A SEAM factory must produce a commercial product to sell at a profit. It could theoretically make anything but I think manufacturing machines (lathes, mills, 3D printers, grinders etc) is a good first focus. The SEAM factory needs these machines to operate so being able to make them increases its percentage replicability. Also, if you sell these machines it means you are making them at a value to cost ratio that is better than the competition meaning the SEAM factory will be economically viable.
So, designing a simple process that can make these machines from a few common raw materials seems like a good starting point. The thing that all these machines have in common is the electric motor, a lathe is a horizontal electric motor, a mill a vertical electric motor, a grinder is a portable electric motor, a 3D printer is just three controlled motors etc.
So, in order to make the machines we need to make the motor. This can be done with a single metal 3D printer and some assembly. Magnets and filament will also be needed (although there are some low efficiency motor designs that don’t use magnets). This guy has designed a good electric motor (80% efficiency) but winding the copper wire is too tedious and hard to automate. www.instructables.com/…/600-Watt-3d-printed-Halbach-Array-…/However I think we can 3D print the windings and just slot them into place which could be automated. This is something I’m definitely interested in working on. I also think this process could be adapted to make stepper motors needed for robotics and 3D printers.
So how do we get the metal 3D printer? Of course, there are lots of fancy, expensive options but Keith directed me to the Virtual Foundry which creates 90% metal filament that can be used with any cheap FDM printer. You then sinter the print in a vacuum furnace which I think could be as simple as hooking up a vacuum pump to a microwave, again this is another project that I’d love to work on.
So that’s it, one $500 FDM printer, one $100 microwave and one $200 vacuum pump + a bit of assembly and you can make any other machine be it a lathe, a mill, a grinder, or a 3D printer, etc. Of course, the quality of these machines will initially be low but they will be 10x cheaper than the current cost for these machines and the quality will quickly get better over time.
Some of the people do recognize that, so as I have seen the guy, so as there are others, and seems one pdf on that topic is somewhere inside the group posts(also 3d printing variations) – the point is, that is recognized as one of the keys pieces and it is an example of a good way of thinking.
I thought about that and decided against it. I always open for suggestions, but it should come with a road to implement it.
There are 2 reasons for that. Bill of materials and production complexity.
And as a secondary set of reasons – requirements for electronics which manages to make those engines to work, which probably turned to be a deciding factor, considering current constraints.
There are solutions for some combination of those problems, sort of pick 2 of 3 type solutions, but no good way to address the problem as a whole trough reusing of technologies, especially at the initial phase. It jumps to including quite a big chunk of technologies or to the inventing different approaches or to the need of significant improvement of existing technologies of 3d printing or to vitamin it with a lot of money. So you can’t claim and guarantee the successful result at this point because of – resources available for the task, technologies available to us today which are proven and are in use. So there is no solid base to place our claims or make promises with a high degree of confidence.
Generally yes, the direct brushless drive is a good stuff – power is there, RPM’s are there, a flexibility of RPM’s is there, size – they are good and convenient to use but there is a price package which comes with it.
As production of engines themselves, with brushes or 3phase motors – it is a simpler task. And yes, sure have to be recognized as one of the objectives, at some point, so as other types of engines and motors.
There are machines which wind coils and push them inside of rotors and stators, so as it is an absolutely different problem for big ones(done differently). And this is another reason for me to avoid reinventing stuff about electric motors, there already are solutions, which may ask/need for some improvements for our use case, but still, most of the work is already done and those solutions already exist, as a design for electric motors so as equipment involved in their production.
Winding coils as a problem existed for a long time and it has multiple solutions, some of which can be adapted. But accessing that data/information/designs is the next level of development, it needs resources and people.
So even if it is quite clear why would we like to make electric motors, jumping right in depths of the problem without preparing other stuff, and facing the necessity to reinvent things is premature, even with the excitement about 3d printing – some folks did that and they even succeeded but quality not up to the stuff you can do with conventional means.
There is another angle to approach the problem. Hydraulis.
It includes hydraulic pistons and hydraulic motors, and hydraulic motors with variable ratios.
As with any approach, it has its cons and pros, but if we think about the approach to the whole SEAM problem from the mechanical side, metal shaping metal turning/bending/welding/cutting etc, then pros of hydraulics as the main driving horse can outweigh the cons. So as the metalworking approach is historically grounded, it was done that way in the past – the development of our technologies.
There is a company which makes/designs hydraulic arms for underwater uses so as for other applications(quite a fresh one, sort of early stage, in Korea; ohh man, I wish we had resources right now to hook them). With specs on pair with currently used arms.
With hydraulics, good stuff is that torque is there in quite a compact form, RPM and force changes are possible and exist. Pressurising units can be hooked to any type of driving force(including powerful 3Phase motors which can be produced relatively easy so as their electronics isn’t that complex), most of the components are metals, which is relatively cheap. The stuff is widely used and has an okayish price tag(hydraulics in general). It is more about precision than quantity. And you can control a 100HP unit with simple 2 coils and varying voltage from 0 to maybe 9V – thus electronics like Arduino can do(which is cheap).
So controlling systems for it can be quite cheap. And it is a problem for high power electric stuff – low power electronics is abundant and relatively cheap, high power electronics is expensive and contains elements production of which requires the whole field of new technologies.
Hydraulics isn’t ideal, but considering how many fields of production it cuts off, and focuses on metalworking and precision metalworking it probably is a good tradeoff. (if we consider moon implication it also not ideal, but there are solutions for that and it is not that important at this phase)
Precision metalworking is one of the stuff which still is expensive today, it is the direction in which people still make money, and the field where they aren’t exactly doing it by hands but through tools and experience.
And when we do not think in conveyer therms but do that in terms of more flexible solutions, we can get our margin from that. The direction where once encoded experience works for you without requiring $20-40-50+/hour. Where a standby with humans is expensive, compared to a potential equipment which can change the stuff they do or not do.
Where today’s relatively cheap measuring tools, or expensive measuring tools providing precision which makes your eye to opens very wide – where that all helps for implementing of SEAM stuff.
It is the field where human hold mostly because of experience and skills by the nature of the processes, or where relatively simple but tedious processes may give you a precision one could not think is possible. (famous 3 plate method, or scraping when people shave 0.5 um by hands and different variation combinations of that and other approaches like grinding – really people do art pieces, from a machinists/toolmaker point of view)
A big brother CNC which makes parts according to G-code and is capable of 1/10000th precision, it is great – flexible, fast, complex and expensive – it cost like an airplane(small one). It is expensive for a reason, for a good reason, but it does not mean it does stuff better(it does some, complex shapes where it shines for sure, but not all parts are complex shapes, most of them aren’t). It is a tradeoff small footprint, high precision, complex shapes, high speed of production, flexibility in one package vs high price. The speed isn’t that high with complex shapes – 1-5h per part it is a norm – it replaces a lot of other operations allows producing some stuff in the way it wasn’t possible with old means – so there is a tip of the iceberg where the stuff really shines, but for the most part it is just the same good old stuff we using for last century(I mean isn’t that much better if at all).
10 or in some cases 100 times cheaper equipment and adding finishing process can produce the same parts as that big brother CNC. Not in all cases, but in most cases.
And for the ability of production, to be able to make certain stuff, not for the speed or for a human hours usage or footprint – combining few processes will result in the same quality of production as the big brother CNC. Not talking that for a hefty percentage of potential production using BB CNC is overkill.
so there is a potential to cut capital investments, and dropping the human from the equation allows to approach footprint stuff in a different way. So as different requirements for finishing and production process may allow organizing things differently in terms of footprint and speed of production, compensating some of the potential inefficiencies.
Installing cost, flexibility and low standby cost are important parameters for spreading of the system and be able to implement it in fresh locations. Not talking about obvious – no need for people with specific experience as labor to be available in the location.
There are problems to solve, and it needs to avoid the problems we can’t chew right now.
What an arm is good for, when it can position itself with a precision of a few micrometers but it can’t put a blank in a chuck. What’s good about it if you have to program it each time. An operation which stuff with the precision of 1mm and computer vision can do easily.(there already are movements in that direction cheap arms simple operations)
And we do not have to have a general computer vision problem recognitions etc, which is successfully solved in some implementations including some of the industrial settings. We may reduce the complexity of the problem because we can have it to be assisted by a model known by software, where it also can learn how to do stuff for a particular operation, without our involvement.
In that regard, the problem isn’t in the choosing what we should produce first, as part of SEAM installation, but what an existing process/product we can take and kick it a bit in a way which allows us to focus on essential solutions which can be developed further for use in more complex setups and processes. In a way where we reuse most of the currently existing new or old equipment, and build addon on top of it.
Measuring and quality control is an essential part of the SEAM process, so any currently existing production line could be a testing setup for us. And sure there are automatic solutions for that, but they sort of embedded in production lines and are specific for those lines, they aren’t general enough as we need it. For production lines, it doesn’t make sense to make them general, but for SEAM is one of the essential problems – the ability to verify you are producing what you expect to produce at any stage of the production process.
Or take flats as an example – it is a measuring tool, so as it is a product. It is essential for different stuff including calibration of existing machines and equipment, and the same function it may have with SEAM installation and be a product.
People did them by hands in the past, so it isn’t that complicated technology in terms of operations or actions, but there are some problems which it was hard to address back in the days. Today they are addressed by expensive, but very precise measuring tools, and it still did drop their cost btw.
But back in the days you needed an eye and paint to do the same – and until not so recent we could not address that problem in other fashion, but with the state of computer vision as for today, we can have the way to address the problem in an old fashion way – counting spots, I did that in photos of a sample in a software 20 years ago, today it is even a lesser problem.
Our time provides us with a new look at old problems, a new way to address them differently, maybe in oldish ways in some cases but with similar or the same quality of production as it is today on the market with lesser capital and operational cost.
It needs to recognize those options and take them as the first step. And there is plenty of stuff to take. And those which need a notch of changes are of interest to us for initial stages. Maybe it is a small percentage of positions of products available today on the market, but in terms of absolute numbers, even 10 positions, it is good enough(and there more than that). It is a possibility to devote financial streams and begin to fill our bucket to be spent on increasing complexity and applicability.
Essentially it may lead to producing solutions for a lot of stuff. We need to reshuffle existing solutions and add smart interconnections. And motors and other stuff will be just a point on the road.