I saw your PM and gave a try this morning, but I’m afraid we are reaching the limit of Shapr’s capabilities (or mine’s ).
Your angle is much bigger than the one I used, so the curvature is also much bigger, and the program struggle to do it right.

I tried to revolve the concave surface instead of the convex (it is easier to compute the revolve I think) and it gives you a better shape, but it does not solve the wrong ending of the slot (you must complete the ending by move 4.5mm + rotate 160° the milling tool but it does not perform better).

I tried also to do several loft, with multiple 3D guides, but didn’t succeed to get the loft working, for the same reason of high curvature I guess.

So at present, I have no idea how to do it better. I will keep thinking about it in background, and will PM you if I find a better solution.

Sorry if this is OT, but I find it interesting to play around with…

The slot , made with a mill ( OD 2 mm end mill) would not have a trapez shaped slot in a section view, so this would be confusing. The only way to make it like made on a mill. A cylindrical shaped body must be swept throug the larger cylinder in order to “cut” out to the section.

But if you sweep a cylindrical pin you do get a trapezoid shape:

I simulated the mill (cylindrical shaped body) 16,8mm traveling 90° in a slope of 51° with a cylinder diam 100/50 and the resulting slot (radial) becomes a trapezoid 20,7x29,7 with slightly curved symetrical sides resembling something between a (half) ellipse 12,7x0,8 and an a catenary/parabolic arch in an angle of 2x10,17°. 16mm pin inserted.

But when it comes to real manufacturing, and I made this part pretty often, there is no trapezoid form and a manufacturer will have problems to do the programming on the cnc mill or he will be confused. I appreciate your support, unfortunately the pin inside the slot must rest on a perpendicular wall of the slot and not on an edge, and a end mill like i used make no trapezoid slot.

I think @BKE is right and @Chris07 experience is not wrong : the shape is not a perfect rectangle, but with the parameters used by @Chris07 it is so close to a rectangle than the difference is probably not noticeable : about 0.05mm.

Here is an animation using @Chris07 parameters, with 5° steps from 70° to 90°, with the cutting plane set at 80°.
At 70°, the mill machines the bottom left corner, at 80° it machines the top left+right edges, and at 90° it machines the bottom right corner.

The superposition of all the cuts gives the final shape, which is not a perfect rectangle but very close to it at the intersection with the tube.

I did this animation trying to understand the wrong cut ending, and I think I found a way to correct it, even though I don’t understand why, which is not totally satisfying. I will post the solution soon.
PEC

Long story short, for some reason I don’t understand, when the angle of the helix is a large number, the start angle of the half mill tool that I first revolved is no longer vertical but must be rotated by an empirical value.

The yellow (1) is the one used in my previous post, the green (2) is the one that gives the correct result.
The end part of the mill is unchanged.

As the rotation angle is determined by trial (thanks to parametric version :)) in order to have just one angle to change in the step named “corrected angle”, I add a new revolve step using the end face of the previous body and revolve back with a new angle.

I also use the concave face to do the tool’spath and move the initial shape to the end to finish the tool’s path.

But this 3D is possible to create in Shapr3D. (Though exact mill of this is complex as the mills movement has to be combined with an axial movement, otherwise not creating the correct symmetrical sphere.)