SIMPSON'S FIRST PRINT
![]()
Here is Simpson showing off his creations. I didn't print all at the same time. In fact, I won't be able to until I work on bed leveling issues.
![]()
Left: 30mm/s, .3 layer height, no standard bowden tweaks (The filament drive had a loose set screw.)
Middle: 30-90mm/s, .4 layer height, no standard bowden tweaks (This is my printrbot's configuration file.)
Right: 30mm/s, .25 layer height, bowden tweaks (I could push the speed up. With stiffer springs, I expect to be able to print at 100mm/s without diminished quality.)
![]()
You can see the near perfect print on the right. Not a striation looks out of place.
I still have a few issues to work out.
*The spring should probably be twice as stiff and the springs attaching geometry should be analyzed with math (Oh no, not math.) to make sure the spring is effective in the whole print envelope.
*Work out the math for quick calibration. I had to spend a lot of time on calibration. There is no reason I could not have a sheet of paper with a few dots on it and move the tip of the extruder to each dot and then enter in all the position data and have the computer figure out the exact geometry. Some of the distances that need to be measured are hard to measure to the accuracy needed in the traditional way.
*Come up with a good way to make it easy to string the pulleys and hard for the string to slip off. (The string never slipped off during a print but as I was calibrating I would occasionally have the printer go the wrong direction and then everything would get messed up.)
*I don't have firmware written yet. I tried to modify repetier and failed. (I spent 1 hour. Time is very tight right now.) In the interest of time, I wrote a super simple python program that transforms standard gcode to match my geometry.
Surprises:
*The x and y rigidity are amazing. There is just a little hysteresis in the z direction but nothing that really shows up in a print and it will disappear with stronger springs.
*The central hub doesn't really need to tight when you have two connections per arm.
*Homing by turning the motors off and letting the printer go to its apex under spring assist and then zeroing all the motors works amazingly. It is very repeatable.
*I can probably remove half of some of the double arms and keep the rigidity.
*Even a poorly calibrated machine can print if you can find a way to get past the first couple of layers.
*Long moves (30-40mm) don't bow much even when I don't break a move into a bunch of little segments. All the long moves are inside the part and it bows so the highest point is in the middle of a move. The way it is now, there is little need to segmentize unless I am print perimeters with straight lines. My test squirrel had no straight lines.
*Ignore the fact that I didn't fully tighten the set screw, my new filament drive worked great. I is a very compact unit. It is designed to sit on the table and to pull filament from its side and push it out the top. It isn't much bigger than the stepper itself. I will be refining it and publishing it on Thingiverse soon.
*Assuming this printer isn't perfectly made, it is over constrained. That is usually a bad thing but in this case it uses the flexibility of the system to average between the fighting constraints. This allows a system with low rigidity to have a rigid effector.
*The machines prints get better with time. It literally works out its own kinks.
Crazy Ideas:
*You could flip this upside down and get rid of the springs and add enough weight to the head so you can get repeatable positioning.
*You could get rid of most of the bearings, if not all. Close tolerances paired with bolts and stop nuts would probably be good enough. This design is very tolerant of slop.
For those of you that are following all my designs and redesigns, the other design I built with the simpler arms is going to get revisited. Now that I have a large 3D printer, I will be able to print out 10 inch arms that are optimized for torsional rigidity.
Gada Prize News: I almost pushed out a bunch documentation and parts for the current iteration of my printer. I realized that it doesn't really stand a chance as is and I don't want to proliferate a substandard design. I got the bright idea that I could ask for an extension since I have only been at this for 23 days. I just got news that I am cleared for a 15 day extension. Based on my current trajectory, I can do a lot in that time.
Here is Simpson showing off his creations. I didn't print all at the same time. In fact, I won't be able to until I work on bed leveling issues.
Left: 30mm/s, .3 layer height, no standard bowden tweaks (The filament drive had a loose set screw.)
Middle: 30-90mm/s, .4 layer height, no standard bowden tweaks (This is my printrbot's configuration file.)
Right: 30mm/s, .25 layer height, bowden tweaks (I could push the speed up. With stiffer springs, I expect to be able to print at 100mm/s without diminished quality.)
You can see the near perfect print on the right. Not a striation looks out of place.
I still have a few issues to work out.
*The spring should probably be twice as stiff and the springs attaching geometry should be analyzed with math (Oh no, not math.) to make sure the spring is effective in the whole print envelope.
*Work out the math for quick calibration. I had to spend a lot of time on calibration. There is no reason I could not have a sheet of paper with a few dots on it and move the tip of the extruder to each dot and then enter in all the position data and have the computer figure out the exact geometry. Some of the distances that need to be measured are hard to measure to the accuracy needed in the traditional way.
*Come up with a good way to make it easy to string the pulleys and hard for the string to slip off. (The string never slipped off during a print but as I was calibrating I would occasionally have the printer go the wrong direction and then everything would get messed up.)
*I don't have firmware written yet. I tried to modify repetier and failed. (I spent 1 hour. Time is very tight right now.) In the interest of time, I wrote a super simple python program that transforms standard gcode to match my geometry.
Surprises:
*The x and y rigidity are amazing. There is just a little hysteresis in the z direction but nothing that really shows up in a print and it will disappear with stronger springs.
*The central hub doesn't really need to tight when you have two connections per arm.
*Homing by turning the motors off and letting the printer go to its apex under spring assist and then zeroing all the motors works amazingly. It is very repeatable.
*I can probably remove half of some of the double arms and keep the rigidity.
*Even a poorly calibrated machine can print if you can find a way to get past the first couple of layers.
*Long moves (30-40mm) don't bow much even when I don't break a move into a bunch of little segments. All the long moves are inside the part and it bows so the highest point is in the middle of a move. The way it is now, there is little need to segmentize unless I am print perimeters with straight lines. My test squirrel had no straight lines.
*Ignore the fact that I didn't fully tighten the set screw, my new filament drive worked great. I is a very compact unit. It is designed to sit on the table and to pull filament from its side and push it out the top. It isn't much bigger than the stepper itself. I will be refining it and publishing it on Thingiverse soon.
*Assuming this printer isn't perfectly made, it is over constrained. That is usually a bad thing but in this case it uses the flexibility of the system to average between the fighting constraints. This allows a system with low rigidity to have a rigid effector.
*The machines prints get better with time. It literally works out its own kinks.
Crazy Ideas:
*You could flip this upside down and get rid of the springs and add enough weight to the head so you can get repeatable positioning.
*You could get rid of most of the bearings, if not all. Close tolerances paired with bolts and stop nuts would probably be good enough. This design is very tolerant of slop.
For those of you that are following all my designs and redesigns, the other design I built with the simpler arms is going to get revisited. Now that I have a large 3D printer, I will be able to print out 10 inch arms that are optimized for torsional rigidity.
Gada Prize News: I almost pushed out a bunch documentation and parts for the current iteration of my printer. I realized that it doesn't really stand a chance as is and I don't want to proliferate a substandard design. I got the bright idea that I could ask for an extension since I have only been at this for 23 days. I just got news that I am cleared for a 15 day extension. Based on my current trajectory, I can do a lot in that time.