All those firmwares work fine, or even better, over USB. Of course, there’s also the option to simply buy a kit. No idea if these people are legit, but the tech itself looks simple enough, a circuit board with contacts that let the linkage make a connection.
I am not quite sure I’d be ready to recommend it, but your more adventurous patrons may want to experiment. These keycaps are PBT, a cousin of polyester. They are not particularly pleasant smelling when heated or especially when burned, but they’re not as unhealthy as ABS (the other common plastic for keycaps) and certainly not as bad as the straight up poison gas that comes from PVC. I use a basic 5W blue diode laser, coat the keycap with an “infusible ink” pen from Cricut (most of their infusible products are polyester-based), put it in an alignment jig, then laser a raster image “low and slow.” My particular laser seems to do best when I do two or three passes at 2% power and 45mm/minute. The idea is to heat it roughly in line with the crafting heat presses without letting the heat spread and color in areas beyond the beam. I experimented with actually burning or engraving, and that sort of works, but (1) it’s stinky, and (2) the ash wipes away and you’re left with a mostly colorless letter-shaped indentation. The “dye sub” technique produces barely any fumes at all. There are a few people on youtube who’ve tried similar techniques, and quite a few who have used different heat or dye sources.
Aesthetically, the process was only marginally successful, though I’m optimistic about the longevity, at least compared to other low-end manufacturing techniques. I’ve been using a similar set of home-lasered keycaps for about a month with little to no wear. My jig was not as good on that set, AND I tried to center the keycap legends, meaning every fraction of a millimeter was painfully obvious. These legends didn’t end up exactly where I might have liked either, but they’re all off by the exact amount (about 1mm), so being consistent, the alignment isn’t too bad.
It should be doable. The way these things are wired, you wouldn’t use a common. You’d instead wire a matrix with diodes to avoid ghost presses. The firmware on the arduino or RPi microcontrollers will constantly scan for keypresses. So much would depend on the exact mechanism of your typewriter, but you could find a place where a keycap moves parts in close enough proximity to make your own switches, or if some part of the mechanism presses straight down, you could just have that actuate mechanical keyboard switches.
For wireless, you’d probably want ZMK. QMK is the most famous, but ZMK supports more wireless MCUs. I use KMK, a firmware where everything is human-readable python. I understand it has some wireless support, but I’ve never looked into it.
Doing it on the keyboard could very easily work, but I have a laser-cut jig that holds twenty-six 1u keycaps, and has one open-ended spot for anything up to maybe 3u. I considered 3D-printing the jig, but 30 minutes on the laser made more sense than 5 hours on the printer.
This plastic didn’t love my “Infusible ink” pens, so the legends are duller than the last time I did this, but the jig helped a lot with alignment, as did adjusting my ambitions and expectations. Much less disappointing to land 1mm off when you can pretend you wanted it there all along, and that is much easier to do with corner legends versus centered. :-)
Thanks. Kind of a 2004 Apple and Logitech thing going. Probably need another couple of years before it’s in vogue. white filament on the printer though, so white keyboard it is.
putting another tab button on the numpad
Horizontal spreadsheet data entry, my friend. it’s probably one of the less crazy parts of this layout! Of course, ten minutes on the laser and/or 5 in the software, and I can make it any key you need. Programmable keyboards are a godsend when you’re winging it on the layout.
At my rather beginner level, designing single parts for a 3D printer or laser engraver, it behaves almost exactly like most other parametric-history CAD apps in the broad concepts. The devil is just in the details, really. Shortcuts are different, terminology is different, Certain QoL and UI elements are either missing or somewhere else. The workbench model is not unique, but some of the kruft that has built up around FreeCAD’s benches and the defaults (better in recent versions if you look at the start screen) can make a new user “nope out” if they have other options. I guess assemblies in particular remain a fragmented area and lag behind the commercial packages, and I can say for certain that it still requires “good design practices” in a way that some of the commercial apps manage around, toponaming the biggest among them.
If all the negatives kill your workflow to the point that you want to pay for commercial software or live with the limitations (current and potential) of their free tiers, then that’s absolutely understandable. Commercially, it’s doubly so, and with addition of the “business reality” that there’s also no one to blame or sue if FreeCAD is not working for you. Hell, I don’t use it for all my stuff either, as I find no-history modeling still mostly works for what I’m doing and I have some free or cheap options in that space that are decent, but I can see the appeal as I’m starting to make things that could benefit from tweaks after the fact. What I get frustrated by is claims that FreeCAD “is no good” or “will never be useful”. I call BS. It’s already good and useful for many use cases, and anyway the number of free parametric CAD suites that do not restrict your use of your designs is exactly ONE. Otherwise, you’re looking at an absolute minimum of $300 a year to subscribe and hope that Shapr3D’s new history functionality doesn’t break, and that neither they nor Alibre gets gobbled up.
It’s getting better, and it’s critical that it do so, if for no other reason than to raise the floor that commercial offerings have to surpass to retain small-to-medium customers. I haven’t committed to it, but I’m rooting for it and following it closely.
White filament and a sharp pocket knife hide many sins, but thanks. It came out quite okay, and the cheap hardboard from Home Depot cuts very nicely with even a cheap laser.
It’s actually not secured yet. I included a couple of hexagonal openings in the long legs of the print that should accommodate a heated brass standoff just about right; it’ll be very similar to your alignment pins. I used the idea on my last keyboard and it worked well. The gap will be closed a bit more after securing the screws from the bottom plate. It will be nowhere near “perfect,” and but the end result should be a bit cleaner than this in-progress state. I am fairly pleased how decorative grooves draw the eye away from the seam a bit, but I’m not really trying to hide that it is 3D printed, and that in two pieces.
Tree supports and careful use of an exacto knife could also make it work. Post processing can be a bit much sometimes, but it doesn’t have to be the enemy either.
BLENDER’S NOT CAD! ;-)
I mean, depending on one’s definition, either it is or it isn’t, and with CAD extensions I imagine it’s perfectly fine for 3D printing where everything ends up a mesh anyway, but pretty much everything I’ve ever made was better suited to a solid modeler. I downloaded it once, but there was a lot going on there, and since it wasn’t my priority to learn to sculpt, I set it aside.
When is the last time you tried? 0.21 is better, and the new Ondsel build based on 0.22 is even a little bit better still. The skills (and mine are, frankly, below average) from Fusion360 should translate; under the shiny pull arrows, they’re obviously very similar. It’s all the same concepts, and constraints are less critical than they first appear for making one-off parts.
Then, if you don’t share my concerns with OnShape, it should work very well in Linux and after hitting F11 you might just forget it’s browser-based.
I understand the 3D experience includes some sort of browser-based riff on SW; maybe that would be a halfway useful tool for Linux. I dunno. :-)
There is a roadmap and the adults in the room are getting the Devs to eat their broccoli and fix long standing issues - looking at you topological naming issue.
To be fair, you can’t really call out part-time devs working for free too harshly, but I do think there was a focus on what they wanted to be able to pull off, versus what they wanted to empower random idiots with an Ender 3 to do, much less small-to-medium businesses. I think it led to the infamous “FreeCAD way” attitude amongst devs and power users. Topological naming, for instance. That arises out of the software implementation of 3D kernels, making it an issue of proper use of inherently limited design software, not of design itself. Software that accommodates it well is is not “cheating,” and it’s not lazy design to rely on well-coded apps, but I’ve run across that sentiment before, if not the words.
The push from Ondsel (for as long a it lasts… as the CTO mentioned, they have a very narrow path to support the underlying project while finding sufficient monetization), and the structure provided by the nonprofit should help. Autodesk too seems to be doing their part by making all but the most rank beginners (and some of them too) second guess the choice to lock themselves into Fusion 360.
They seem to offer a very generous scheme and seem like they are really in it to make a product they can be proud of. It seems like it’s a really good direct modeler, especially with the grasshopper addon. I’m a couple of decades too late for an educational license though.
Yeah, that is sort of what I’ve been noticing, and also with my experience initially wrapping my head around constraint-based drafting. I think the thing I’m still seeing though is that FreeCAD fails less gracefully than some of the commercial packages if you do things that are difficult to constrain, and in brief experimenting it seemed to be less willing to autoconstrain while drafting, particularly with easily changed dimensional constraints. I still feel comfortable saying it’s a bit “persnickety,” but agree that it’s not the hell I used to think it was.
Does the OpenSCAD workbench count?! 🤣
For straight up ease of use, maybe try the current Direct Modeling version of Shapr3D (i,e, not the Beta). The free tier is useless for printing anything, but it’s a great way to see if the workflow fits you. DS Mechanical, limited as it is, is also free and pretty usable if you have literally no import/export needs from other software, and its STLs are fine to print. Both are a little more intuitive for non-CAD people. Oddly enough, the big thing missing in the FOSS space is a decent direct modeler_. It’s kind of a shame, too, since so much hobbyist use has minimal need for strict constraints, collaboration tools, or parametric history.
Now, because high quality direct modelers are a bit niche, and often use sketches anyway, it makes sense to get a basic grasp on sketch-based CAD, whether fully parametric or not. The paradigm that was so different for me to get used to is that idea of “sketch and extrude”. You don’t just plop down a cube or sphere.
Instead, you go into 2D mode, select a “workplane” which (you often start with the generic XY plane), and you draw the cross section of your solid. Then you use extrude or pull or pad or whatever the app calls it; this adds the third dimension to make it a solid. If you wanted a cube, draw a square of n mm, then extrude to the same height. Oh crap! That needed to be a rectangle? In a parametric modeler, go back to the sketch and change it, then (if needed) do the same for the extrusion’s height. For various flavors of direct modelers, you can pull that face, or click the sides to change the size, just do it over, or add the extra volume and boolean them together. The next big concept is placing workplanes. There’s typically some button that pulls up a tool that lets you do your sketch directly on a face of the solids you already have (or on an “arbitrary” plane placed exactly in line with a face… NO TOPOLOGICAL ISSUE HERE!) Once you do THAT, you can extrude down into the old solid to cut stuff out of it (sometimes this will need to be done with booleans, but often the apps have a “remove removed” set of buttons or are smart enough to guess what you meant), or extrude out from the old solid to glom stuff onto it. Fillets and chamfers, when supported, usually work by selecting one or more edges and telling the app to calculate them away at a certain radius/distance.
In the end I feel you on wanting to like freecad, I’d really rather use a Foss solution for my personal work.
I just downloaded The Ondsel “flavor” of FreeCAD. It’s based on the upcoming 0.22 release, and it includes an addon that integrates to their tiered PDM, but everything else is still completely free. I don’t see anything mind-blowing, but it’s a very nice update. UI improvements, mostly on the Part Design workbench, nice and legible dark theme, floating tree, better launch page, and floating user-enterable dimensions when sketching (pretty cool, and long overdue). I question the viability of their business model in the timeframe their VC’s likely want, but for now they’re plowing some amount of money into FreeCAD development.
the mouse controls are a bit whacky to me
Is it true that SW ties everything to the Middle button with various modifier keys? And doesn’t let you change anything? I’ve used a LOT of different 3D software over the last 45 days or so, and honestly that would be (if true) among the wackiest I’ve run across. FreeCAD has 10 preset styles now, and one or more of them might work for you.
Solidworks Maker
Here is the cheapest deal I’ve found on it (it’s well known, I didn’t have to do any detective work or anything). $38 for a year is probably worth it just for the removeds and giggles, but as I mentioned, there’s just no reasonable path for the home-business that manages to pull in $3-5k of profit. Lock-in is the bread and butter for all the companies, big and small, and a big part of this exercise has been to see just what I’d be locking myself into.
My very first “design” was cord wrangler that fit the exact number of things I needed to charge and was the exact thickness for my no-name pressboard nightstand with a gap that matched an opening in the back. It was exactly what I needed.