> The ten thousandth one costs $15
As a software engineer, I'm used to being able to build essentially whatever I want, with the only cost being my time. I'm often unpleasantly reminded that software is an outlier in that regard and as soon as physical items are concerned, especially hardware, it turns out that manufacturing even a simple thing is prohibitively expensive if it's done at a small scale.
But yeah, going from prototype to production with hardware is much, much harder, I agree with you.
[1] the project changed scope and we had to do a large refactoring of the codebase as a result, if it had been based on microservices the refactoring would have been an order of magnitude simpler. Luckily the monolith was relatively well architected with separated service providers so it wasn't super bad.
Though the thing that does stick out to me as a software engineer is that you need parts and thus can't iterate on your ideas nearly as quickly. You can't just pull in a library or something, you have to order a part and wait for it to physically get to you. And the cost of any mistake is in actual real money because wiring something incorrectly can destroy things.
That's the same in software engineering. You are losing your own time. If you are just breaking your prototypes and not the machines to build them, your time lost is probably multiple times more valuable than the material.
I guess, once you have billed a certain number of hours to clients, the relation between time and money just ingrains itself so deep inside your mind, that you might as well start telling the time in USD.
Don't you still work for time? If you have achieved something, you can do it again or something similar you might enjoy. Unless what you are doing is completely directionless and you are basically just randomly soldering stuff together, I would say there is still esentially the same loss function as with money.
I give you that the abstraction often makes you feel less pressure in the moment itself. But in the end it is really just an abstraction. Money is just a virtually constructed number we use to control people's time.
> If it's your own project, your mistakes cost you nothing but you learn something from every one of them.
You also learn form professional mistakes.
You would just need to 3D print a case and somehow cast your own silicone keypad. DIY electronics is actually not too bad cost wise.
I hand-placed some 0.4mm QFNs before (LVDS to TMDS converter IC) (admittedly, only a ~90% success rate) so this has to be at least twice as easy.
Even PCBA companies usually just place and reflow the chip and may only use x-ray to validate the initial build- and then rely on process consistance + functional testing at the end of line.
Woodworking scratches many of the same itches as programming for me, with one key difference: it’s unforgiving of mistakes, so requires a lot more care.
At least if you have a grinder, welding can sometimes be more forgiving.
Their pricing and supply times factored in a pretty significant failure rate, and we were basically asking them for highly precise screws and nuts for a couple thousand each.
But to check THAT tolerance, you have these metal screws that they need to securely attach or be unable to attach to. And since the metal fittings will be used for batch testing entire production runs of life or death parts for years to come, they go a little overboard with specifications.
I'm not sure if it really needed to be THAT precise, but that's just what it was
I'm not sure if this command will work? Well I may as well just try it anyway. Something broke? I'll throw in a breakpoint and try out a bunch of dummy variables and alternative methods until the errors stop.
And while it's a common complaint that compile times force some extra care, with hardware that extra care needs to happen outside the medium itself, mucking around with CAD software and such, before sending designs off to workshops. It'd be like if the solution to avoiding compile errors was very carefully handwriting code on paper
> (Western Canada, slang) beset with unfortunate circumstances that seem difficult or impossible to overcome; screwed.
https://en.wiktionary.org/wiki/hooped
I’d love to know its origin. I picked it up from family.
But working with wood a crippling anxiety hangs over me.
So the software equivalent would be if you oursourced a large chunk of your project to someone else, and in that case, I think $10,000 might not be too far off.
The short answer is yes you can, but it's a PITA. Among other problems you have to deal with shrinkage and drooping. I'd only use that approach for a quick and dirty prototype that just needed to hold parts together and survive rough handling.
A better option is to build a case from parts you cut out of a sheet of plastic (anywhere from 1mm on up) that you then glue together. Once built, rounding the edges will make it a lot nicer to hold, but it will still basically be a box, and the glues you use are generally pretty toxic so don't breath in the fumes OR get it on your hands.
More involved would be making the case by vacuum forming a heated sheet. The forms (one each for the top and bottom instead of the two you need for injection molding) are typically hand sculpted or machined from wood, filler and primer rather than metal, and vacuum forming is something a hobbyist can accomplish at home with a DIY setup. You can get a case nearly indistinguishable from an injection molded part, at least from the outside. The more complex the shape the more labor intensive the fit and finish will be (for example, getting the edges of the case halves to match up).
However you make the case, there would be a lot of keyboard holes you'll need to carve out with something like a dremel rotary tool. It might be easier to make the case with a single large hole that you glue a thinner insert into that will be easier to cut holes in with a hobby or utility knife, but I've sliced my hands up a lot when doing this sort of work. A rotary tool seems scarier if you're not already familiar, but with as little as 30m of practice you'll probably have much better control than you would over a knife and lower risk of injury (as long as you use eye protection, etc. You really don't want the plastic shavings flying around getting into your eyes).
Not a bad way to go, but I found that for hand assembly, I had a really hard time with the glue out of control; I probably needed a better applicator than the default tube of epoxy.
Instead, I would go with VHB (Very High Bond) tape. Also weird to use, if you are using the thin 1mm width, but I can position the tape more carefully before I press it into place.
- machine a form out of wax or delrin (or clay), pour a silicone mold, and then pour liquid plastic into the mold, either a not-yet-polymerized resin such as a thin epoxy, or a molten thermoplastic;
- similarly, but use plaster instead of silicone and burn out the form in a kiln before pouring (not applicable for clay);
- 3-D print the case, and if you don't like the layer lines, smooth them out with either a coat of paint or by smoothing an ABS print with acetone vapor;
- 3-D print a form for molding.
Even for injection molding, you might start with an aluminum mold, which won't last as long as hardened steel but will cost a lot less.
But I don't know anything about this stuff so maybe it's not applicable to this case?
Unless there are products I'm not aware of, mostly thermo plastics require high pressure injection that isn't feasible for a hobbyist setup.
> a not-yet-polymerized resin
There are lots of products like this (a bewildering variety, really), but in general they aren't particularly convenient to use, and you'll have a lot of failures. Factors that can mess you up include temperature, humidity, air bubbles (so many air bubbles!), mixing ratios, over mixing, under mixing, contaminates on the mold surfaces or your mixing containers, pour holes too small, not enough exit holes, entry/exit holes in the wrong places, molds held together too loosely leading to leaks, too tightly leading to distortions, etc.
Except for the possible exception of relatively simple shapes and small items (like dice), just about everything you can imagine going wrong will, and you'll end up with a lot of waste before you get something you can use, and disposing of the failed attempts responsibly is another challenge.
There is a caveat though: if you're trying to make a soft part (whether in a hard or soft mold), the materials are generally a lot more forgiving of errors, but that's tangential to the challenge of making a case.
As far as thermoplastic molding goes, aside from the vacuum forming that someone else suggested, there seem to be a lot of experimental options. There is a YouTuber who has built a couple of DIY injection molding machines (http://www.youtube.com/user/AndysMachines), and Star Simpson did successfully mold recycled HDPE plasticized with vegetable oil in an open-top mold (https://www.instructables.com/HomemadePlastic/), but the results are not super smooth. My recollection from melting plastic cups in the oven as a kid (probably polypropylene) is that you end up with a puddle with a pretty flat and smooth surface except for whatever lettering or imagery was on the outside of the cups, so I feel like you might be able to get a smooth top out of an open-top mold, but I don't know which plastic that was, and melt degradation might be an issue. https://youtu.be/KA1UELu33FI shows some apparently successful open-top molding with powder made from recycled polypropylene (skip to the last minute of the video) but doesn't test it for strength. https://www.youtube.com/watch?v=2OYbyzgErvo recycled polypropylene into a two-piece water-cooled mold improvised out of aluminum baking tins, but seems to have suffered major melt degradation possibly due to overheating it in a can on an open fire. https://youtu.be/Ae8_OWju8RM recycled (probably heavily filled) HDPE into a billet in an aluminum can in a thermostat-controlled electric oven at 180 degrees and had a lot of porosity due to high melt viscosity, then turned the billet down on a lathe; the outside of the billet seems to have suffered enough melt degradation to make it machine much more cleanly than the center, which produces the kind of long stringy chips you'd expect from HDPE. https://youtu.be/blSvUqfuiuo makes an HDPE sort of gingerbread man in an open-top silicone oven mold, the kind you'd use to make gingerbread-man-shaped muffins, suffering melt degradation severe enough in parts to turn the HDPE brown, warping, and significant porosity due to high melt viscosity. https://youtu.be/-igxhoGEQFU shows really nice marbled results melting HDPE in a panini press with a silicone mold release, then pressing it into a billet in a particleboard mold with a bench vise, but still suffering some problems with porosity. By turning it on a wood lathe they are immune to problems of warping etc.
Polycaprolactone deserves a mention because above its glass transition you can mold the thermoplastic with your bare hands, but pouring it into a mold is out of the question because of its high viscosity, and as a result of that low Tg you tend to get a lot of creep, and parts become unusably weak in a warm room.
So I think you've convinced me that thermoplastic casting without injection molding equipment is not a practical option for making low-volume parts. Thank you for sharing your experience!
It is much easier to get good results with a bit of in-person guidance, and in a controlled environment. There are also optional materials (eg. mold release) and equipment (pressure pot, vacuum chamber, or vibration table, all of which help with both mold-making as well as casting) that make success easier which are harder to justify for a low volume or one-off project.
> Which I think implies a cost of first success well under US$10000?
Well, sure. But you aren't comparing like with like. If you outsource the production of a prototype to someone else using resin casting, it will also probably be cheaper than $10k for the first one (say, $1-5k, because resin casting more or less requires you to have built one with some other method before you can even make molds) but scaling up is going to be more expensive (dozens or hundreds of $ per unit), in part because silicone molds wear out quickly and have to be replaced after being used as little as 5-10 times.
Conversely, if you already have your own machine shop, you may be able to produce your own metal injection molds for much less than $10k and hand it to someone to do a small run.
The two methods trade off (access to) expensive equipment for expensive materials, and up front cost for per unit cost at scale, but if you really only want one neither is attractive.
Still, it is really hard to beat vacuum forming for approachability in making a one off: The only equipment you don't have laying around can be made with a few chunks of wood. If you have a small oven (even a toaster oven) that only limits the size of your parts. If you have a vacuum cleaner with crappy suction, that just means you can only form thin sheets.
BTW, all these techniques combine. In the case of resin casting, I mentioned that you really need to have made a one-off in order to create the silicone molds. In fact, I have in the past vacuum formed a blank shape (this was before 3D printing was a thing), sanded or machined it to more precise tolerances, added details (dedents, screw posts or holes, flanges, ribs, ridges, textures, etc.) and then used silicone molds and resin casting to make duplicates. Sometimes you actually do need several identical parts in order to just make one object...
In some cases I think there are materials that are a lot easier to shape to make a mold pattern than whatever you're going to make your final part out of. Wood is the traditional choice, except that wax is the even more traditional choice, and clay, styrofoam, foamed waterglass, and pumice also come to mind. I guess in theory polycaprolactone might work, too, and as you say, poly(lactic acid) is a popular modern choice, sometimes painted or sanded to smooth out the print lines. And I've been wondering if it's feasible to strengthen aluminum foil by spray-painting or cathodic mineral deposition in order to make such one-off items, because aluminum foil is very easy to form, usually too easy. Lots of stuff to try!
There are a couple of things you can try:
Squeeze the glue (of whatever sort) onto a disposable surface (a scrap of foil generally works for me) and use a wire or whisker of some sort to pick up and apply small amounts in a controlled fashion.
Tack the parts together with minimal dots of cyanoacrylate, and use gobs of the real adhesive to put it a fillet on the inside corner.
You can also use a jig of some sort to position the parts you're gluing.
Solvents instead of an adhesive are a bit more forgiving of accidental over application (you just spill it off and wait a moment for the excess to dry).
> Instead, I would go with VHB (Very High Bond) tape. Also weird to use, if you are using the thin 1mm width, but I can position the tape more carefully before I press it into place.
An interesting suggestion, and not one I've tried. Thanks, I'll keep it in mind.
The F1C100s is $5 for one piece on AliExpress, all the parts seem to be solderable by hand/manual reflow in an hour (the passives can be populated by JLC for like $30 for a batch of 5 boards, saving your time), the board would be $2/piece for a batch of 5 at JLCPCB (seems to be 2-layer), the enclosure can also be 3D-printed on a professional printer at JLC for tens of $, the only thing I don't have experience with is the conductive-rubber keyboard.
Then if you want to actually make a business out of it, another $500K for Lawyers, overhead, advertising, manufacturing contracts, headaches and so on.
But, the computer is very cute, and nice work. I was halfway expecting YARPIP (Yet another Raspberry PI project)
Very cool it was all done from scratch. Done that myself for a tablet prototype and it is a shit ton of work.
You can overcome some of the scaling costs with 3D printers. It's pretty cheap to make fairly robust plastic parts with a 3D printer if you can do basic CAD.