Welcome! Log In Create A New Profile

Advanced

Hangprinter version 3

Posted by tobben 
Re: Hangprinter version 3
August 07, 2018 11:19AM
"Slack line can run across a load cell undetected"

Could this be solved by weights? The line would never get totally slack, because weights are pulling and tighten the line?
Re: Hangprinter version 3
August 07, 2018 11:55AM
I do agree that it is preferable to correct for line flex, and to measure and use tension data for that. See for example this blog post for some Hangprinter related line flex compensation work. I think it contains something close to the model you're proposing.

Two following two scenarios made me think it's not easy:

Tight/Slack Lines Due to Mis-calibration
All installed Hangprinters will have some mis-calibration that will make lines go slack or over tight when moving far away from the origin. (We can make mis-calibration very small, but we can't rely in it being zero.) So the machine will experience slack lines and over tight lines during operation. Take the situation of slack lines. If you tighten one line, all the others will also tighten. How to decide which axis to tighten in order to wind in the slack?

Upon second thought, we could just use a simple rule, like "tighten all axes by an equal amount until we have the tension we want in all lines". We could use the equations in the linked blog post to calculate the wanted tensions. The simple rule would compensate an unknown error with a known error. This would be ok. It would give us a confidence interval for accuracy and enable the constant radius spool to work smiling smiley

Obstacles, Tripping In Lines and Other Unexpected Forces
See from second 22 in this video for the behaviour we want when the nozzle hits something. If the firmware blindly tries to enforce a pre-calculated tension, the nozzle would just tent to stop when meeting an obstacle.

A way to mitigate would be to detect that the nozzle has met an obstacle. When nozzle meets obstacle, one/two of the ABC lines get tighter and the rest gets slacker. Upon detecting this, we could make the firmware force the nozzle across the obstacle with greater force, while keeping the rest of the lines tight. I guess line tripping is a different story, but I'm sure we could detect it and find reasonable custom behaviour.

So yup, measuring and using tension data would be possible and make the Hangprinter better =) It would require some work, but all development is work... tongue sticking out smiley The tension feedback loop must work really well to properly avoid any slipping ever around the effective spool, but maybe it's easier than I imagine and turns out to just work?

Edited 1 time(s). Last edit at 08/07/2018 12:10PM by tobben.


blog
Re: Hangprinter version 3
August 07, 2018 12:03PM
Weights, or any kind of spring load system, along the lines will start jumping around and make good print quality very hard to achieve. Even the little suspended 623 vgroove bearings I used back in 2015 vibrated enough to tattoo their eigenfrequency into every printed surface. Pic of suspended bearings here. Also see this comment about keeping a constant torque on one axis.


blog
Re: Hangprinter version 3
August 07, 2018 04:46PM
smiling smiley
One quick thought....

"Upon second thought, we could just use a simple rule, like "tighten all axes by an equal amount until we have the tension we want in all lines". We could use the equations in the linked blog post to calculate the wanted tensions. The simple rule would compensate an unknown error with a known error. This would be ok. It would give us a confidence interval for accuracy and enable the constant radius spool to work" smiling smiley

And if we fix the mover to the 0,0,0 position while tightening we could calculate position from tension.

Edited 1 time(s). Last edit at 08/07/2018 04:47PM by Dahah.
Re: Hangprinter version 3
August 07, 2018 05:09PM
What if the effective spool would have a spiral shape? The friction would be much higher by the longer distance.
Re: Hangprinter version 3
August 07, 2018 07:04PM
" Also see this comment about keeping a constant torque on one axis".

Klipper seems very promising. Will the Raspberry perform well enough for complex code of a tension based hangprinter? Better than Arduino for shure, but maybe one should think of something like the UP squared ( [up-board.org] ). That could keep everything in place. No additional PC for sim etc. in the best case? Like a HAW hardware + printer control. So one could set up and use the hangprinter eg. with a mobile phone or with a monitor and keyboard/mouse.

It's maybe too enthusiastic but........more computing power would open up future possibilities for cameras, tracking systems (e.g. pixy cam [pixycam.com]) and other sensors to enable for example printing on a given unknown complex surface.

Edited 7 time(s). Last edit at 08/07/2018 08:27PM by Dahah.
Re: Hangprinter version 3
August 07, 2018 07:40PM

This is just a quick concept drawing.
If one would use tubular spools and move the idlers very far away the effective radius would be guaranteed, if the angle is right.
No compact top plate! sad smiley
Maybe with aluminium or carbon rods one can build a sun shaped top plate with the D spool in the middle. But it would be huge in xy size. sad smiley
Or loose idlers on the sealing: Difficult to measure....

Edited 6 time(s). Last edit at 08/07/2018 08:04PM by Dahah.
Re: Hangprinter version 3
August 07, 2018 08:15PM
Obstacles, Tripping In Lines and Other Unexpected Forces
Sounds good!

How often do those collisions happen? Why does it hit so hard? Whats the reason for the wrong nozzle position?

Edited 1 time(s). Last edit at 08/07/2018 08:16PM by Dahah.
Re: Hangprinter version 3
August 07, 2018 08:44PM
From Vitana: Some flex comes from line modulus, but adding infinitely stiff lines alone won't solve our problems because ceilings, floors, ceiling unit sheets and plastic parts also flex. Very stiff lines are also thick, building more on the spools and making the imperfection of our buildup compensation more severe

Is it easier to use a thicker line to clamp/rubber wheel it on the constant effective radius spool or is a thin line better?
VDX
Re: Hangprinter version 3
August 08, 2018 03:36AM
... have you already read this old thread from 2007?

Maybe the image in the first post can give you some ideas too:




Viktor
Re: Hangprinter version 3
August 08, 2018 05:33AM
If the effective spool is a straight cylinder, then we will have "cable walk" like explained by a lecturer here.

One way to keep the lines enter/exit points still is to use a threaded rod as the effective spool, and let this threaded rod move up and down through a nut when it is turned. The same lecturer talks about helix shaped grooves here.

Calculating position from tension would require extremely low friction in all idler bearings. Again, hard but I'm sure it's possible.

Yes, I think the Rpi has enough cream to do tension compensation. But if I used Klipper I could just run it on one of my other computers, no need for the Pi.

These collisions happen 0-100 times per layer. It doesn't hit very hard. If we naïvely enforce a fixed line tension, then an obstacle can force the mover/effector to a full halt because more tension is needed to get across the obstacle.

A thicker line will grip better around the effective spool. Be careful though. Your compensation idler might have to compensate many cm of line if the collector spool is small and the line is thick. Attaching an example that shows how much the current firmware would compensate if line was 0.75 mm thick, and anchors were 3 m from the origin along x, y, and z. All relevant parameters are found in the source file.


blog
Attachments:
open | download - compensation_plot2.svg (20 KB)
open | download - compensation_plot2.gp (3.1 KB)
Re: Hangprinter version 3
August 09, 2018 02:14PM
Very good lecture. Thanks for the link!
Grooved or ungrooved cylindrical spools and idler positions that enable line walk seem too complicated and the compact top plate is such a good feature of the hangprinterV3 design. I will keep in mind that cylindrical spools are a very good way to have a constant effective radius.

I will make a prototype of the constant effective radius spool v4 for the 0.39 line. It's time to gain experience!

Edited 1 time(s). Last edit at 08/09/2018 02:14PM by Dahah.
Re: Hangprinter version 3
August 12, 2018 06:08PM
Just a quick idea in memory of a hangprinter failure: Cables clocked in gears.
What if we use gears instaed of a round constant effective radius spool and rubber wheels? Of course this would stress the cable incredible hard. Would gears provide enough friction? Maybe rubber coated gears?

Edited 3 time(s). Last edit at 08/12/2018 06:11PM by Dahah.
Re: Hangprinter version 3
August 13, 2018 03:52PM
The other day, I had a similar idea. If it would be possible somehow to bite into the line. Like to force it through two gears. Obviously that would deform the line and you would see the pattern in the print. And the line wouldn't last long. But then I thought about a belt drive with a smaller pinion attached to the stepper motor driving a bigger gear, like in this quick sketch.

The line would be guided to enter between the belt and the gear at one side and it would exit on the other side. The exit line would then be spooled up on another spool like Dahah suggested.
Now the trick would be the deep grooves on the big gear which would guide the line and prevent it to be squished. The belt would barely touch the line, just enough to build up the needed friction.
I'm still not sure if this would work. Most likely the belt would wear out quick and then you would lose the grip again...


Re: Hangprinter version 3
August 13, 2018 07:26PM
If the cable gets just friction from the gears teeth and doesn't bend in zigzag pattern it acts like a round spool with less contact area. An alternative could be a wave-like pattern and simular rubber wheels. I mean an extruded rounded star (extremly rounded) or a rounded gear and one or more rounded star shape rubber wheels that hit the cable in the valleys (inner points of the star.) If the curve of the valleys and the hills (outer points of the star) is soft enough to not brake the cable it could maybe - it is a guess or a hope again - allow more friction compared to the round wheel because of the angles between two points on the hills and one point in the valley when the rounded star shape rubber wheel hits the line. I will think about it and maybe I will make a drawing after I finished the first prototype.

At this point the V4 seems the best design to test first. Right?
I found one possible and solvable problem in V4: The friction that is provided by the rubber wheel is in relation to the force the motor has to move to turn the spool. Is this true?

A difficulty is the very thin line. That would mean we need to have a groove in the effective spool with a flat surface area of 0.39 mm. The rubber wheel needs a flat surface area with 0.39 as well. Very thin for rubber and pressure. Rubber bends easily.... So for the first prototype I will use a larger flat surface area. It will be not as precise as a 0,39 surface area but easier to realize and it shows if it is possible to generate enough friction this way. Later prototypes can focus on precision. First version will be a proof of concept.

Edited 5 time(s). Last edit at 08/13/2018 07:55PM by Dahah.
Sorry, only registered users may post in this forum.

Click here to login