Manufacturing is far more complex than most people imagine. I laugh when I see commercials showing craftsmen lovingly making YOUR mattress or some such thing when you know damn well that it looks much uglier and far less custom/manual in reality.
Take the task of manufacturing and bringing to market an electronic product. Here's an incomplete list of some of the tasks requiring attention:
- Schematic design
- Mechanical design
- Plastics
- Machined metal
- Sheet metal
- Wire harnesses
- Optics (for lights, light pipes, etc.)
- Label, decal or printing design (legends, etc.)
- DFM (Design for Manufacturing
- Component sourcing
- Staging according to lead times
- Strategy to ensure long-term availability
- Testing methodologies
- PCB Layout
- Electronic Assembly
- Surface mount
- Through hole
- Manual soldering
- X-ray inspection
- Automated optical inspection
- Device Programming
- Testing
- Packaging
- Packaging design
- Cardboard
- Foam
- Injection molded
- Cut and glued
- Box
- Labels
- International variants
- Regulatory testing
- CE
- TUV
- FCC
- UL
- Other countries
- Mechanical manufacturing
- Injection molded plastics
- Machining
- Sheet metal
- Stamped metal
- Mold making
- Jigs and fixtures
- Software development
- Embedded
- FPGA
- Workstation
- Documentation
- Manuals
- Website
- Brochures
- Support
- Develop testing and methodologies
- Support scripts
- Support systems
- And more...
The list is not complete of course, but it provides some idea of how complex it can be. The idea of reducing this to an API or solving it via 3D printing is nice, but, today, a fantasy.
There are issues that are not obvious from simply reading this list. For example, entrenched systems and software that does not inter-operate. Disciplines that require massively different infrastructures (machine shop, plastic molding, packaging, PDB assembly, wire harness fabrication) and are sometimes islands in and of themselves.
Just looking at CNC machining is an example of how arcane things can be and how difficult it can be to change the way things work. CNC machine manufacturers use G-code (which is older than dirt). You'd think that by now this would be a smooth system. It isn't. Software vendors produce tools that require customization for each machine type and, even then, can produce g-code that is actually dangerous to run.
I had exactly this experience when I decided that we needed to have our own CNC machining in-house. We bought hundreds of thousands of dollars worth of (mostly) Haas equipment. All g-code was generated with a CAM tool that ran inside of Solidworks. I'll fast forward to the day that one of the machines --with a 20 Horsepower spindle-- happily churned aluminum with a one inch roughing bit like it was butter. It was amazing to watch, and it sounded horrible. The CAM program simply put out the wrong instructions because the engineers who created it in India decided to have their own creative interpretation of how G-code should work.
Anyhow, the point is that manufacturing is hard and complex. That's why you don't see Y-Combinator style "Here's $15K, go start a company" incubators doing real hardware. In the hardware world, outside of the trivial, $15K won't even cover the tooling.
Maybe in a few hundred years we'll have Star Trek style replicators. How cool would that be.
One can see in this list that there are many ways to ease the pain of bringing an object to market without directly getting hands dirty in manufacturing. It'd be so much better if the burdens of sourcing, certifications, shipping, liability insurance, etc. could be taken off the engineer's plate.
Shipwire does this in fulfillment but the cost can be a barrier, and mfg.com or Alibaba are a step in the right direction but don't necessarily address the trust you need to have with your suppliers in lieu of becoming an expert. It's too much of a Wild West right now, and we need a sheriff.
Alas, it's not yet as simple as the "Print to China" that manufacturing-as-a-service promises. There are currently too many human translation layers that would impede reliable manufacturing APIs.
Yea, CAM is its own kind of hell - with it's pre/post-processors - crazy array of options, machine-specific GCode and bugs, etc.
A big issue manufacturing faces is that none of the people starting companies are software guys - all of them are MechE's or something who think hardware first, so the CAM ends up far worse the controller (probably a FANUC...), which is far worse than the physical machine - you do what you know.
Certainly, trying to go from CAD --> finished consumer product ain't happening next year, but we're seeing signs that we'll see that with more simple processes pretty soon.
Adjusting a CNC machine is very complicated and requires a lot of experience and skill.
One lesson learned is, that craftmans that learned on shitty old russian manuel lathes did a lot better on modern machines than the ones that learned on super precise automated modern western machines.
The abstraction layer between metall and man that modern machines make does impede the feeling for the process.
The feeling and understanding of the cutting is essantial to make a good cut, weather it is automated or not.
Craftmansship is not automatable, weather it is software engineering or manfucatoring.
I am one of those rare engineers who is equally at home writing embedded software, web software, iOS software, workstation software, designing electronics hardware (from simple analog go GHz range FPGA's) as well as mechanical.
And, of course, as a total geek, I also jumped into the manufacturing processes for all of the above. I am equally at home in front of a SMT pick-and-place machine as I am in front of a VMC or a table-saw.
I have to tell you that some of the most gut-wrenching design work I've had to face sometimes wasn't how to design the thing but figuring out how to actually make it. From specialized fixtures to custom tooling or a weird manufacturing process.
Frankly, this is some of the most valuable knowledge that our country is loosing to other shores as manufacturing continues to be decimated. It's one thing to know how to use a VMC. It is an entirely different thing to know how to engineer, setup and execute on an end-to-end manufacturing process.
Just remembered a potentially tragic event as we were getting our CNC machine shop installed and running. The CAM software vendor sends me the pre-processor for the Haas VF3-SS. I loaded it and programmed a simple square pocket for a test.
First thing the damn pre-processor does is a rapid traverse in x and y to the center of the pocket, WITHOUT raising the tool to safe "flight altitude". I was lucky in that I was using a vacuum chuck and a block of machinable wax for these tests, otherwise it could have been really ugly (broken tool and potential collision of the spindle with clamps leading to torquing the column out of vertical).
Fun stuff.
As a programmer who's used nearly every computer programming language from machine code, Forth, APL, C++, ObjectiveC and everything in-between I have always felt that G-Code is stuck somewhere in the Mesozoic era. Sure, it works, but one could certainly talk to these machines in far better ways.
This speaks to some of the low(-ish) hanging fruit in manufacturing that software can help with. I think we're closer to having the LaserWriter of 3D printing than the Pagemaker.
It's a huge understatement to say that CAD/CAM software sucks. Start by chipping away at small, solvable but labor-intensive parts, in the vein of Stripe.
Make tools that provide constraints that reflect the real world, in the hands of the designer. How often do you see beautiful design concepts that have zero draft or unmoldable overhangs? If those are things that are irrelevant with additive printing, then render the preview in no greater resolution than what the printer can produce, and identify the grain along which your part will break.
This will lead us to what my partner calls augmented fabrication. He prototyped Part Preview, a "Print Preview" for milling machines, projecting toolpaths onto your material or using a Kinect to intersect the blank with the model (unfortunately, no documentation except for http://www.metropolismag.com/pov/20111010/lab-report-2 and a thesis paper).
We need tighter integration between the software used to design objects and the hardware that manufactures them. Protomold is a step in the right direction. Because they control much of the process, they can tell you how your uploaded design needs to change to be manufacturable. That's an API for manufacturing - it digitizes something about the physical process. And it's a great educational tool. It's all about reducing the length of the feedback loop so that we can make mistakes and learn faster.
This is a really good list. Another thing to consider is that one seemingly simple change request, let's say from user interface (adding a button), can affect almost every category listed. And hopefully you haven't tooled up yet, because there goes $50,000+.
On the other hand, there is definitely some fun to be had in hardware for the low volume budget hacker. We have arduinos, raspberry pi's, all kinds of components available at outlets like digikey and mouser, relatively cheap custom pcb manufacturing, and somewhat low cost high resolution 3D printed custom enclosures.
Take the task of manufacturing and bringing to market an electronic product. Here's an incomplete list of some of the tasks requiring attention:
The list is not complete of course, but it provides some idea of how complex it can be. The idea of reducing this to an API or solving it via 3D printing is nice, but, today, a fantasy.There are issues that are not obvious from simply reading this list. For example, entrenched systems and software that does not inter-operate. Disciplines that require massively different infrastructures (machine shop, plastic molding, packaging, PDB assembly, wire harness fabrication) and are sometimes islands in and of themselves.
Just looking at CNC machining is an example of how arcane things can be and how difficult it can be to change the way things work. CNC machine manufacturers use G-code (which is older than dirt). You'd think that by now this would be a smooth system. It isn't. Software vendors produce tools that require customization for each machine type and, even then, can produce g-code that is actually dangerous to run.
I had exactly this experience when I decided that we needed to have our own CNC machining in-house. We bought hundreds of thousands of dollars worth of (mostly) Haas equipment. All g-code was generated with a CAM tool that ran inside of Solidworks. I'll fast forward to the day that one of the machines --with a 20 Horsepower spindle-- happily churned aluminum with a one inch roughing bit like it was butter. It was amazing to watch, and it sounded horrible. The CAM program simply put out the wrong instructions because the engineers who created it in India decided to have their own creative interpretation of how G-code should work.
Anyhow, the point is that manufacturing is hard and complex. That's why you don't see Y-Combinator style "Here's $15K, go start a company" incubators doing real hardware. In the hardware world, outside of the trivial, $15K won't even cover the tooling.
Maybe in a few hundred years we'll have Star Trek style replicators. How cool would that be.