PropellerBraider - Towards a P2-Controlled CNC-Braiding Machine
Christof Eb.
Posts: 1,572
Hi, this is work in progress and an experiment, as it is new. - At least I have not seen this kind of setup.
Perhaps nevertheless it might be inspiring or funny or interesting....
So what is it about?
Braiding is known for hair and also yet another textile method. Others are weaving, knitting, crochet,... (in German there are also knüpfen and wirken, I don't know the English words.)
The basic setup and movements for braiding are quite simple and yet there are many many ways of variation. And this makes this technology interesting for a CNC project, I think.
Variants can go as far as "klöppeln", doing braiding lace:
https://commons.wikimedia.org/wiki/File:Ursuline_lace_2.jpg#/media/File:Ursuline_lace_2.jpg
This picture shows some basics: You need a number of bobbins with strands on them. You will just twist these. Direction and order of the twists creates the patterns.
Here is a very nice site showing lace pattern possibilities: https://www.theedkins.co.uk/jo/lace/patbegint.htm
Well, doing flat braiding lace would be the far end, but there are far simpler but nevertheless interesting goals in between.
There are braiding machines. Simple ones for fixt patterns or even ones for lace. They are optimised for speed and do as many movements in parallel as possible. Here are two videos, where you can see the basic movements and also a machine, which is controlled by punchcards: http://das-kloeppelmuseum.de/die-kloeppeltechniken/
Such fast machine is rather big. One of the reasons for the size is, that each bobbin carrier has some means to maintain tension of the yarn while the distance from the bobbin to the braiding point varies. This is done with a weight or a spring in parallel to a break. This link shows an impressing mechanical round braiding machine and his bobbin carrier in detail: https://fraensengineering.com/3d-printed-braiding-machine/ The height of his carrier is about 165mm. I have studied patents for bobbin carriers. There are quiet a lot....
This picture shows a my very basic means to do a braided cord:
A nice German pdf about this can be found here: https://www.regensburg.de/fm/RBG_INTER1S_VM.a.253.de/r_upload/Onlineprogramm - Freundschaftsarmband.pdf I have learned, that this method is called "Kumihimo". (Actually the way of braiding used here without twists is rather special. It is funny: I have tried to vary the number of strands from 7. But this is the only number that works in this way! Some sort of miracle.)
It shows some basic principles of the way, I want to go here:
- Have some strands of yarn, maybe with different colours
- Each strand will be held on a bobbin
- The bobbins sit on positions on a round plate.
- The fabric or cord leaves downward through the middle of the plate.
- Tension is controlled while moving bobbins.
- Movement of bobbins is not radial but vertical to the plate and in circumferential direction. (I hope, this keeps the question of tension more easy)
- The plate is rotated during work.
For my design, I have decided to split the movements:
The plate with the holding places for the bobbins will rotate and therefore I will call it carousel. A pickup-spindle can grab the bobbin with an electromagnet, has the ability to rotate the bobbin to tighten the yarn and can lift the bobbin.
There will be two pickup-spindles to be able to swap two bobbins more easily.
The idea is to be able to have the possibility of a larger carousel, but to begin with a smaller one. (Have a limited disaster, if it does not work....)
So, here is a picture of the current design, only one pickup-spindle shown:
Diameter of the carousel with 16 positions is about 300mm.
There are:
3 step motors, one for rotation via belt drive and 2 for lifting
2 DC motors for the tension and
2 electromagnets.
A 4th step motor will be used to wind up the cord. (Not shown)
I think, that for this project, to have a P2 controller with 8 cores is quite helpful. The actuations are related but not like Bresenham in one movement.
As said before, at the moment this is all theory. I have begun to produce first parts. I don't know, if it will work. 
One bigger question is, if the electromagnet will hold and also let go the spindle. It is D20x15mm and said to hold 2,5kg in ideal cases. For solenoid armature, fixed to the bobbin, I hope a steel washer will work. ??? I will provide means for AC voltage and PWM.
Perhaps some comments, ideas?
Have fun, Christof
Edit: Have done some testing with the solenoid. Yes, it will hold and also release the bobbin. But it tends to de-centre the bobbin.
Comments
An update: It's braiding!
So far I have done braiding with 3 strands and now round braiding with 4 strands, Kumihimo Style.
Have had big problem to 3D-print larger parts from ABS. Awful warping and cracks. A different type of ABS solved this. This even led to a failing main bearing unit....
Second difficulty was to find the right shape of the chamfer at the lower end of the bobbin. When the bobbin is put down onto the carousel, it's position shall be realigned. The chamfer leads it then. The picture shows the bobbin. The carousel plate is 12mm thick. The chamfer of the bobbin is also sanded for low friction. The chamfer must allow sufficient variance of position, must not stick when the bobbin is put down and the length of the cylinder must not be reduced too much. Under the carousel there is the belt drive wich must not collide.
At the upper end of the bobbin a steel washer is glued to the face. To prevent sticking two layers of Tesa film adhesive cover the washer and form a magnetic gap.
The braiding point is now elevated from the carousel plate, which decreases variation of tension. The tensioning motor is not used. Yarn tension is so low, that the motor is not moved, even with no motor current at all.
To pull down the braided cord, a peg with weight of about 8,5 grams is attached to the knot at start. When the bobbin is moved down on the carousel, the yarn tension is high and then sometimes the bobbin spins a little bit to release perhaps 1mm of yarn.
And some "product":
This is acryl wool. Diameter of the picture is about 8cm. I am quiet happy, that the braid has this good regularity. This is very sensible to the tension when the bobbins are wound.
Cheers, Christof
https://youtube.com/shorts/oZYtT4F1XFk
So now there is a video showing a 8-strands round Kumihimo style being done.
Meanwhile a dedicated COG is used to do the stepping of one movement at a time. Sequence of movements is controlled by a cooperative task running in the main COG. With this setup the system remains fully accessible during breading.
Have Fun, Christof
Great project video Christof, an impressive machine!
I have posted the actual version of P2XCForth. https://forums.parallax.com/discussion/176239/p2xcforth-a-fast-hybrid-xbyte-and-c#latest Control of PropellerBraider is done with this Forth, see Block50. One of the nice things about Forth is, that you extend this language. So now for example there is a new command "home", which lets the machines step motors go to the start position. You can type this into the terminal or you can use the new word in a new program. So the description of a breading pattern is a program, that can be compiled and executed. You don't need to write an interpreter, it's already there.
The last line here sets the former bobbin position number 2 to be number 0 from now on.
So evoking "round8" will do one cycle of movements.
The braiding point is elevated some more, which gives better consistency and also makes the braid more dense.
Two variations of the same 8 strands round Kumihimo. You can vary the pattern, if you have different colours at different bobbin positions at start. https://friendship-bracelets.net/patterns/kumihimo?SortBy=&MinColors=&MaxColors=&MinStrings=&MaxStrings=8&MadeByUser=&Keywords=&SimilarToPatternId=&OnlyPatternsWithPhoto=false&OnlyPatternsWithVideo=false lists 80 "proven" nice patterns with 8 strands. This site also has a design tool for patterns.
Great photos.
I am still experimenting with geometry of the braiding point. This version has a smaller diameter, which shall give a braid more dense. Last trials have been with a combination of cotton yarn and some thin acryl. Both have very low elasticity which makes things more difficult.
Also I have now activated one tensioning motor. It sits above the spindle. The electromagnet acts both as a gripper and a clutch. The motor can turn the spindle to tension the yarn. It is driven via a L298 bridge. The motor is rated for 12V but we only use 5V and smartpin PWM to be able to fine tune torque. This setup works. :-) However the actual diameter of the spool has influence on the tension. Well, that's mechanics....
Some update on speed:
) in P2XCForth and faster variables. My old step routine using smart pulses had unwanted lags, this is solved using smart transitions.
I have invested quiet some time to make the machine faster. There is now a simple Assembler ( as far as I needed it
There has been a "speed barrier". Above it, there was something like a resonance on all axis. It was around 70kHz microstep frequency = 2.1kHz full step frequency. I did and do not understand it. Normal resonance frequencies are supposed to be lower. Also the vertical axis are very different from the rotation yet the limiting frequency was rather similar.
I had used TMC2209 step drivers. I do love that they reduce the current at stillstand. Also they are running very smooth.
I had the impression, that the resonance problem had to do with the chopper frequency of the current regulator. (????)
But after solving the lags issue and swapping the driver modules to DRV8825 I can go 3 (!!!) times faster on all axis. The DRV8825 need active cooling though.
Also acceleration does now use something like dt= c1/(sqrt(d+d0)). d = the smaller distance to the start- and endpoints. dt= step time. Nice, that P2 has Cordic for division and sqrt.
The downside is, that the machine is so fast now, that you cannot press the STOP button in time, if something goes wrong....
There is a unplanned safety feature: The tensioning motor is only clamped, so if there is a vertical collision, it will slip in it's fixture.
I guess this was so called "mid-band" resonance. It normally occurs around the frequency where the torque begins to drop. Below that frequency the motor runs in constant current mode and the torque is nearly constant. Above it the winding inductance causes the current to be limited to a lower value. The torque drops off at an 1/f rate. Around the "corner frequency" the current control loop can become unstable because the current to voltage phase relations goes from 0° (resistive) to 90° (inductive). Back-EMF changing by small speed fluctuations can get close to 180° phase shift and causes positive feedback and oscillations.
It can be damped either with an external speed feedback (encoder) or by adding a small phase shift proportional to the derivative of the motor current.
Video, please!
https://youtube.com/shorts/D3WCXPby8L0?is=j5OQmkOUTmmqgaGL
looking good, a fascinating project!> @"Christof Eb." said:
Oh, and I think that your machine has pretty better precision at the same time!
One difficulty is to bring the bearing unit into the exact center of the wooden carousel. But I am quiet astonished that you can 3dprint a gear for a belt with 2mm pitch. And it does not show wear so far, made of ABS.
Well, if you feel so, you could let your machine do braiding. It's all about pick and place.....
Thank you very much for these insights!
And as allways, if you know the right wording, you can Google much better. This has been discussed some years ago here https://forums.parallax.com/discussion/139086/stepper-motor-midband-resonance-compensation