A redesign of the GUS Arm for easy external stringing

[ekaggrat]

got the older arm verison up and running for now on marlin... i didnt get segmentize still correct on the second run and the cube has curved walls.. i will need to redo the segmentize and then later maybe go to the new arm design and smoothie..





attachment=0]86156571730588231.jpg[/attachment]

477645446220746298.jpg (98.89 KiB) Viewed 48964 times

676239795814143331.jpg (54.14 KiB) Viewed 48964 times

[Jfriesen]

Looks awesome! The cube looks pretty smooth. I just got done printing my calibration cube as well. Attached an image from my webcam (taken via remote desktop) so I'll measure it tomorrow and report back. I already had a half printed cube when I noticed something that needed adjusting on my printer, and that one measured 74.7 x 74.7 mm so I was off by about .3mm it looked like. I keep getting some very subtle curves along the edges of straight parts, having trouble getting to the bottom of what the cause is.

By the way I've made a few more minor improvements to the CAD in various places. I'll be doing another commit to the repository sometime this week once I clean everything up. The biggest change is replacing the elliptical endstops with screws. Much easier to get minor adjustments in my opinion. I've reprinted my arms and I love the change.

[ekaggrat]

cool . my cube is smooth but exists in a wraped space .. i will redo the calibration soon and reprint... for now it is on hold . also i suspect the print orientation of the arms makes a difference in accuracy of the printer.. i am currently working on a corexy with all metal parts to ensure accuracy. once i get it done i will print the new arms and recheck the accuracy.


i think more the versions built the better it will get

[Jfriesen]

I couldn't agree more. I tossed all my old arm design prints today because they were taking up too much space. I must have had about 20 of them. I think I am finally converging on a printer which produces usable parts. Many of the base pieces of the printer are now printed by itself now haha.

[Jfriesen]

Alright, a quick update for you guys. After a host of testing I'm still not happy with the current performance of the printer. Thus I am in the process of doing a revision 2.5. The 2 major changes I am looking at are replacing the pulleys and changing the gear profile. I think the scaled up STL gear profile I used is not accurate enough and is a major contributer to inaccuracies in the print.

To address this I have taken the time to CAD up my own herringbone involute gear profile with a 45 degree pressure angle. The profile is parameterized so I can easily change the number of teeth pitch and pressure angle by just altering three parameters. I'm in the process of printing and testing a 40 tooth 125mm pitch gear and am going to test how accurate I am at holding center to center distance. I'll report back with results soon.

Additionally I have found an ideal replacement candidate for the 623v pulleys. They can be purchased through amazon:

http://www.amazon.com/FORWORD-10pcs-Pri ... ge_o00_s00

I have bought 50 of these and their quality isn't even comparable to the 623v pulleys because they are just so much smoother with zero wobble. At $15 for 10 they are a bit more pricey but well worth the cost vs. performance benefits here. I just strung up an arm with them and it moves like butter. I'd say by hand it has about 25% of the friction resistance as compared to the 623v arms when I try to back-drive them. Their outer and inner diameter is about 1mm larger than the 623v but they are the same width (slightly thinner really as the 623vs have are wider than spec). This means we will need to use m4 screws as pulley axles as well so 'll have to size up the holes appropriately.

A big github commit should be coming very soon once I finalize and clean all the changes. Keep an eye out!

[ekaggrat]

why use herringbone instead of the cycloid gears like the original one? the cycloid gears have in theory zero friction.

Did the errors start with the new arms or did they exist on the older ones also.. one thing i have noticed with my other projects is that gears printed in xy orientation kind of tend to have the tooth profiles shrink on the corners. So the printed gears will be a bit squished in profile. Maybe that could be the difference between the old and the new.

i will re calibrate mine and check it again.. i have a idea for calibration. I will build a probe in the place of the head. Then upload a Cartesian firmware. finally using a python script command the arms using transformed coordinates in real time to probe points at various locations . Using these points then do a leastsq to get the true shoulder arm and max z values...

i dont know how well i would work till i try it..

[imrahil]

Additionally I have found an ideal replacement candidate for the 623v pulleys. They can be purchased through amazon:

http://www.amazon.com/FORWORD-10pcs-Pri ... ge_o00_s00

these bearings are available on Aliexpress too - http://www.aliexpress.com/af/ugroove-4* ... ove+4*13*4

is it possible to buy as cheap as possible u-groove bearings and replace inner bearings with proper ones (e.g. 4x9x4 ABEC-5 2RS)?

BTW: love your design - I'm waiting for final-final version to reprint my oryginal GUS

[Jfriesen]

Hmm didn't realize the original were cycloid. I will investigate this thanks for the tip!

[Jfriesen]

OK, did some research and I think I agree that a cycloid gear profile makes sense here.

Just to be more clear, what was used on gus was the intersection of an epicycloid and a hypocycloid. Descriptions and equations are listed here:

http://mathworld.wolfram.com/Hypocycloid.html
http://mathworld.wolfram.com/Epicycloid.html

I was able to CAD up a gear using these parametric equations in solidworks. I think one of the biggest problems previously was that I used the same gear profile used on the regular GUS arms but I scaled the gear to be much larger. A normal GUS arm was designed with a 30 tooth gear profile. I scaled the arm by 125% so I think I am going to use a 40 tooth profile to get back closer to the original tooth size without having a bunch of messy numbers.

I attached an image of the sample profile I plan on testing today. Would love feedback!

[Jfriesen]

Alright guys, I finished my redesign cleaned it up and pushed to the repository:

https://github.com/Jfriesen222/GUS_CAD_Files

I still have yet to print and test this new design so use at your own risk for now. I feel pretty confident about the changes but if you wait a few days I can print and confirm functionality. Results will be posted soon, Enjoy!