Why do my aluminum parts always come out with visible tool marks on the face cuts?

The diagonal swirls on a face pass are almost always the stepover, not the feed. If your finish pass is stepping over most of the tool diameter, you get that scalloped, swirly look because each pass overlaps the last one only slightly and the tool deflects a little at the edges. On a hobby size machine the spindle and gantry have just enough flex to make it worse.

What worked for me on similar parts was dropping the finish stepover way down, somewhere around 10 to 20 percent of the cutter diameter, and taking a very light last pass, maybe half a millimetre or less. You were right to suspect running the finish at the same depth as roughing. A heavy finish cut loads the tool and the machine, and the surface suffers. Light cut, fine stepover, and keep the same feed you would expect for that depth. Slowing the feed way down can actually make marks worse on aluminium because the tool rubs instead of cutting.

Two other things worth checking. Climb milling on the finish pass usually gives a cleaner face than conventional, and if your CAM lets you choose, spiral or morphed toolpaths on a flat face come out nicer than plain raster in my experiance. Also make sure the tool is sharp. A two flute carbide that has done alot of aluminium will start smearing the surface before it looks obviously dull.

As for sanding, yes, plenty of people do it and there is no shame in it. A quick pass with fine wet and dry on a flat block, followed by a scotchbrite pad, evens out any remaining marks and gives a nice brushed look. On a functional part I would not even bother unless it bothers you. For the next one, get the CAM settings right and you may not need to touch it.

So to answer your stepover question directly, start around a tenth of the diameter for finishing and see how it looks. It adds a bit of machine time but on a small plate it is minutes, not hours.

I agree the finish settings are critical, but on a smaller hobby rig, the machine flex is often the real culprit for chatter. When I switched from working on thick aluminum blocks to thin sheet material, I had to stop worrying about the G-code and focus entirely on how the part was held. I used a simple scrap piece of 2020 extrusions as a fixture, just to sandwich the aluminum between two clamped pieces of steel. That small step stabilized the gantry just enough to make the tool run much cleaner, even with a standard finish pass. It stopped the micro-vibrations from making the surface look like it was textured with sandpaper.

I tried changing how deep the cutting tool goes on the pass, instead of just changing the path settings. Like, I went from cutting only a little bit off the surface to pushing the cutter through much deeper into the block. It seemed like the bigger cuts made the metal behave a little better when the cutter was really loaded down. I wondered if running a deeper cut on the final pass helps the finish look better too? Or does the depth of cut always have to be super shallow for a clean look?

I have always noticed that even when the stepover is sufficiently minimized, the cumulative effect of inadequate chip evacuation can still degrade the resulting surface finish because the aluminum tends to pack heavily against the cutter flank. This leads me to question if adjusting the feed rate in smaller increments, rather than simply reducing it, might allow the tool to shear the material more aggressively when the cutting forces are highest. Furthermore, when implementing a climb milling strategy on the finish pass, meticulous attention to the tool entry and exit points is arguably as critical for preventing micro-chatter as maintaining a consistent path thru the bulk of the material. I wonder if experimenting with varying the feed while maintaining that tighter stepover whould yield a more distinct improvement in the visual result.

The issue might not be the pathing as much as the spindle itself. I used to run everything on a column machine and the tolerances were tight enough that if the spindle wasn't rock steady, you'd see vibration even with a perfect stepover. On these hobby rigs, everything moves a bit. Have you checked the coolant setup? Just a damp bar is fine, but if you're running dry, you're just smearing the aluminum. I always favored a good flood for a reason. Another thing is that even if the cutter is fine, a slight misalignment on the tool holder can ruin a whole surface run and give you a pattern that looks nothing like swirl marks.

So I tried reducing the stepover significantly, and the diagonal marks finally cleared up. However, the surface finish still doesn't look quite right. It looks a bit fuzzy, almost like chatter is still occurring, even with the path settings optimized.