One of the earliest solutions I saw was the Air Struder by Jassper. Later, Trick Laser debuted the Fly-N-Strude. These are both adequate for running a single extruder. I wanted three (for a Diamond hot end) or three plus one up top for support material (for a Kraken), so earlier this year I designed an arm-mounting system for Trick Laser arms:
This worked fairly well, but it didn't let the extruder bracket swivel, so it was always pointed in the same direction. As a result, it worked very well for small prints, but the tube would have to bend a lot on larger prints. That added friction - sometimes a lot of friction! After using it for a few months, I decided to design a triple-extruder flying platform:
This works a lot better, but is still flawed. One thing it does much better, but not nearly ideal, is lower bowden tube flexion than the arm mounts. The more bent the tube is, the more friction it adds to the filament path, wrecking retraction performance and making it that much harder for the extruder to push the right amount of filament at the right time. When the effector is dead center, there's no flexion (or very little); but the further we move from the center in the XY plane, the more the platform wants to "fall over." Another issue: the extrusion steppers will tend to "stand on" the hot end's effector platform, transferring their weight down to it through both the filament and the Bowden tube. As with flexion, the weight exerted varies with the XY position of the effector. I have the paracord and bowden tube lengths balanced so that it's just perfect when centered, but as soon as it moves off center, it will exhibit this pushing effect.
In practice, this doesn't hurt accuracy that much - on brand-new 300mm Trick Laser arms. I can get a nice calibration on my printer with an average Z error of ~20 microns, which I think is excellent. However, the differential loading is likely to make the joints wear out faster. I think we all prefer to have printers where the arms aren't considered a wear item!
One solution would be to use pairs of paracord "arms" rather than a single "arm" that lets it flop around so much, but the kinematics still won't keep the extruder platform centered perfectly above the effector. If the paracord comes from the same elevation as the delta arms, the cord must necessarily be shorter than the arms in order to maintain the extruder platform above the effector. That changes the equation of where the platform ends up: it never moves away from center as fast as the effector because its "arms" don't allow it to go as far. If the effector is 100mm out, the extruder platform is perhaps 85mm out, for example. It can't help but lean over in that configuration.
Then, last night, it hit me: Take the existing carriages, arms, and effector, and make a "copy" of them that would go higher up on the belts. To visualize this, look at the above picture and imagine a second set of carriages a few inches above the first, and delta arms projecting from the upper carriages to the triple-extruder platform. The platform would be modified such that it would have joints at the same position as the effector, on the curved white crossmembers that connect the extruder brackets. Spacing between the upper and lower carriages would be achieved with a printed part that would be fastened to both of them.
Advantages:
- You'd get the same performance as you would with a direct-drive extruder, or very close to it. Some people have abandoned the delta platform altogether over this issue.
- Same-length arms and identical joint positions on extruder and effector platforms make the kinematics for both identical. The extruder platform is GUARANTEED to be exactly above the effector.
- No bending of the Bowden tubes.
- No differential loading exerted on the effector by the extruder platform.
- Requires relatively little PTFE tubing.
- No careful tuning of the length of the paracord is required.
- Drivetrain exerts positive control over the massive stepper platform at all times. It will not bounce around, or risk coming to rest on one of the arms when the effector is close to the print radius.
- The belts are under constant tension, which (maybe?) helps with accuracy.
- It looks freaking cool.
- It costs more money to do it this way.
- You lose a few inches of print height (~3-4?) because the hot end carriages have to be lower on the belts than they were before, although this could be overcome with a sufficiently clever design (see below).
- As currently designed, the extruder platform is wide enough that it could interfere with the lower (effector) arms when the effector is close to one of the towers. It seems possible to redesign it to be more "concave," but you still might lose an inch or so of print radius.
- No room for a fourth extruder. One could be mounted up top, or somewhere else. It wouldn't get the benefit of the super-short Bowden tubes. I figure you'd probably use it for support material, where perfect feed is less important.
- Requires a counterweight system so that the hot end doesn't dive into the glass at 15MPH when you turn the printer off. (This is not a hard problem - I'm working on it now.)
- Heavy-duty rod ends, something that could actually take 3-4lbs. of weight (divided by six of course), could be used, and the extruder platform could sit on top of a short tower that would bolt to the extruder platform. Three fewer carriages, six fewer arms, and no height lost, although it would make it significantly harder to keep the effector platform accurate. Tensioned ball arms (chromed steel on either end of the rods, riding in acetal cups on the effector and carriages, tensioned by paracord or similar) might be good enough. Another solution would be to use tiny NEMA 11 geared steppers, as suggested by bot. They wouldn't add as much weight. Unfortunately, the only ones I know of are sold by Warner Berry, and fully 3x as expensive as the TriDPrinting geared NEMA 17 solution. Then again, we would be saving money from not having to buy another six arms.
- Alternately, keep the two-sets-of-arms-solution, but push the carriage joints inwards 2" towards print center. In this way, they would totally clear the endstop switches. The lower (effector) joints would be at the same elevation as now, and the upper (extruder platform) joints would be mounted on a mast that projects upwards. The carriages and their masts would have to be engineered to withstand the cantilevered supports projecting upwards several inches. When the printer is homed, the upper joints would be several inches above the endstops!