I designed a polar printer

nicholas.seward Wrote:
-------------------------------------------------------
> Here is the picture of my design that has up to a
> one inch diameter dead zone in the middle of the
> platter. I love the simplicity. I don't love the
> hole. (This is a functionality drawing. It is
> not meant to reflect how I would actually build
> it.) I am scrapping this idea. I am still in my
> brainstorming phase.
>
> [forums.reprap.org]
> ilename=turn_table.png

I was working on a similar design, except I did not move the hot end except in the up and down. I moved the bed in both rotary movements instead. That way I did not have a hole, even with the 4 bar linkage for the hot end.

However, once I saw how Wally could come together, I put that design aside since Wally was simpler to make in a low cost RepRap way.

I am trying to visualize the bed being able to rotate in two ways. Is it like the feather printer that has a forum post in the general forum?

Nicholas, I was working on a design more like this one in the link, but much simpler and only a 3D printing head. Sort of a cross between this one and yours. I only had it sketched up in several versions on paper for brainstorming ideas. The key is that the X translation from the 4 bar linkage of the head in the Z has to be tangent to the polar motion of the partial rotational axis (not the turntable axis). Everything was to be built off a flat base plate.

Polar Bot You Tube

I really like that mill/print machine. It is becoming evident that we still have some crazy designs to bounce around. I will focus on Wally and Simpson for now but if you want to start any new threads, I would love to pile on.

Back to Unlimitedbacon. Can't wait to see this beast.

Man, I think you are crazy (in the good sense of the word), but I'm glad to see you believe in your design and you have commited yourself to make it work. It certainly will be a spectacular machine to see working with the 4 extruders.

Congrats!

Quick update:

This has morphed into a school project. Follow our progress here: [csulbmae472group7.weebly.com]
The github has moved here: [github.com]
Experimenting with some software: [github.com]

Looking Awesome!!

Edited 1 time(s). Last edit at 11/09/2013 02:59PM by Jo1212.

It prints!

Here's a video of our first test [youtu.be]

We tested the operation of a single arm. The object is a 20mm cube. It is deformed because the firmware hasn't been modified to do the coordinate conversion. This is the result [i.imgur.com] on the right and the same STL printed with my Makerbot on the left.

I like it! That print looks very good considering the firmware has not been modified. Do you just make an estimate for the steps/mm in the region you are printing, and then treat the axes as cartesian?

Correct. I figured the steps/mm based on the gear ratio and an 8cm radius. The farther you get from the center, the more cartesian-ish it acts.

New test! Two extruders working at once: [youtu.be]

Very cool!
Looks like you could rotate the table and print 2 colors on one part?

Very nice project, looks really cool! Have you already started modifying the firmware?

I would also like to build a polar printer, quite similar to the PiMaker and the R-360. Although they claim the projects to be open source, I could not find the firmware they are using. Does anyone know if they really published their firmware?

Quote
A2
Very cool!
Looks like you could rotate the table and print 2 colors on one part?

You mean like this? [youtu.be] [i.imgur.com]

Quote
Fabricate
Very nice project, looks really cool! Have you already started modifying the firmware?

I would also like to build a polar printer, quite similar to the PiMaker and the R-360. Although they claim the projects to be open source, I could not find the firmware they are using. Does anyone know if they really published their firmware?

I created a fork of Repetier here [github.com] . The plan is to rewrite the Delta printer parts to do the polar conversion, but I have to admit I really don't know what I'm doing. If you can track down the firmware for those other projects it would be really helpful.

You might be able to use some of ttsalo's code: [github.com]

He has modified Marlin firmware for parallel SCARA configuration, which is probably closer to your polar design than a Delta is.

wow awesomeness overflowing! Just wondering, what is the advantage other than faster printing of cylindrical shaped objects?

How far are you with firmware and bipolar adaptation?

The math is all worked out and documented here [github.com] . I also wrote a program in python to simulate the motions and help me understand how it will work [github.com] .

I've been playing with various firmwares, trying to wrap my head around them, but it turns out to be much more difficult to read other people's code than to write your own. Thanks, MattMoses. I looked over ttsalo's patch a little. Having an example to go by will make things much easier.

Quote
unlimitedbacon
The math is all worked out and documented here [github.com] . I also wrote a program in python to simulate the motions and help me understand how it will work [github.com] .

I've been playing with various firmwares, trying to wrap my head around them, but it turns out to be much more difficult to read other people's code than to write your own. Thanks, MattMoses. I looked over ttsalo's patch a little. Having an example to go by will make things much easier.

Don't know much about firmware myself, but next project is a bipolar printer, so when i have the hardware build I will try to port firmware for it. But kind of hoped that you have got some firmware working just a bit. ;-)

Quote
Kenzu
Don't know much about firmware myself, but next project is a bipolar printer, so when i have the hardware build I will try to port firmware for it. But kind of hoped that you have got some firmware working just a bit. ;-)

bipolar printer? You mean with two hotends?

I think (although I did not run any experiments) the big problem atm is that the arduino has not enough computing power. For this reason I am going to wait for the RAMPS-FD or RADDS boards to be available, as the Due seems to have a lot more power!

Calculating the polar coordinates from x and y positions should not be to complicated (http://en.wikipedia.org/wiki/Polar_coordinate_system#Converting_between_polar_and_Cartesian_coordinates), if the device is powerful and the libraries exist.

But I am not quite sure if we shouldn't apply specific slicing for the polar bots and have some new G-Code commands (and have some G-Code Preprocessors in the meantime). So instead of G92 X10 E90 you could say G92 R10 P36 E90. R, S, T, U, V, W could be 6 extruder positions, P is the angle in degrees and Z and E, F, H, I, J, K (G is not allowed here I think) are the extrusion rates for these extruders. This would allow to have multiple extruders working simultaneously.

Quote
Fabricate

Quote
Kenzu
Don't know much about firmware myself, but next project is a bipolar printer, so when i have the hardware build I will try to port firmware for it. But kind of hoped that you have got some firmware working just a bit. ;-)

bipolar printer? You mean with two hotends?

I think (although I did not run any experiments) the big problem atm is that the arduino has not enough computing power. For this reason I am going to wait for the RAMPS-FD or RADDS boards to be available, as the Due seems to have a lot more power!

Calculating the polar coordinates from x and y positions should not be to complicated (http://en.wikipedia.org/wiki/Polar_coordinate_system#Converting_between_polar_and_Cartesian_coordinates), if the device is powerful and the libraries exist.

But I am not quite sure if we shouldn't apply specific slicing for the polar bots and have some new G-Code commands (and have some G-Code Preprocessors in the meantime). So instead of G92 X10 E90 you could say G92 R10 P36 E90. R, S, T, U, V, W could be 6 extruder positions, P is the angle in degrees and Z and E, F, H, I, J, K (G is not allowed here I think) are the extrusion rates for these extruders. This would allow to have multiple extruders working simultaneously.

No bipolar as both rotating table (polar) and a rotating/sving arm with the hotend. (also polar) There for bipolar. Se the doc linked by Unlimitedbacon [github.com]

Quote
Kenzu
No bipolar as both rotating table (polar) and a rotating/sving arm with the hotend. (also polar) There for bipolar. Se the doc linked by Unlimitedbacon [github.com]

Ah, ic. This is a little bit more tricky, but should also be possible, if you have enough computing power.

But why don't you take a bit more mechanics to make the bipolar printer a polar printer? The actor-mechanism could be quite the same but you would need a linear rail (or smooth rod) where your extruder slides.
I am not sure if you can get the desired precision of the extruders (on the Z-axis) with the bipolar approach (when you have a large radius).

@unlimitedbacon: what kind of bearing did you use for the rotating axis? Do you have plans for upgrading your bed with a heatbed?

Good news, everyone!

Now theres firmware! [github.com]
Here it is printing an octopus [youtu.be], and the final product [imgur.com].
This print took about 1.5 hours. The feedrate was 20 degrees/second which is roughly 55 mm/s. I think that the machine is capable of at least 40 degrees/second.

Quote
Fabricate
I think (although I did not run any experiments) the big problem atm is that the arduino has not enough computing power. For this reason I am going to wait for the RAMPS-FD or RADDS boards to be available, as the Due seems to have a lot more power!

Computing power does not seem to be an issue (yet). As it turns out, the trig functions provided by avr-libc are extremely fast. We ran a test and found that tan() executes in about 13 cycles. An examination of the source code shows that this is because they are using some hand optimized assembly and lookup tables.

Quote
Fabricate
@unlimitedbacon: what kind of bearing did you use for the rotating axis? Do you have plans for upgrading your bed with a heatbed?

Funny you should ask. We just upgraded it with a heated bed. The electrical connections are passed through a slip ring [www.adafruit.com] which also acts as the central bearing. The heater is a big round kapton heater thats made for delta printers [trinitylabs.com].

The slip ring looks great, I wondered how the bed would be heated. Do you run thermistor voltage through the slip ring as well?

---

What is Open Source?
What is Open Source Hardware?
Open Source in a nutshell: the Four Freedoms
CC BY-NC is not an Open Source license

Quote
unlimitedbacon
Computing power does not seem to be an issue (yet). As it turns out, the trig functions provided by avr-libc are extremely fast. We ran a test and found that tan() executes in about 13 cycles. An examination of the source code shows that this is because they are using some hand optimized assembly and lookup tables.

Very nice to hear that! Also the results on your printer look amazing, well done!

Quote
unlimitedbacon

Quote
Fabricate
@unlimitedbacon: what kind of bearing did you use for the rotating axis? Do you have plans for upgrading your bed with a heatbed?

Funny you should ask. We just upgraded it with a heated bed. The electrical connections are passed through a slip ring [www.adafruit.com] which also acts as the central bearing. The heater is a big round kapton heater thats made for delta printers [trinitylabs.com].

Great that you do all the experiments that I planned for myself ! Now I can use your findings, thanks a lot for that !

I also found the slip ring you are using, but I am not quite sure if it can be used for this application. They say that it is rated for 2A each wire, but I think this is an absolut maximum rating for quite a short period of time (sadly I found no datasheet specifying everything precisely). I think the main problem are the very small wires, that are never rated for 2A. Anyway, nice to hear that it works for you! Also the heatbed looks great!

I would have used a slipring designed for wind turbines to transmit the power (something like that) and have the temperature control done via a wireless connection.

I never thought about using the slipring as a central bearing! This is a great idea, how well does it work? I am worried about wobbling on the bed resulting in differences on the Z axis.

Quote
unlimitedbacon
Good news, everyone!

Now theres firmware! [github.com]
Here it is printing an octopus [youtu.be], and the final product [imgur.com].
This print took about 1.5 hours. The feedrate was 20 degrees/second which is roughly 55 mm/s. I think that the machine is capable of at least 40 degrees/second.

Quote
Fabricate
I think (although I did not run any experiments) the big problem atm is that the arduino has not enough computing power. For this reason I am going to wait for the RAMPS-FD or RADDS boards to be available, as the Due seems to have a lot more power!

Computing power does not seem to be an issue (yet). As it turns out, the trig functions provided by avr-libc are extremely fast. We ran a test and found that tan() executes in about 13 cycles. An examination of the source code shows that this is because they are using some hand optimized assembly and lookup tables.

Quote
Fabricate
@unlimitedbacon: what kind of bearing did you use for the rotating axis? Do you have plans for upgrading your bed with a heatbed?

Funny you should ask. We just upgraded it with a heated bed. The electrical connections are passed through a slip ring [www.adafruit.com] which also acts as the central bearing. The heater is a big round kapton heater thats made for delta printers [trinitylabs.com].

Congrats with the firmware :-) I just found this firmware with support for single polar, but also with these optimized features, maybe you could get some inspiration for your own firmware :-)

New continuos rotating print mode.

It allows the printer to see the plate as a continuous space. When it gets to the "end" of the Y axes, it can just continue rotating forward.

Smart path finding

On a disc in some cases there are two paths to travel (one to the left and one to the right), now R-360 goes on the shorter one.

Configurable Centre Crossing Detection

When moving through centre, things could slow down and a bit too much filament get extruded. Not anymore, as R-360 will speed up Y motion to compensate. This is fully configurable setting two values by the user. Give you the power to fine tune your printer.

Rounding Error Workaround

Arduino is an amazing board, however it has some shortcoming on rounding large float numbers. We added a workaround to solve this problem which auto detects rounding errors on rotations and compensates.

[github.com]

But great, just great, now I really need to get started with my own build.

Jesper

Edited 1 time(s). Last edit at 01/13/2014 03:31AM by Kenzu.

Quote
Fabricate
I also found the slip ring you are using, but I am not quite sure if it can be used for this application. They say that it is rated for 2A each wire, but I think this is an absolut maximum rating for quite a short period of time (sadly I found no datasheet specifying everything precisely). I think the main problem are the very small wires, that are never rated for 2A. Anyway, nice to hear that it works for you! Also the heatbed looks great!

We're doubling up the wires to let us pass more current. That slip ring has 12 connections total, so we can use 5 for each connection on the heater and the other 2 for the thermistor. This brings the limit up to 10A (120W). The slip ring hasn't melted yet, so it seems to be working.

The wind turbine slip ring is probably a better solution in the end. I don't think you need to do the temperature sensing wirelessly. That one has 3 connections. You could combine the ground for the heater and thermistor, then use the third connection for the sensor.

Quote
Fabricate
I never thought about using the slipring as a central bearing! This is a great idea, how well does it work? I am worried about wobbling on the bed resulting in differences on the Z axis.

Wobbling is a big problem at the moment, but it was like that before we added the heated bed. So its not because of the slip ring. I think I can fix it by changing the arrangement of the support rollers.

@Kenzu: Figures that someone would find a firmware as soon as I write my own. Definitely going to implement some of these things.

Quote
unlimitedbacon
We're doubling up the wires to let us pass more current. That slip ring has 12 connections total, so we can use 5 for each connection on the heater and the other 2 for the thermistor. This brings the limit up to 10A (120W). The slip ring hasn't melted yet, so it seems to be working.

Nice to hear that it is working at such a high current! I did not expect that. There is also a version with 24 wires 2A each on ebay, just in case.

Quote
unlimitedbacon
The wind turbine slip ring is probably a better solution in the end. I don't think you need to do the temperature sensing wirelessly. That one has 3 connections. You could combine the ground for the heater and thermistor, then use the third connection for the sensor.

You are right, but after reading from issues that start with high current on the same connections as the temperature sensor ground, I don't think that this is a good idea!

Quote
unlimitedbacon
Wobbling is a big problem at the moment, but it was like that before we added the heated bed. So its not because of the slip ring. I think I can fix it by changing the arrangement of the support rollers.

I thought about that for quite a long time and I came up with two solutions:
- There are special bearings for this application, igus has some not too expensive ones. Have a look here (but it is German, sorry). This kind of bearing should work pretty well, I guess.

- The cheaper, but a bit more complicating solution would be to take a big roller-bearing that can take force in both directions (axial and radial) (like the 6209-2RS1 from SKF) in the center of the turntable. Clamp the outer ring of the bearing to the turntable and the inner ring to the floor (or vice versa). Not sure if this works better, but it was just a quick idea that came to my mind.


Anyway, I just wanted to inform you that your printer has been presented today at a large german-speaking website about 3d-printing. Have a look here!

Just a comment for your firmware. You have removed homing of the plate, I know does not need to home, but it could be nice to have a starting point. Say you want to pause a print for filament change or resume a failed print, then you need a known starting point of the plate. You could use a hall-effect sensor with a magnet on the plate or a opto-sensor.

Edited 1 time(s). Last edit at 01/27/2014 06:19AM by Kenzu.

Pretty cool! Looks like it's using a version of the Bipolar Coordinate System. You might wanna make a Gcode preprocessor. This looks like a good source.