Cherry Pi ball sockets: corner cubes?
Posted by dc42
|
Cherry Pi ball sockets: corner cubes? May 25, 2015 04:30AM |
At least two people have had problems using my differential IR probe on the Cherry Pi. The reason in both cases was that the effector is not staying in the same plane (i.e. parallel to the bed) as it translates. Since the IR sensor has to be offset from the nozzle a little, any tilt of the effector changes the relative heights of the sensor and the nozzle, leading to height errors.
Using the nozzle itself as the probe (e.g. with FSRs or with Andy's effector with integrated Z probe) avoids this particular issue; but if the effector is tilting then that is going to make it impossible to get a flat printing plane anyway. So the issue needs to be addressed.
I took a look at the Cherry Pi files on Thingyverse, and I see that it has printed spherical sockets to accept the balls. I can't see how you can ever get enough print accuracy to get the balls to be a good fit in the sockets with no slop - and to get the sockets in each pair exactly the same depth. I have seen the suggestion to use a ball-end mill, and I guess that might work if the mill is exactly the right size for the balls and the holes are milled very precisely - but that is likely to be difficult to achieve.
Has anyone tried a corner cube (i.e. 3 planes intersecting at right angles) for the socket, instead of spherical sockets? The corner cube would need to be oriented such that the ball is held in positive contact with all 3 planes at all normal angles of the rods. I'm thinking of one plane inclined at perhaps 5 degrees to the horizontal, falling towards the centre axis of the machine. This would allow for the rods being vertical or nearly so when the effector is close to a tower, with the rod angle falling to 10 degrees from the horizontal when the effector is opposite a tower. Then two more planes at right angles to that and below it on the carriages, above it on the effector.
It's possible to have the planes intersect at angles sharper than 90 degrees, if 90 degrees doesn't provide enough movement range while retaining positive contact with all 3 planes. For example, imagine two planes intersecting at about 80 degrees in a vertical line. Then put the third plane across the top, angled at about 10 degrees to the horizontal. Like this:
I guess another way of improving the sockets would be to use a CNC machine and a ball-end mill (exactly the same size as the balls) to mill the sockets in a block of PTFE, which is then screwed to the effector or the carriage. But that makes it less of a DIY design.
Edited 1 time(s). Last edit at 05/25/2015 07:26AM by dc42.
Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod
Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].
Using the nozzle itself as the probe (e.g. with FSRs or with Andy's effector with integrated Z probe) avoids this particular issue; but if the effector is tilting then that is going to make it impossible to get a flat printing plane anyway. So the issue needs to be addressed.
I took a look at the Cherry Pi files on Thingyverse, and I see that it has printed spherical sockets to accept the balls. I can't see how you can ever get enough print accuracy to get the balls to be a good fit in the sockets with no slop - and to get the sockets in each pair exactly the same depth. I have seen the suggestion to use a ball-end mill, and I guess that might work if the mill is exactly the right size for the balls and the holes are milled very precisely - but that is likely to be difficult to achieve.
Has anyone tried a corner cube (i.e. 3 planes intersecting at right angles) for the socket, instead of spherical sockets? The corner cube would need to be oriented such that the ball is held in positive contact with all 3 planes at all normal angles of the rods. I'm thinking of one plane inclined at perhaps 5 degrees to the horizontal, falling towards the centre axis of the machine. This would allow for the rods being vertical or nearly so when the effector is close to a tower, with the rod angle falling to 10 degrees from the horizontal when the effector is opposite a tower. Then two more planes at right angles to that and below it on the carriages, above it on the effector.
It's possible to have the planes intersect at angles sharper than 90 degrees, if 90 degrees doesn't provide enough movement range while retaining positive contact with all 3 planes. For example, imagine two planes intersecting at about 80 degrees in a vertical line. Then put the third plane across the top, angled at about 10 degrees to the horizontal. Like this:
I guess another way of improving the sockets would be to use a CNC machine and a ball-end mill (exactly the same size as the balls) to mill the sockets in a block of PTFE, which is then screwed to the effector or the carriage. But that makes it less of a DIY design.
Edited 1 time(s). Last edit at 05/25/2015 07:26AM by dc42.
Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod
Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].
|
Re: Cherry Pi ball sockets: corner cubes? May 25, 2015 06:54AM |
My only concern with using three planes in a printed socket is wear.
For kicks and giggles, I just looked up the cost of PTFE and graphite sheets to see if it was practical to CNC an entire replacement effector plate and carriages. Unless someone can find a kitchen cutting board or some other domestic product that uses one of these materials to use as raw stock, the material probably costs more than the machining labor.....
For kicks and giggles, I just looked up the cost of PTFE and graphite sheets to see if it was practical to CNC an entire replacement effector plate and carriages. Unless someone can find a kitchen cutting board or some other domestic product that uses one of these materials to use as raw stock, the material probably costs more than the machining labor.....
|
Re: Cherry Pi ball sockets: corner cubes? May 25, 2015 07:10AM |
@dc42
I abandoned magentic / spring loaded ball stud system like below image because of the issues you described. And I've designed effector and carriages.
I designed the ball seat that has flat top. The ball stud sits on top and sinks by given angle of seat module.
OpenSCAD code
90 degree
45 degree
30 degree
@vreihen
In my case PLA adaptor of diagonal rod will wear out.
Regards,
Hughe
I abandoned magentic / spring loaded ball stud system like below image because of the issues you described. And I've designed effector and carriages.
I designed the ball seat that has flat top. The ball stud sits on top and sinks by given angle of seat module.
OpenSCAD code
ball_diameter = 10;
ball_radius = ball_diameter/2;
theta = 90;
sunk_height = ball_radius * sin(theta);
seat_radius = ball_radius * sin(theta);
wall_thickness = 2;
seat_height = 4;
$fn =50;
module ball_seat()
{
difference()
{
cylinder(h=seat_height, r1=ball_radius+wall_thickness, r2=seat_radius+wall_thickness, center=true);
cylinder(h=seat_height+2, r=seat_radius+0.1, center=true);
}
}
translate([0, 0, (ball_radius-sunk_height)/2+ball_radius/2])
# sphere(r=ball_radius, center=true);
ball_seat();
translate([0, 20, 0]) rotate([0, 45, 0]) ball_seat();
90 degree
45 degree
30 degree
@vreihen
In my case PLA adaptor of diagonal rod will wear out.
Regards,
Hughe
|
Re: Cherry Pi ball sockets: corner cubes? May 26, 2015 09:53AM |
This is what I might have a play with...
The red bit is an oilite bearing like this one, set 3mm into the printed part (yellow) - although I'd probably try to find a cheaper source! The hole in the printed part would have to be a snug fit as these bearings are impregnated with oil and almost certainly ungluable. I could make something in phosphor bronze or steel that could be glued, but I'd rather buy something off the shelf if possible. They may well need to be sunk deeper into the printed part to avoid wobbling. I'm not sure what the ideal inside diameter would be - this one is 6mm, but maybe it should be a little larger.
I shall have to experiment on a test piece.
Edited 2 time(s). Last edit at 05/26/2015 11:58AM by David J.
The red bit is an oilite bearing like this one, set 3mm into the printed part (yellow) - although I'd probably try to find a cheaper source! The hole in the printed part would have to be a snug fit as these bearings are impregnated with oil and almost certainly ungluable. I could make something in phosphor bronze or steel that could be glued, but I'd rather buy something off the shelf if possible. They may well need to be sunk deeper into the printed part to avoid wobbling. I'm not sure what the ideal inside diameter would be - this one is 6mm, but maybe it should be a little larger.
I shall have to experiment on a test piece.
Edited 2 time(s). Last edit at 05/26/2015 11:58AM by David J.
Sorry, only registered users may post in this forum.