Hello all, My name is Kevin Harrington and I am working with a team

Hello all, My name is Kevin Harrington and I am working with a team to develop a whole new 3d printer architecture. We have replaced stepper motors with closed loop control, eliminated ground rod, linear bearings and belts, and got our controls below $10/axis. This design should have far fewer vitamins then any previous design.

Have you looked at the Kossel? It’s also a T-slot based delta, and they’re using Spectra fishing line as the belting. It works pretty well actually.

The Kossel is a Rostock derivative. We do not use any linear slides of any type. All joints on the Delta-Forge are rotary using 8mm bearings. We are trying to minimize the number and cost of the vitamins kit.

Ah. It’s very hard to tell that from your blog. Most of the entries are so very zoomed into the platform, and you don’t show off the upper end at all. After you mentioned not using linear slides, I found the older post showing the whole bot. Very nice. :slight_smile:

I’ve been hearing about the fact that you use continuous rotation servos instead of steppers for this project – is that a price or function choice?

Price mostly, we wanted to simplify the control electronics by having a single pin low current way to add more axis. Servos come in around $4.5 in bulk and have a gear box, bearings, and the high current circuitry built in. I also like that they are a standard form factor allowing for a lot of expansion with better hardware without changing designs.

We also chose to use a feature of the metal geared servo, the horn connector, as our pinch wheel. Our ServoStruder contains no hobed rod nor gear. We even were able to use a printed opposing wheel where the magnet is. As a result, we detect material slip on the fly and the control loop compensated by bumping the motor speed.

IWe made or own encoding using a 12 bit hall effect encoder and talking to it over SPI. Our cost for axis is around $10-20 depending on volume. We get 12 bit precision out of it, and use a gear train to get that up by 16x. This gives us a theoretical 0.005 degree accuracy translating to less the ~.1mm anywhere in the workspace, a bit better in some regions, and a bit worse, but reachable “outside the workspace” since it is non rectangular.

We plan on releasing control kits for developers wanting to convert Cartesian designs to use the servo system. I think it has the potential to drop the cost of printers markedly. All source and open-hardware files will be released when the kickstarter ends, success or not :wink:

What does that mean: “fewer vitamins”?

Vitamins are pieces of the printer that aren’t printed. Threaded rods, motors, belts, etc.