Okay, but like... this just worked? Why didn't I do this to start with? These just clicked together perfectly and the basic dimensions are already locked in ???

To be fair, I have the benefit of hindsight of months of developing the original version of these, and that to me is what being a #maker is all about - collecting knowledge and experience so that what takes you two hours the first time becomes five minutes the next. #3DPrinting

This was always my vision for #Thrixels - modular bases that clip together in a way that allows the tiles on top to just seamlessly... tile... to form a pattern. The tiles are very easy to locate and snap in, there's enough room in there for bigger patterns to not collide and warp apart, not a single individual tile popped off in repeated drop tests, and while a ton of dimensional tweaking obviously remains, all the core functionality is suddenly there.

This has been a good day.

#3DPrinting

This might be where I pause for today. Base spacing is correct, but I need to extend the joiners/edge supports to the corners, otherwise they bend when you press tiles onto them (and if I've found myself doing that others will too).

I'm also not happy with how tiles are ending up rotated in place slightly - I'm happy with registration and spacing, but alignment I'll have to think about.

PLA is very easy to print and tweak, but it also wears quickly, so that's another challenge.

#3DPrinting

I think the diagonal orientation lines are probably the best option here. The tiles don't look *terrible* if they're slightly rotated... until you put them next to a set of tiles that *are* perfectly straight, and suddenly it's night and day.

I don't love introducing chirality into the design of the tiles, but they are surprisingly easy to manipulate in hand and spin around to the right direction. I'll go with it and see if it ends up being more annoying than beneficial.

256 tiny (4x4x4mm!) individual PLA parts successfully printed on one PETG raft, with 100% successful yield. I've switched back to a generic orange PETG from the official #BambuLab green PETG which oozed terribly and seemed to have worse raft adhesion, which I wasn't expecting to be the case at all.

Today I'm tweaking some baseplate dimensions - the alignment diagonals could do with some embiggening, and I'm still tweaking the slots for the joiner/support pads underneath. #3DPrinting

Here's how the baseplates have turned out. Changing colours mid-print to highlight the direction of the diagonals is trivial, and the joiners and supports are working perfectly too.

This has made me realise I could do the same colour-swap trick with the tiles themselves - white for the underside, then the intended colour for the top. That would give me all the visual contrast I need to quickly orient and plug in each individual tile.

And here's what this version of the board looks like with some tiles on it. I made sure to assemble these separately, to simulate doing 10x10 boards on their own so tiles from neighbouring boards can't help with orientation, and....... I can't tell without looking at the base pieces where the seam between them actually is.

There is some bowing in the middle, but that's because the joiners are a couple of layers thick and are quite flexible - it wouldn't be visible in a completed, framed work.

So, clearance is a fun thing to get right in #3DPrinting. If you have a 10mm cube, and a 10mm hole for it to go into, it won't fit - 3D printing is just too variable a process for things to work like that. You'd need to build some clearance in - probably 0.25mm on each side, so you'd end up with a 9.5mm cube; my #Thrixels have a 0.1mm gap between the tiles and the part of the baseplate they grip to.

But stuff gets weird when you go small - like single digit multiples of your nozzle width small.

My problem the other day was that I'd made the slots in my base parts 1mm wide - which was fine - but the same width for the tabs on these joiners, which was not fine, because it was trying to fill a 1mm gap with 0.8mm of material, and there just wasn't enough there to keep a good grip in that width.

But squishing a 0.8mm tab into a 0.8mm slot will work - there's little enough material there that it's still malleable, yet enough that it fills enough of the space to get an interference fit.

It's a weird and tough problem to solve, and I'm not totally convinced I've worked out all the kinks, but this is a good enough system to move on with - which means the next step is to print hundreds of Thrixels two-point-oh, stick them on these bases, and see how they line up.

Time to attempt to print 512 parts in one go. YOLO, right?

I regret to announce I have invented voxel teeth.

In better news: That's 512 v2.0 #Thrixels successfully printed at once - not the most individual parts I've ever printed together, but not far off it, either.

These are at a 0.16mm layer height, while the originals needed to be at 0.12 to work reliably, so that + the simplified geometry makes for a much faster print. Four hours for ~30g of parts sounds slow, but that's still under 30 seconds per individual part!

Not only does this look like there is no seam at all, when you insist to your brain that there definitely is one, its best guess is a row or two away from where it actually is.

The sense of satisfaction I'm getting from this is on par with building a complicated gaming PC from scratch, and doing all the cable-tying before pressing the power button and finding it boots and works perfectly the first go. I am ridiculously happy with how this project is unfolding. #3Dprinting #Thrixels

Well, that's the 2x2 test done - totally seamless, but there is significant bowing, a lot more than I was expecting. I saw this effect with the Dopefish mosaic, and cutting down to 10x10 boards was intended to minimise it.

I think I can counter this by making the tiles narrower in the middle and taller, and only widening to the full 4.0mm wide on the topmost layer or two. There might also be another strategy I can take with the parts that join and support the boards from underneath.

And just like that, the bowing is gone - there might be a fraction of a millimetre or two in it, but that should be invisible in a finished piece.

These taller thrixels with the ridge around the top is 100% the way to go, they're so much easier to handle. I also had the idea to switch filament part way to make it easier to see each piece's orientation, same as I've done with the bases.

Next steps: More of the same, to retry validating being flat and seamless in a 2x2 grid. #3DPrinting