I have run some controlled experiments and I do get a distortion on one corner of your test cube.
I have set up the cube:
- your way
2.point supports along the edge
- linear support along the edge (my preferred)
see attached .form fileBILL TEST 01 FLF otaolafr.form (397.1 KB)
Now the way the print works, the cube starts to print as a pyramid balancing on a single support, and this single support has to take all the loads until the next support is met.
Now for most people this does not really matter as geometrical tolerances are not that important for their needs.
My first solution worked better and this was not disernable distortion across the 6 faces when tested on a engineering square. What I dont like is the chip out on the edges. This is a higher density where it matters.
My second solution supports all three edges nearest to the print platform all the way (I just picked 0.4mm thickness as first guess - you can go smaller) my support structure is given further support by the Preform program. I use this technique a lot and there was no distortion. I clean up edges by scraping with a scalpel under a microscope and finish off with 2000 grit wet and dry resting on thick piece of float glass.
Now I did run this with clear at 25 micron layers and not with the HT, but it shows the same distortion, so it is a processing issue and not a material issue. The F2 is great but sometimes you have to work with its limitations when you are pushing the limits.
Now coming back to your original component. If it is essentially solid and not shelled out it will be relatively massive and with a 0.1mm layer thickness this distortion will be exacabated as there will be a little distortion on every layer.
Now printing flat on the build platform is an obvious solution but you have to allow for some layer distortion in the first few layers. What I tend to do is to add an extra say .75mm to the base then machine or sand off afterwards by 0.75mm.
I hope this helps. Let me know how you get on.