doesn’t work, no practical funding path besides commercial that is sustainable, plenty of competent devs but that’s not where the money is spent, usability wins markets, not code.
I’m concentrating efforts on the production side, an ecosystem that can be used by complete novices and experienced machinists alike. The difficulties and mistakes made are common regardless of experience level, that points directly to a process improvement and optimization issue. The math, physics, material science is decently solid on hardware, but process really hasn’t changed in half a century.
I get all those and they are handy for the beginner, but an intermediate or advanced option as well.
The struggle also becomes the never ending update to the software and sometimes loss of aesthetic or polish feel because it is a never ending beta in true nature.
What if you would like to incorporate a spiral toolpath with a specific stepover for the floor or some intentional scallops on the surface finish. Most easy going software is limited b/c it takes a lot of work for it to become a powerful option and then even more so to be intuitive.
Process as in being able to have a novice sit down at a machine and be making parts within a few hours and the experienced operator being able to learn the system in the same amount of time and output higher quality at increased volume if needed, or being able to run more projects in the same amount of time. Levels of difficulty within design (beginner, intermediate,expert) aren’t effective, people no matter the existing skill level, benefit equally from an optimized workflow.
The industry doesn’t have a technology or feature problem, it has a usability problem. There will always be another company able to undercut on cost, but it’s practically impossible to copy an ecosystem. This particularly benefits smaller machine manufacturers since their support load drops significantly. It’s more intensive on the front end but gains are long term.
Maybe I’m missing something then because the instances Will mentioned replicate what you’re saying to some effect.
Can you provide examples elsewhere where a process made such a drastic improvement?
3D printing seems to have this sort of process, but there are far less variables to account for in the end.
Material size/location, workholding method/size/location/holding ability, machine rigidity/speeds/feeds/work envelope/repeatability/accuracy, tooling size/geometry/effective F&S/stickout, controller processing power/memory, etc. Once all of this is addressed you change a few pieces of hardware and a whole new range of parameters to adjust or tune. This is a lot of data to capture and have readily available to limit the knowledge needed by the end user. A tremendous amount of data within a system
We’re probably talking past each other, a box of parts doesn’t make an engine unless assembled well.
You’ve enumerated a load of information that a user currently needs to have available to him. Have you ever questioned why? Every one of those issues you’ve mentioned are an impediment to the end user.
I suppose it comes down to how a companies customer base values their time and how big of a market they’d like to capture. If you’re comfortable being strictly in the tinkerer type of market where customers value their time at a low rate then it works, with the downside being lack of mobility upward and new learning curves as the user outstrips production capacity, willing to put up with downtime. With an effective workflow provided by the machine vendor the user won’t experience the start/stop sort of activity that has become prevalent at the small side of part manufacturing. This is a business exercise, straight time and money calculation, if it’s a hobby it’s still a time / money calculation, you’re being paid in having fun.
Nearly every industry has benefited greatly by optimizing how the user interacts with a piece of equipment. The work I’m doing may be a little upstream but it is one of those areas where it scales downward as well. Reliable understandable low barrier production beats break / fix cycle every time.
I understand you are suggesting a new process, but it is all vague mentions of improvement.
If the industry is at need are you trying to cater to hobby and scale up, or start with high end production and scale down?
The high end production arena has some very impressive ecosystems which alleviates much of the information needed by the end user. It comes at a price, and much of the time is proprietary since the companies do not tend to give it away.
I’m curious what the next steps are, actionable items, creating the process.
Examples of previous instances in other industries, specific instances, to further understand this whole exercise.
Just like a word processor does not make you a typographer, a beginner can produce stuff but to do things beyond the basics, you need knowledge to operate a CNC. It is not a Star Trek replicator, it is a tool that can be used to produce some parts using some materials when you know what you are doing.
If I want to produce a widget, I first need to understand how this part can be realized. A beginner will be able to produce simple 2D parts on a CNC, 2.5D will require more in depth knowledge while a 3D part will require not only require advanced knowledge but also significant equipment upgrades to be able to make the part by adding at least one or two axis so the parts can be be rotated to be milled.
Before wanting to change everything,I would suggest you get a system, try to see what works and work on improvements from there standing on the shoulders of people from this and other forums. You need to build up that experience to get that understanding.
Trickle down, it’s much easier to charge more when you have demonstrable time savings, increased profit and less waste from a solution, not much sense in fighting upstream. Interestingly the hobbyist type market indirectly benefits from the improvements, but for different reasons. The widening skills gap in the manufacturing industry worldwide is making these changes mandatory, the upside being easier to use equipment at the smaller end of the spectrum.
I’m not suggesting that tech will replace everything, but hearing the same chronic problems from a wide variety of experience levels within the same vertical indicates an opportunity for improvement. It takes time and effort but I’m confident of a successful outcome.
It’s not clear to me where you’re going with this. Are you trying to solve all ineffiencies in machining? Trying to build a better benchtop machine for the same market as Shapeoko 3, X-carve, openBuilds, etc?
There will always be limits and there will always be people pushing the limits, you can be confident in that.
On the back end of the process - I’ve always like the Datron hardware and software. I’ve never used it but having tools like the probing/camera system available on a cheaper machine would be great. The software leaves a lot of other systems in the dust in my opinion
Obviously having having an understanding of the front end of design & toolpath creation is critical. but for someone who is half decent at that side of it, I always yearn for something like the Datron system for day to day use. I’d be quite happy to forego some accuracy to get the cost down
With an app like CNCjs where the camera is already connected and it is made to control the CNC and probe, some great people maybe able to add similar capabilities. I know there are several steps to get there but there is something to start with.
I’m not sure it’s possible to solve all inefficiencies in any process. Benchtop is a very small segment of the market, physics limit the possibilities at that form factor although the small form factor machines can benefit from the improvements.