When do you stop modding a cheap printer and just build the real thing?

I ran my old i3 clone for almost four years and it printed everything from PETG pet shelters to ABS enclosures, mostly stock with a new hotend and a belt tensioner I modeled myself. The whole time I told myself the same thing you're saying, that each fix taught me something. And honestly it did. I learned belt tension by ear, learned why my bed corners sagged at 60 degrees, learned that my z wobble was a bent lead screw and not my printer's fault for once. But here's the thing I only admitted later. The frame was the bottleneck and I knew it by year two. Every mod I bolted on after that point was working around the machine instead of improving it. When I finally got my Voron kit printed and built, the first benchy came out cleaner than anything the old one managed in four years, and I realized how much of my 'tuning' had really been compensating.

That said, I never got rid of the old one. It still lives in the corner and prints ABS parts for the Voron upgrades, since it handles the enclosure heat better than I expected. If yours is genuinely printing the brackets that improve itself, let it run that loop as long as it makes you happy. The kit will still be there when the frame finally gives up, and you'll appreciate it more for the wait.

I was in almost the same spot last year with an old Ender that had a printed shroud, a printed fan duct, and a printed drawer full of spare printed parts for the printed parts. What finally got me was realizing I'd spent three weekends in a row printing brackets and shims for a machine whose frame flexed every time the bed moved. The plastic parts weren't the upgrade path, they were the machine begging me to stop.

So I stopped patching and built a Trident instead. Used the old printer to print its own parts, which felt fitting, then let it retire once the new one did its first benchy. Took me about two months of slow evenings, and yeah, the wiring was the part I underestimated, not the printing. Here's my take on where the line sits: if you're printing structural parts for a machine that has to be accurate to print them, you're chasing your own tail. A cable guide is fine. A new carriage or a motion upgrade on a bent frame is just polishing somethig that can't get better.

Nothing wrong with the sentimental route either, the guy above clearly got years of joy from his. But I don't think you're rehearsing anymore. A rehearsal you keep extending stops being a rehearsal.

Seconding the thing about working around the machine instead of improving it. My breaking point was spending an evening trammed the gantry, then watching a printed bracket sag on the plate mid print and having to redo the whole alignment. That was the moment it clicked that I was fixing symptoms. The old machine still runs as a spare, but everything structural stopped making sense after that.

Gonna push back on the idea that a cleaner first print on the new kit proves the old machine was all compensation. Of course the fresh one printed better, it had new belts, fresh rails, and every nut torqued that week. That's not the frame being the bottleneck, that's new parts beating worn ones, which would happen on any printer. I ran a squared-up bedslinger next to my kit for a good while and side by side prints came out closer than people here would expect. If the frame is rigid and the bed doesn't rack, most of what you're paying for in a kit is time and novelty, not capability.

The 'printing its own upgrades' loop is how half of us started anyway, back before anyone had heard of Klipper. Mine was a delta held togather with zip ties and prayers, calibrating by folding a sheet of paper under the nozzle and hoping. Honestly the machine tells you when it's done. Mine told me when the printed carriage started shearing screws mid print. If yours still runs true after a month of sitting, keep going. Mine didn't, and that was the whole answer.

Quick update: I finally figured out what the big bottleneck was. It wasn't the overall structure, but the lack of rigidity in the gantry assembly. That idea that I was fixing symptoms rather than the root cause really stuck with me. Swapping out the existing linear rails for upgraded ones was the real turning point.