I used these calibration pieces over the weekend, and also tried with a modification that has pegs and holes. Curiously, adding pegs and holes changed the calibration. See http://repstrapdk.blogspot.de/2018/01/lots-of-calibration-prints.html?m=1
@Lars_Clausen Nice post on your experiments. Did you put clearance gaps between your peg and hole features? That might account for your perimeter adjustments needing to be 75%. Make your hole a total of 0.5- 0.6mm wider than the peg in diameter (0.25-0.3mm on each side) and it should provide a perfect fit.
Try the second set of clips that I posted with the interlocking curves that simulate holes and pegs. On the interlocking clips with curves there is a 0.2mm gap between the mating surfaces for the straight parts (like the “C” clips) and a .25mm gap for the curves. You should get a snug but easy fit for both interchangeably when properly calibrated. For moving parts add an additional 0.05mm gap minimum.
Remember the parts do not have to fit flat. The first layer squish should not be included in the fit test. For this reason they are 3mm tall so you can have at least 2.80mm of fit area.
I didn’t add any clearance. I like my prints to come out the way they are defined. I’ll have to check some other prints, but it looks like the smaller perimeter setting really does make clearance unnecessary.
@Lars_Clausen In engineering design and manufacturing you need to provide clearance for parts to fit. There are always manufacturing tolerances you are dealing with so you will not get a perfect or even the same fit every time. The smaller perimeter does the same thing as adding clearance. For a 0.4mm nozzle you will reduce the size by 0.1mm at 75% which provides your 0.2mm minimum clearance. Either way will work so pick the one that works best for you. 