Yeah there were a bunch of things.
First, a worn nozzle. I did dimensional accuracy tests before I started, and everything looked good. It was only after I did several real world prints where pressure advance was doing work that I realized something was wrong. The wheel pieces didn’t fit together, no matter how much I sanded. I ended up throwing out like 4kg of parts. After replacing the nozzle, my extrusion multiplier went from around 1.015 to 0.98.
Second, a warped toolhead part. My printers use a printed toolhead mount. This mount can warp if the hotend isn’t cooled adequately, or if the screws are overtightened. This resulted in significant tilt of the toolhead, which knocked over supports and caused print failures. I had to re-print and rebuild the toolhead to correct.
Third, just profile tuning stuff. I found that large support gaps (~0.3mm) work well for removal, but some parts, like the wheels, don’t like that distance. The part of the wedge that gets printed only on supports curls up and eventually gets knocked off by the nozzle. This is solved by reducing the support distance, but that makes removal harder.
Fourth, the TPU. I tried for a while to figure out the stringing and blobbing. Best I could do are wispy hairs and scars on the surface where travels happen. I’ve resorted to printing objects sequentially. This is a decent compromise.
Fifth, the TPU again. I tried some really aggressive retraction, which helped with the blobbing and zitting, but it turns out that only works for low-retraction objects. For objects that have a fair amount of retraction in a short period, it pulls filament so far into the heatbreak that it solidifies and clogs.
So yeah, like you said, so many variables. But I was able to work through all of them. Now I’m working through more 
it never ends.
