Hi, Does anyone know how to rotate 30 degrees in Magic?

The "M" in magic stands for "Manhattan," which means horizontal and vertical. No matter how you look at Magic, it will resist non-Manhattan geometries. That said, there is a "splitpaint" command that goes halfway toward arbitrary angles. However, some design rules are limited to certain angles (metal layers have fewest restrictions).

Thank you for your response. the command is like splitpaint dir layer [layer2] What layer2 means?

The command paints the selected rectangle on both sides of a diagonal, painting the "dir" corner with layer and the rest with layer2. For example, select a 3 x 4 area and type "splitpaint sw comment m1" (see image in this message with the result). This will fill the sw corner with black (comment) and the ne corner with m1. It looks like I was wrong in a previous message. The comment is drawn with a 30 degree angle, but the m1 gives a DRC error saying only 45 degree angles are allowed (maybe other metals have different rules). (If you don't specify layer2, magic will not paint that portion of the rectangle at all and you end up with a polygon.) You can do some pretty interesting stuff with the split commands, but you can spend hours trying to figure it out.

Most foundry processes (like SkyWater, for example) do not allow arbitrary angles because the design rules don't permit it. All layers in SkyWater are restricted to 45 degrees. Even if you can do arbitrary angles, endpoints must be on the manufacturing grid, which prevents you from, say, taking a cell and rotating it 45 degrees. Magic's use of the "split tile" allows non-orthogonal angles but forces all points of polygonal shapes to be on-grid. The only way to produce an off-grid point in magic is a shape at an angle overlaid on another cell with a shape at the opposite angle. Then, the point where the angles intersect could be off-grid, and this will be flagged as a DRC error by magic.

Tim: Your sentence with "sadly" stuck with me and I got the idea to explore the boundaries of the design envelope with the GF180 pdk (based hpretl's Docker from a month ago). I was able to use splitpaint to create a wire with a bend of a weird angle and a transistor with poly and diffusion as 45 degree wires (so the FET was at angled at 45 degrees -- see image). I messed with this for some time and I do not think there were any DRC errors. The extraction also looked OK. When I tried to make an annular transistor and I got DRC error that I did not understand ("only 45 degrees allowed on transistors" -- which is confusing: what about 0 and 90 degrees?), but did not clearly indicate that such a transistor was unconstructable. My overall point is that I cannot find documentation on the limits of the "design envelope" and using the tool gives surprises. Empirical evidence suggest that Sky130 is more limited that GF180.

@Erik DeBenedictis: There are ESD devices in the I/O that have angled edges on the poly. This cannot be considered a manufacturing design rule error, per se. The issue is going to be the characterization of any such device. The width and length are irrational numbers, not multiples of the manufacturing grid. But I'd be quite surprised if Magic came up with anything even remotely close to the correct device width and length (but you said the extraction "looked OK", so maybe it does? There are, however, restrictions on bends in transistors. The layout you show above might or might not pass the Calibre deck DRC; apparently it passes magic's DRC because it doesn't contain an actual bend (maybe?). But you can't make an annular transistor without a bend in the transistor, so it definitely violates the sky130 DRC rules (as does the ESD transistor, I think, which has its own rule exception somewhere). But this is a case where the DRC rules were written to force designers to use only characterized devices (there are no characterized annular devices, at least not that I know of). Some people have submitted designs with annular devices on them on the open MPW runs, and accept the risks---they are creating new devices and intend to characterize them themselves. We have established a rule with SkyWater that SkyWater will only reject one of our reticles if a "manufacturing rule" is broken. That leaves a bunch of rules which everyone is strongly encouraged to follow if they don't have a good understanding of device physics and why they might want to break the rule; but those rules can be broken by anyone who knows exactly what they're doing and why.