Close
0%
0%

Braker One Robot

3D printed single motor robot

Public Chat
Similar projects worth following
Braker One is a proof of concept 3D printed robot based on a novel inexpensive DIY actuator system. Similar to hydraulics, it distributes power from a single power source such as an engine (gas, steam, sterling, turbine) or a motor (AC / DC / brushless / stepper) to multiple actuators. It does so through a series of interconnected miter gears located in every joint. ALL gears spin continuously delivering power to all joints. Two miter gears in each joint that spin in the opposite directions have disk brakes attached to them. Use of MR fluid brakes is also possible.
When a joint needs to rotate, a caliper or an Eddy Current brake located on a limb acts on the continuously rotating disk and by doing so rotates the joint. Instead of converting energy into heat however, brakes convert spinning disk's energy into limb movements. When a brake is not applied, the joint is fully COMPLIANT and rotates freely! Brakes are smaller than motors, providing higher power to weight ratio.

Braker One was born because we need an inexpensive 3D printed backdriveable (compliant) robot actuator.  If proven efficient enough, this technology can allow construction of biped, quadruped, hex walkers and even rolling robots.  It is based on my other project: https://www.thingiverse.com/thing:4327799

There are many approaches to creating robotic actuators.  Several of them are available to hobbyists.  Servo motors are most common but they are not backdriveable (not compliant).  Low gear ratio brushless motor actuators are becoming available but they are expensive and require sophisticated electronics to control them.  Distributing power from a single power source through a transmission mechanism such as hydraulics or pneumatics is costly and complicated.

I am going to work on the robot in the following stages.

1) Creation of a single leg mounted on a stand with 3 joints and a single geared AC or DC motor without brakes to test the efficiency of the 3D printed mechanism.

2) Creation and testing of Eddy current or electro-mechanical brake/caliper mechanisms.

3) creation of 3D printed joint position, load and other sensors based on FiberGrid technology described here: https://hackaday.io/project/167317-fibergrid

Braking actuator which  takes care of the mechanics, is one of many technologies I have been working on for creating affordable robots.  Electronics were replaced by my FiberGrid optical sensor framework (https://hackaday.io/project/167317-fibergrid).  Control system will be based on my distributed computation framework distributAr (https://hackaday.io/project/167317-fibergrid) simulating an Artificial Spiking Neural Network.  The reason I choose a SNN as control architecture is described in my paper "Perception and time in Artificial General Intelligence" (https://github.com/rand3289/PerceptionTime).

All of these ideas are FAR OUT OF THE BOX.  I do not expect you to agree with everything I am doing in these projects, but if something clicks, I am looking for constructive feedback or questions.  PM me on hackaday or email me toandrey@yahoo.com

caliper.png

CAD rendering of a brake caliper

Portable Network Graphics (PNG) - 63.48 kB - 07/13/2020 at 08:48

Preview
Download

Zip Archive - 1.02 MB - 07/06/2020 at 19:05

Download

IMG_3996.JPG

The calaw :) Eddy current brake

JPEG Image - 3.71 MB - 06/28/2020 at 23:07

Preview
Download

Portable Network Graphics (PNG) - 868.55 kB - 06/13/2020 at 22:01

Preview
Download

IMG_3957.JPG

Assembled gearbox and eddy current break

JPEG Image - 955.46 kB - 06/05/2020 at 04:02

Preview
Download

View all 12 files

  • 30 × 608 bearings
  • 6 × 3.5" HD plates Hard Drive plates from 3.5" mechanical Hard Drives used as disk brakes ( or 3"-4" copper disks for eddy current brakes )
  • 1 × Motor of any kind
  • 14 × Brakes Mechanical (solenoid/servo based) calipers or eddy current brakes
  • 5 × Small Nails Nails for use as pins for attaching driveGears to shafts

View all 7 components

  • Other clutch based actuators

    rand328907/24/2020 at 21:25 0 comments
  • Brake calipers

    rand328907/13/2020 at 08:55 0 comments

    I am thinking along these lines for a brake caliper actuated by a servo.

    Jaws will be pushed apart by a spring.  They will be pulled together by cables attached to each jaw and going through the center of the base.  Jaw ends will have 2x8x14 mm slices of pencil eraser glued to it.

  • Eddy current brake failure :(

    rand328906/29/2020 at 00:20 0 comments

    I've made the eddy current brake.  It  has about 14 m of 20 AWG magnet wire wound about a chain link.  It has about 0.5 ohm resistance.

    I tested it by holding the disk spinning at about 450 rpm in the slot of the chain link.  The magnet was in my hand.  I briefly applied about 10 V from a 30 A variable power supply.  I felt NO FORCE during the time the coil was energized.  It got warm within seconds so there was a lot of current ( my calculations tell me about 20 A) flowing through the coil.  At this point I call my eddy current brake a failed experiment.  It could be the magnet, it could be the low RPM or the thin aluminum plate. 

    Rumbling and whining:
    This thing was hard to make and it didn't work.  I had to buy magnet wire on line, drive to buy a chain in the store, drive to the beach to get sand for annealing process.  I've spent about 40 minutes heating the chain in the sand with a propane torch which after all probably didn't change the magnetic properties of the metal much but made it a lot softer to cut and bend.  I had to unspool the wire to a larger spool so I can measure how much wire I will be winding which took about an hour.  Hours of research on winding magnets, trying to figure out wire resistance, max currents, weight and other crap.  Then I had to wind the coil which took me about two hours.  I didn't want to connect it to the power supply directly since it had such a low resistance so I had to wire some resistors in parallel to lower the overall resistance because I didn't have any low ohm resistors.  Wire the coil with resistors and a switch in series.  Connect it to the power supply.  At this point I think my meter died because I tried measuring current through the low current connector.  My 30 A power supply's current indicator also died! After all that work I get no force whatsoever from the brake.  At this point I think I need to take a chill pill called fu*kitall and move on to a different type of brake.

  • Eddy current brake take II continued

    rand328906/19/2020 at 06:09 0 comments

    I got 14 links of 5/16" (8.8 mm thick) zinc coated chain.  It is 18.25" long.  They weigh 533 grams.  Link's outer dimensions are about 28x48 mm.  It was US $2.97 per foot @ homedepot.  

    I also bought 1 lb of 20 AWG magnet wire for around $16 US on Amazon.  1 lb of wire should be about 323 ft and I am hoping to use it for 3 breaks.  At about 1 Ohm each.  I need 14 total for 7 DOF.  $5/brake $75 total.  Each brake will be about 190 g.  2646 g (5.8 lbs) total.

    This is getting heavy and expensive... (OR not https://temcoindustrial.com/20-awg-copper-magnet-wire-5-lb-mw0027-magnetic-coil-red-soderon155.html  OR https://temcoindustrial.com/20-awg-copper-magnet-wire-5-lb-mw0167-magnetic-coil-gpmr200.html)  I was hoping to make the whole robot for under $100.

    Next thing to do is to anneal it in sand using a torch to improve its magnetic qualities.  This should also strip the zinc coating.  For this reason I bought zinc coated and not galvanized chain.  It is harmful to inhale zinc vapors.  Do it outside !!!

    After annealing I will cut a slot in each link and try to separate the chain by bending them open.  Close the links back and leave about 3 mm gaps for disks to spin through.

    Looking at a 1 lb spool of wire it seems big.  I hope I will be able to wind 107 ft (32.6 m)  around the link.  It will take some time :)

    I don't  know how much current a 20 AWG wire will take.  Reference materials on the web state anywhere from 1.5 A wound (http://www.magnetwire.biz/magnet_wire_current.html ) to 18 A for a single conductor in air.  The brakes will not be operated at 100% duty cycle.  I am guessing 25% max.  This might need to be enforced in software.  Spinning disks can be equipped with blades for air cooling the brakes.  Therefore, I am hoping to be able to operate the brakes at 12 V through  IRFZ44  mosfets or MCP14A1202 or even MCP14A09?  Anyone has a schematic to recover the conductor EMF energy after it is turned off?

  • Four-bar linkage

    rand328906/13/2020 at 21:52 0 comments

    By using a four-bar linkage to replace a knee joint it may be possible to eliminate spinning disks at the knee joints!  The number of breaks would stay the same. The two from the knees would just move up to the previous (hip) joint.  This will increase the efficiency because the number of spinning gears and disks will decrease.  Allow disks to spin at higher speeds.  Reduce weight and make limbs lighter by moving the breaks closer to the body.  Reduce the bulk of the robot.  A bipedal configuration will require only 6 spinning disks.  This will however put a higher thermal load on the hip joint disks.  Also it will reduce the angles by which the joints can turn.  This change however leads to a flat hip frame which will be easier to print.  And here are my Hand Aided Design (HAD) drawings:

    In this sketch brakes would be mounted on A and B.  Shafts and how A sits on the shaft are not shown on the left.  Disks are not drawn to scale on the left and omitted completely on the right.

  • Eddy current brake take II

    rand328906/10/2020 at 03:58 0 comments

    Copper Disks:

    It would be nice to replace aluminum HD platelets with some copper ones to increase efficiency.  Maybe make them smaller and spin them faster? Should I make them copper rings to reduce the amount of copper used and weight?  On the other hand disks have larger thermal capacity than rings.  Aluminum disks with copper rings attached on both sides?
     I need about 10x10 cm of copper sheet to make a disk.  Where do I get cheap copper?

    Plan B for getting an eddy current brake to work:

    Buy a short length of steel "coil" chain.  Anneal it to improve magnetic properties.  Cut it into individual links with a grinder or a hack saw or a sawzall.  Individual links will form cores for the eddy breaks.  Disks will spin in the groove made by the cut.  Wind an insulated copper wire around the link to form an electromagnet.  Should I get 1/4" or 5/16" or 3/8" or 1/2" chain? Would it be better to use short pieces of 3/16" thick  1 1/4 square tubing instead?  I wish I could find a source of soft iron.  Steel stays magnetized.

    Now where do I get some cheap copper wire? I don't have any transformers laying around... Scrap metal stores?  Aliexpress sells it for about $50/Kg shipped - WTF?  Amazon for $32/Kg shipped.  Current market price of copper is USD $5.65/Kg.  Time to start melting pennies and making my own wire lol...  Can I make a copper disk by soldering copper pennies together?

    What diameter of wire do I need?  What length of wire will fit onto one coil of chain?  How heavy is it going to be?  If I want to use 12V, I need 106 ft (32 m) of 22 AWG (0.64 mm) wire to get 7A.  Alternatively I can use 133 ft (40.5 m) of 24 AWG (0.51 mm) to get 3.5A  How much current do I need to create enough eddy current?  What amount of impedance will it provide?  Is this why people get EE degrees???  Does this whole thing make any sense?  Am I just designing my own motors here???  HELP!!!  I am lost!!!

  • Eddy current brake

    rand328906/05/2020 at 03:27 0 comments

    I've constructed and tested an eddy current brake.  The test was a failure.  I used about 8m (30 turns) of 0.33 mm (#28 AWG) insulated copper wire wound into an 85 mm coil.  It has about a 1.5 Ohm resistance and rated for 1.4 A max current. Therefore I did not use more than 3V to test it.  I will try boosting the voltage a bit more since I don't care about destroying it.  However, at 3V and about 450 RPM I did not see any breaking (movement of the tibia).  This is not a standard eddy break coil configuration.  I wanted to try a core-less design.

    I am also working on a new Femur designed to print in one piece.  Both eddyBreak.FCStd and femurBlock.FCStd are available in BreakerOne_experimental_2020-06-04.zip

  • Gearbox

    rand328905/31/2020 at 07:02 0 comments

    I made a gearbox with about 29:1 ratio that drives the spine shaft of the robot.  It contains two 58:11 teeth gear sets reducing the RPM and increasing the torque. The gearbox has 3 3D printed axles supported by 6 ABEC-7 608zz bearings.  All gears have a 1.3 mm module.  Cad files are available for download. 

    I am using a Johnson DC motor rotating at 14000 RPM and consuming 0.7A @ 13.8 VDC without a load.  When installed in the robot the motor spins at about 13000 RPM (calculated) and consumes 1.2-1.3 A @ 13.8 VDC. The motor runs directly from a 2.5 A power supply.  Disks spin at around 450 RPM one way, as measured by my PYLE laser tachometer and about 520 RPM the other way ( reverse DC polarity ).  This is a bit high.  Ideally I want under 300 RPM at the disks.  This thing is loud and exhibits vibrations.  After 10 min of continuous rotation the motor got hot to touch.  Motor gear (first gear) and its bearings got a bit warm.  All other parts remained cold.

    As a result I do not recommend using the gearbox to reduce motor speed.  It is best to use a commercial geared DC motor with output shaft speed under 300 RPM.  Use a belt or 1:1 gears to connect the motor output shaft to the spine shaft within the gearbox. A small gear on the spine shaft will eliminate the reduction.  (see gearBoxOneToOneGear.zip)

View all 8 project logs

  • 1
    In general

    Along with pictures which you will be able to see in the gallery, as work progresses and the design changes, I will be uploading dated zip files with the latest FreeCad and stl files for printing all parts on a 3D printer.  I use Creality Ender 3.   Minimal print bed dimensions you need are about 200x200 mm.  

    Print everything with 0.3mm layers.  Print shafts and hips with a brim.  One of the gears (driveGear) requires supports.  I do not have any ABS however it might be best to print gears with disk breaks out of ABS since they might get warm.  Be careful, I've heard ABS shrinks a bit.

    If you decide to print a non-final design, there are two problems that need to be fixed. Gears do not have a 45 degree miter.  (Use untested gears from BreakerOne_gears_2020-05-27.zip )  Second, when two hips are put together, hips' shafts do not align on the central gears the way I want them.

  • 2
    Part counts

    Each leg requires the following number of part copies: 1 hip, 2 femurs, 1 tibia, 2 driveGears, 3 endGears, 4 bearingGears, 1 of each shafts.  holeTransferPattern can be used to mark holes on the 3.5" Hard Drive plates.  Hip is about a 3.5 hour print, gearbox body is about 3 hours,  femur is about 1.5 hours and the rest of the parts are under an hour each.  All miter gears have a 1.3 mm module.  All involute gears also have a 1.3 mm module.

  • 3
    Part cleanup

    I use a 22 mm hole saw to clean up all bearing holes before inserting bearings.  Sand all 3D printed shafts with sandpaper.  File the brim with a file.  Holes of all gears need some filing with a small round or flat file.  Clean up the pin holes with a 2 mm drill bit and screw holes with a 4 mm bit.

View all 3 instructions

Enjoy this project?

Share

Discussions

Cat wrote 07/05/2020 at 16:41 point

Interesting.

Can you make one of the actuators stop rigidly in one position while the rest still rotate?  I don't mean just not drive that actuator, but make it solidly keep one position?

  Are you sure? yes | no

rand3289 wrote 07/06/2020 at 05:20 point

It might be possible to balance external force applied to the joint with the force generated by braking to keep one position.  Like muscles balance forces while holding an object in place.  This claim needs to be tested.  

However, brake actuators are compliant!  Since rigidity is a synonym of stiffness and compliance is an inverse of stiffness, by definition, this actuator can not stop rigidly.

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates