Hey everyone,
I yet live. And have not stopped working on this, but I ran into unusual problems with some yet undiscovered solutions. After spending _quite_ some time re-modelling the main finger cluster to be a) more adjustable and b) easier to (dis)assemble, at least that part appears to be somewhat acceptable now.
Here are some modifications I believe to be improvements:
1) I eventually ditched the vertical, sliding magnet layout for the central key; it seemed to be prone for getting stuck and sometimes unrecoverably so (even via filing) and have now implemented a solution I thought impossible at first:
By moving the magnet into a top cover plate and shaping the key itself somewhat like a hook, the magnets now pull up rather than to the side. The cover can be printed with different horizontal magnet displacements, thus greatly affecting upwards strength of this key. I have experimented with displacements up to 0.7 mm, which produces an almost ethereal key - still without getting stuck. Down to <=0.2 mm displacement the keys get rather harsh to press on the long run.
The top cover plate is held down by 2x4 1/16" magnets. All magnets are recoverable unless glued.
2) The central key is now located in a removable tube, which experiences the most frictional wear (apart from the key itself) and can be filed/replaced individually.
3) To assure cleaner sides I split the key itself into first two, later three parts, printing the central shaft horizontally to use the print sheet's perfectly flat surface. The cap part is designed around similar ideas. Since the shaft has to be pressed into the cap, that part needs to have a bottom opening, which creates an overhang if not split, which causes a ton of problems. Hence, this part is split as well. The two cap parts are actually designed to clip, which works pretty well at least when printing with a 0.25mm nozzle. Although I had the same idea to combine the cap and shaft, the best I could achieve was a sticking. Of the 8 keys I made 7 stuck just fine, one had to be glued eventually.
4) Since I introduced the top plate to handle the central key mechanic, this also increased cluster height, unfortunately. That, in turn meant the keys had to be longer, without leaning that much further than their smaller counterparts. This, in turn, meant they had to increase thickness with height, however not all to the way to the top, thus making the old way of printing very bad. I eventually split the keys into two parts each. The version I still mostly use (and which can be seen in the above schematic) requires glue. I have since developed a key part setup which should clip, but at least with a 0.4mm nozzle it's more like a sticking, requiring some initial pressure to hold. I have not used them in practice; I do not know if they hold for longer periods of time.
5) Growing tired of fiddling magnets into the holes, I eventually got the idea of making the magnet holes removable in the form of a ring-like shape. It is printed separately and can be removed and reused again in a different cluster, if necessary. The shape makes installing magnets heck of a lot easier. And, if nothing else, the magnets can be recovered easily. Since it is held by the top plate, it need not be glued or magnetized to the cluster.
6) The whole cluster is designed to fit to a PCB board, taking care of the local logic. After some mishaps and redesigns on that front, it seems to be working reliably now. There are openings for the connector and resistor network. I replaced the original 8 pin resistor network with a 6 pin one, which fits well into the width of the cluster. Out of the latest 8 clusters none have shown problems or had to be re-soldered.
Much of the above modifications are designed around reducing the risk of re-soldering anything. Once soldered, all force adjustments and replacements can be made with reprinting just one or two parts. The soldered parts remain unchanged.
7) Everything is designed in Fusion 360 now, and adjustments can be made using parameters. It's quite nifty but I think Fusion is at its limits now. I managed to produce a few freezes and crashes. Still, this is worlds apart from the software I used before.
The consequence of all the (assumed) improvements means a cluster now has heck of a lot more parts:
And assembled:
(I added shoes to increase traction of the metal feet).
Still fits to the original shafts and screws though. Nothing has been changed on that front (yet).
So, with that part covered, my attention moved back to the hand rest which needed some serious work. After many, many modifications and re-prints, I am currently at a four-part layout:
Basically, the bottom part (bottom-left most, white), allows altering the inclination (along both axes) and base height of the rest without reprinting everything else. I have managed to make it adjustable via user parameters without having to fix the whole history afterwards.
The part is held by magnets, and to get to those with pliers I had to remove the central filling, which reduced traction and had to be added again with a removable inlay part (next to the right, black). When installed, it is pressed against the metal ground, ideally using a rubber surface in-between.
The center part is mostly still the same, although I keep modifying it anyways, mostly to get more space for the key clusters next to it. The top part is also continuously being adjusted to fit my hand. The main reason for all of this is that my hands are not steady, and need to be at perfect rest. Otherwise they shake and I press keys I do not want to press.
While the solution, as is, should work for most people (although with some adjustments I guess), I am currently working on some really crazy stuff to keep the fingers afloat without impairing their movement. I can adjust key strength to keep my fingers at rest but then the keys are too strong on the long run. If I make them weaker, eventually my fingers can no longer rest on them. But this, I think, is more of a personal problem most people don't have.
I dubbed this thing 'Po's Paws'. The visual similarity wasn't actually planned
I think I'll just upload this stuff as is. I'm just not looking forward to documenting it properly. I'll keep you posted.
