How to solve the heat problem？My pcb heat is very serious, can not work for a long time, at most can only run about 10s, because it is too hot

Finally implemented this into a project.

https://hackaday.io/project/189880-magnetic-kinetic-art-display

Hi, I had your PCBs manufactured (0.6mm thickness and flex PCB). I am trying to get the magnets moving. I have 4pcs N52 magnets 5mm diameter, 2mm heigth and TFB6612FNG stepper drivers. Can you please advise on some of the parameters you used in the video, like motor voltage. Also parameters in the code like steps per revolution, speed, number of steps.

try smaller magnets .... 2mm I am not remember all now. it is too old project

How did you decide on the trace spacing? Was it just trial and error with the magnets you chose?

trial / error, 2mm magnets

Ah thanks. What are your opinions on the SRI patent as I'm interested in making some open source model diorama animation boards for figures and selling them?

I think this project using same driving like SRI, but this https://hackaday.io/project/154496-2d-actuator-move-micro-robot-in-xy-2d-space using other driving scheme (3phase, unipolar driving) and is open source. I have plans for sale in future,  I want separated driving and moving board but it is not finished now.

Can I buy a pcb for experimenting?

yes you can but I have not testet latest boards. also design files are on github you can order in any cheap manufacturer

I checked with my local supplier. (Not cheapest) and had to pay more than 60 eur, it would be more efficient if you ordered multiple anyway and had one to spare (20 eur is better than 60).

Btw, I designed a small linear variant yesterday evening to test this  friction and related ideas, boards arrive in 7 days. I really see many many cool applications for this:

Lego compatible linear motor to act as muscles for robotic arm etc (much lighter and more compact than any other actuator based on rotating motors)

Scientific manipulators in tight spaces and vacuum

Chess computer with moving pieces (multi piece required)

Clock, braille display as mentioned by others

Old skool led vu meter for audio, but now with dancing magnets

Conveyer belt systems (race track as your other project)

In order to improve performance I also wonder about using smd solenoids (without magnetic shielding or ferromagnetic core) and making a matrix with those. That would allow to control multiple objects, but would make the design quite a bit more components...would that still be attractive? In terms of components, size, cost would be similar to those led matrices.

positioning could be done with delay line principle, send a fast pulse and look for reflection on both ends, can be capacitively coupled on top of the drive currents; saves the hall sensor and might be capable of detecting more than 1 object. Timing will be critical though.

Multiple objects is an interesting challenge: matrix adressing, objects which have a smd solenoid that is only on at certain brief times, resonant lc object that gets energy via induction and that only moves at very specific velocity, different weights and using inertia difference,...

Would love to see improvement in resolution but not sure how: maybe some incommensurate shaped object with the pattern?

For weak y drive: how about putting a magnet on both sides of pcb right on top of each other? That way the field is strong on both ends and the system does not rely (or less) on gravity and might also work in vertical position. Having a ferromagnetic return path would further increase strength and limit stray fields.

two magnets on both sides not work, high friction. if exist any drawing for thorlabs mount I can make next version to fit in. I am working on this project more than year and it is only money and time eater without purpose, I have full desk of prototypes.

for example of Thorlabs 30mm cage system with cad drwaings see here

https://www.thorlabs.com/thorproduct.cfm?partnumber=CPM06

I could imagine the pcb following the hole pattern and fitting on top of such a standard plate. A hole in the center would be nice (for optics) and the pcb could move a very light assembly with e.g, 3 or 4 magnets that drag the ring around. 2 magnets could do x motion and 2 other y, this way all the tracks can be on top in 4 quadrants.

What if friction could allow to do stick-slip movement where a short strong current pulse would move the object a very small fraction of the normal step? Repeating this multiple times allows the puck to crawl around with very high resolution but probably at the expense of not knowing where we are (step may vary)

Anyway, this is probably diverging very much from the original goal, but if this works there might be a market with relatively high spending power ;)

cool project! Wonder if it could be transformed in a small xy table for scientific/optics/microscopy? These are typically bulky and very expensive. Moreover this would be more or less vacuum compatible. For this application it would be nice if it would fit on e g a 30mm optical cage mount of thorlabs. 

I think is no problem make it smaller. I have plan make smaller board because FLEX contest support cover price only for 2sqr inch. I am not sure if resolution / step is acceptable, it is about 0.5mm / step. I think flex board is required to no lost bottom tracking force.

indeed, resolution is limiting for this application, 500 times better would make it highly attractive, but this is already very nice.

Are you going to be focusing more on this version than the mBot version? By using a stepper driver you get the fine controls of microstepping managed for you at least! I was thinking you can use Kapton on FPC material and stretch it on a LEGO frame, would help with the bottom layer strength issue.

Yes I am focused now on this version, I have plan for simple/easy/cheap mBot version with Attiny84

I want to see something like this in a braille display.

y is this called a 2.5 d? really cool work .. If you could control 2 robots simultaneously it would make a cool clock. i hae always wanted a minimilistic clock with just 2 circles moving about on a surface ! 

if you wave flex board you can can move in Z position of course this more joke than practical. you can not control 2 elements now, but you can separate board to more zones. I have no purpose now for this because I have not solved something like "hand" it can just push objects :( if it can carry smt component ... it can be amazing pick and place, but I have no idea how to do.

Well implemented. Neat ! Any plans of making it open source ?

Yes, if I win TOP20 in Hackaday prize 2019 :)

Don't the rules of the prize require you to open source it?

Of course, I will publish all the documents, but now I have to keep people doubt. ... In end of May. I have not tested latest design.

You had me at Lego. #P1 - Lego compatible drag feeder The project is pretty neat too.  :)

Did you notice heating issues in this design when you run it for a longer period of time? What about actuating more than one magnet independently? 

I love 2D magnetic robots! Hope to see more details posted (: