Working BB-8 Droid from Star Wars Episode 7

I'd like to invite all our supporters to vote on the direction of the project.  The basic idea is still for a LEGO Technic BB-8 kit of up to 3000 parts to be bought, built and driven around but there are different ways to achieve the functionality, some more successful than others and some more authentic than others.

BB-8 has been developing continuously for 7 months now, plus a bit before project submission, and we've gone from a small prototype with too many colours to a larger model with more functions, mostly the right colours and some deployable tools.

I've found the limits of what the existing LEGO bricks can do.  The rules of LEGO Ideas say I can't invent a large smooth sphere piece or a new magnet piece so I have improvised and innovated by encasing magnets safely in existing minifig dustbins and by making a composite shell from existing pieces.

BB-8 can drive, and turn to some extent, but when I added the mechanism to the shell, the spinning function did not give the performance I had hoped for.  Also the head is not yet able to stay on top with a magnetic system.  Even the strong magnets are too far apart through a composite shell to have the attractive and repelling forces necessary to hold the head in place, and the shell unevenness means the pads under the head would need a wide radius to ride over the bumps, perhaps wider than half the head width.  Overall I'm sure our supporters, the customers of the project, would like more functionality.  We want BB-8 to drive, turn and spin with his head on top.  I can see another way that I think will deliver more.

So we come to a debate with two options:

• Option 1 is to continue in the current direction and see just how far I can push against the existing issues.  BB-8's body can drive and turn but not spin on the spot.  It is unlikely he would be able to move with his head on, other than a small tilt from the turning function.  I have adapted the base of the fitting rig so that he could turn slowly as a static exhibit.
• Option 2 is to try a new direction that would deliver more functionality for less cost, but would seem like cheating compared to the authentic original movie character.  I have a new mechanism that could drive the shell in two halves and will be able to hold the head on an axle rather than using magnets, with potential to turn the head independently of the body.  I have considered the need to synchronise the tool areas (in the orange circles) so that when BB-8 rolls in a straight line it is always with the circles synchronised alternately from each side.  This is the case in the majority of the film footage I have seen, both in Episode 7 and the celebration events.  The order of the tool areas would vary with the turning and spinning function but would default to an alternately-synchronised position.

The deployable tools will stay the same either way.  There is bit more to do on those later.

Please vote by adding a comment with the Comments tab, including "Option 1" or "Option 2".  Are you willing to trade some of the authenticity for better functionality and a more reasonable cost?  All constructive comments appreciated.

Please let me know which option you would prefer and please spread the word.  Get more people involved.  What would you want to buy if this became a kit?

And the other question this week is "Who will be supporter #900?".  I hope we will reach 1000 supporters soon.  I can only do it with your help.  Many thanks.

Thank you for your continued support.  Please keep on sharing.

Having tested the functions in the fitting rig, I added the static counterweights to bring the total up to 12 blocks, similar to the previous weight that included a full Technic AA battery box.  Weight distribution for maximum leverage is for a further update.  For now their positions have provided a good balance to the motors and battery box.

I made the fixture of the weight blocks more robust than before by using studded beams.  The beams also add rigidity to the mechanism, reducing the level of vibration in some parts including the spinning flywheel drive motor.

An earlier attachment of the static weight blocks was outside the space envelope, so the fitting rig does its job of showing the fault prior to putting the mechanism in the shell.  With revisions the whole mechanism fits well with the turning function set straight.  Limited movement to either side is possible but I will still have to see how close the shell size is to the fitting rig approximation.  The shell shape varies a lot with the different parts.  If I find the shell is smaller in any place then I will add to the fitting rig to improve the simulation.

I have added some half-bushes to the spinning flywheel axle.  It was sinking a bit over a few runs but 6 half bushes should prevent that completely.  There were only a few parts of varying axle grip in the middle of the mechanism - 2 half bushes (good), 2 24mm pulleys (fair), an 8-tooth cog (low) and the wheel bearing (nil).  This compares with the axle grip components in the flywheel of 2 4x4 round bricks with cross hole (very good), 2 2x2 bricks with studs and cross hole (very good), 1 2x2 round plate with cross hole (good) and a 2x2 round tile with round hole (nil).  I hope the wheel bearing with half bush and pulleys above it will take most of the weight and hence minimise strain between the lowest of the 6 half bushes and the upper support beam.  I would have liked to add another wheel bearing instead but there are no axle sizes between 16M and 32M.

So now it's time to fit the mechanism to the shell and see how it goes.  If the plan works well then I'll test the rolling, turning and spinning (with brake) functions.  On the fitting rig with the static counterweights I only got more than 30 degrees of spin on a couple of occasions; it would be less with the shell on the rig, considering the extra weight.  The spin angle on carpet might not be too far behind the rig but I don't expect the spinning function to work on a duvet yet.  I'd like a full rotation eventually of course, which would allow a lower spin speed to give a proportion of that.  It did give me a further idea for display.

Another item for a further update would be adding a second L-motor to the spinning function.  This would help with acceleration and top speed with 8 weight blocks in the spinning counterweight.  The single L-motor has been getting warm from repeated spin cycles.  A V2 IR Receiver can drive two L-motors because it does so in the Technic Crawler 9398.  This would need another rebuild of the internal mechanism!

I'd like to do a video of some good results!  Better charge the battery before I fit the mechanism.

A big Thank you to everyone who has supported the idea of a build-and-drive LEGO BB-8 droid; we are close to 900 supporters now.  Please keep on sharing and let's get to 1000 supporters soon.  I can only do it with your help.  Many thanks.

Thank you for your continued support.  Please keep on sharing.

The latest version of the internal mechanism has the drive, tilt, spin and brake functions all working together now.  The tilt function slides all but the drive motors from side to side and is self-centring with the left Servo motor.  The other motors you have seen before.

I have improved the sliding mechanism because the sliding assembly used to turn too much.  Now it has an axle slide as well as wider-spaced beams to help the movement to be more sliding and less turning.

There is a new counterweight too.  I changed the parts in the middle to use some 4x4 round bricks with holes, which attach to some Technic 2x2 plates with hole.  There's a set of those above and below the octagonal plates that sandwich the weight blocks.  This means reduced vibration compared to previous counterweight designs.  I had to hang the counterweight using a 16M axle because the 8M with bezel was not long enough to go through the counterweight, the wheel bearing that carries the weight and any means of extending the axle.  The 16M axle can go all the way through including the drive cog and the upper support, though I would like to use another wheel bearing to carry more weight.  The use of an axle without a bezel means it needs to devote more length to bushes to prevent the counterweight slipping down with use.  I used further bricks with cross hole between the 4x4 round bricks to get maximum traction on the axle.

The rig is the most obvious enhancement.  With the increased weight of the internal mechanism under test, the rig would either fall over or bend over too easily.  A square frame stops the bending over and I've added another layer of frames to stiffen the base and widen its footing, which stops the falling over.  This means it can now run repeated tests of the spinning function.

With extra weight, despite some of it being in the rotating counterweight, the spinning performance is not as good as I would like, though the friction of turning in the rig turntable is more like the friction of BB-8's body turning on the floor.  Perhaps a more modest but more representative result is not a bad result after all.

I might try with counterweights of 4 and 8 blocks.  With 8 blocks it takes longer to spin up and the L-motor is getting warm with repeated use.  Despite being lighter, 4 blocks would spin faster and have wider mean radius.  This might mean less detriment than expected, though I think the full 12-block weight is needed to counter the drive torque and motor weight, so 4 blocks spinning 8 plus the body may not fare well compared to 8 blocks spinning 4 plus the body.

There is further potential to raise the spinning counterweight, use a wider radius with step plates and have the two octagonal plates closer together.  This depends on being able to get through the middle to put static weights below the spinning part but I think a turntable would have too much friction.  It is also necessary to leave a 2M clearance on each side of the counterweight for the tilting function so that the counterweights don't hit the inside of the shell.  This is the purpose of the fitting rig, to describe the sphere shell locus and help to maintain clearances in the mechanism design.

The rig and the new counterweight work well enough that I might try to video a test.  The mechanism seems to be safe enough.  I have to experiment with a new camera for that.  I could include a summary of the shell and head parts too.  I would like to add the static counterweights to the mechanism before I do a trial fit in the shell.

A big Thank you to everyone who has supported the idea of a build-and-drive LEGO BB-8 droid; we are approaching 900 supporters now.  Please keep on sharing and let's get to 1000 supporters soon.  I can only do it with your help.  Many thanks.

Thank you for your continued support.  Please keep on sharing.

The spinning experiments so far have shown that it is best to keep the weight blocks fixed between the two octagonal plates, so a smaller radius allows it to be stronger.  It is time to try out the mechanism on the fitting rig that is designed to test the space envelope within BB-8's shell.  The idea is that the hoop can rotate to show the fit all the way round, describing the spherical limits and making it easy to see any bits that stick out too far.  Much easier to see on the rig than in the shell itself.

The next two pictures show the previous internal mechanism with drive and turning functions that allowed BB-8 to turn with 0.5m radius.  In this case the turning function tilts a static counterweight made from a battery box and 8 weight blocks.  The battery box remains but the weight blocks, which surrounded the battery box and were attached with the black plates, have been removed as I re-used them in the new counterweight blocks.

Following the spinning experiments, I have rebuilt the spinning and braking mechanisms into another orientation to fit them around the drive mechanisms and form a new internal mechanism.  Rather than tilting the angle of the flywheel, which would induce the Coriolis effect like steering a bicycle wheel, the flywheel mechanism can slide sideways and will be driven by the servo motor of the turning mechanism (to be added).  This should achieve a suitable sideways shift in the counterweight to allow turning whilst rolling, usually when the flywheel is not spinning.  To spin, turn and roll at the same time is for a further test.  The next pictures show the fit of the new mechanism within the space envelope.

Also some pictures showing the spinning brake open and closed.

Some pictures of the drive train.  With limited height between the counterweight and the drive axle level, it was not possible to have all the motors act on one axle (compared to the earlier spinning rig) so the brake is on a parallel axle of the same speed with a 24-tooth idler cog in between.

Some of you have been asking for another video.  Whilst these experiments continue, BB-8 is not ready to roll around, but it would be good to have a video of what works well and I am working towards a point where it might be possible.  I would not show any dangerous experiments on video because those are for adults.  Younger fans need to have a safe toy till they decide to be more adventurous in their experiments.  At least these days we can get a wide range of bricks to replace those we melt, so if you fancy a career as an engineer, let me encourage you to experiment whilst putting safety first - LEGO pieces are replaceable but you are not.  I sometimes use a polycarbonate sheet between myself and any model that might shed parts.  In this case, with the flywheel well below eye level, a magazine to absorb any counterweight impact would suffice.  Every "failed" experiment is an opportunity to learn lessons and this is how we develop great products with new technology to meet the challenges of our society and our world.

The next step is a further update to the counterweight and its attachment method.  If the bezel of an 8M axle were located at the top, in a wheel bearing piece, then the counterweight itself needs plenty of traction on the axle to stay on indefinitely whilst enduring vibration.  Even more so if I use more weight blocks as it would need to solve the earlier pull-down issue.  Once I have this sorted, I hope to do a trial fit in the shell.

A big Thank you to everyone who has supported the idea of a build-and-drive LEGO BB-8 droid; we are up to 860 now.  Please keep on sharing and let's get to 1000 supporters soon.  I can only do it with your help.  Many thanks.

In the third experiment I tried a new flywheel with just four weight blocks at a wider radius.  Reduced weight would reduce stress on the axle, bushes and beam.  The wider radius for the centre of mass would mean the product of the mass and the radius would be similar to the baseline, so that the same speed would give the same spin angle.  A slightly higher speed from the reduced mass would give a bit more spin angle.

I based the new flywheel on a motorbike wheel with the tyre removed.  The motorbike wheel had worked well earlier, with easy and secure attachment to the wheel bearing piece on the axle.  I added the octagonal plates above and below, with plates extending outwards from them to the weight blocks.  This would trap the weight blocks between the two octagonal plates, like the baseline flywheel.  I attached the upper octagonal plate to the wheel with some 2x2 plates with hole and double cross blocks.

The reduced weight had the desired effect, with less pull-down on the axle.  There was some vibration but that could be ironed out later.  The flywheel spun faster as the motor found it easier to spin up; the reduced acceleration time is welcome.

I noticed a bit of flexing in the octagonal plates and filled in the gaps with smaller plates to fix it.  A few spins indicated some similar spin angle results to the baseline.

After incremental spins in one direction I tried the other direction.  A longer spin at maximum speed found a sudden shedding of weight blocks - ouch!  This showed how the baseline flywheel was much better in integrity.  In a LEGO model proposed for a product, safety is the primary requirement so I returned to the safe baseline to think of further ways to optimise the flywheel.

In the fourth experiment I replaced the four central weight blocks of the baseline flywheel with bricks.  This changes the mass and radius parameters, from 400g at a mean radius of 4M to 200g at a mean radius of 5M.  The same test gave a spin of these angles from the various speed settings:
Speed 1: 5 degrees
Speed 2: 30 degrees (15)
Speed 3: 60 degrees (45)
Speed 4: 110 degrees (135)
Speed 5: 170 degrees (225)
Speed 6: 250 degrees (315)
Speed 7: 340 degrees (405)

It is better that the flywheel will turn freely at speed 1 and that three speed settings will each give spins of less than 90 degrees.  The flywheel does spin faster at higher settings, perhaps 1.4 times as fast, so that if it did fail and shed the weight blocks, they would travel twice as far as from the lower-speed baseline flywheel.  I put a magazine next to me, so that I would not be injured in the event of failure, but there were no failures and the flywheel did not loosen, despite finding a couple of speeds at which there was significant vibration.

The flywheel with 4 weight blocks did not pull down on the axle and has the advantage of being cheaper by £19!  I will think about making a safety guard around the flywheel so that any shed weight blocks would not hit the shell directly once it is installed in BB-8.

Further experiments will improve flywheel integrity and reduce vibration.  I will have to see whether the flywheel at reduced weight would be sufficient to turn BB-8 with a heavier mechanism, rather than just the test rig.  It might be that I have to return to the original weight of 8 weight blocks and upgrade the axle pulling resistance.  The counterweight might have to be permanently installed, rather than being removable on 3 pegs of a wheel bearing.  Part 3 to follow.

Many thanks for supporting the idea of a controllable LEGO BB-8 droid; we are over 850 now.  Please keep on sharing and let's get to 1000 supporters soon.  I can only do it with your help.

Once again Thank you for your support.

I have been testing and optimising the flywheel for the spinning function, with some interesting results.  The energy for spinning BB-8's body comes from three parameters of the flywheel, its mass, its speed and the radius of the centre of mass.  The flywheel pictured earlier is the baseline.  It has 8 weight blocks adding up to about 400g and a mean radius of 4M (32mm) for the centre of mass.

In the first experiment I tried increasing the mass of the flywheel in order to spin it slower and/or increase the spin angle.  I added 4 weight blocks with some 2x6 plates, recognising that their attachment is less-secure than the original 8 blocks.  Testing at various speeds showed that there were no detachment issues but the total weight of the flywheel, just over 600g, was enough for it to pull down the axle through the bushes, drive cog and brake wheel.  It would need more bushes.

I tested with more bushes.  Whilst this held up better than the wheel bearing that melted before, it eventually caused a bush to dig into the beam.  It resulted in increased friction through the axle hole as the beam hole melted and constricted the axle.  Clearly a heavier flywheel is too much for the bricks.  After the first experiment I turned the affected beam around so that it now has a peg through it instead.  You can see more of the peg in the red circle on the left.  The circle on the right shows where it was before.

Since last time I added the brake function to the spinning mechanism, using a wheel from 42040 Fire Plane and a clamping mechanism like the one from the Technic Ideas Books, driven by a Servo motor.  This works well, bringing the flywheel to a stop quickly as planned.  The Servo Motor on the right rives the brake through a 24:40 gear ratio so that the brake and servo ranges match up.  The grey 'weapon barrel' wheel is ideal for moving two levers.  The spinning drive L-motor is geared up 40:8 to the counterweight axle, with the brake acting directly on the counterweight axle.

In the second experiment I made a jig to test the spinning function, shown in the first picture above.  This holds the spinning mechanism in a frame of angle beams, which is mounted on a 60-tooth turntable to allow it to spin with some friction and simulate the friction of BB-8's sphere body on the floor.  This provides a measurement of effectiveness of spin from different speeds via the resulting angle.  Spinning up the baseline flywheel and applying the brake and stopping the spinning motor simultaneously (pressing both stop buttons at once) gave a spin of these angles from the various speed settings:
Speed 1: no movement
Speed 2: 15 degrees, inconsistent
Speed 3: 45 degrees
Speed 4: 135 degrees
Speed 5: 225 degrees
Speed 6: 315 degrees
Speed 7: 405 degrees

This is a reasonable result, providing small and large spin angles up to just over a full revolution.  I expect the angles would be smaller for the sphere as it would be heavier including the other mechanisms.

Part 2 to follow...

A big Thank you to everyone who has supported the idea of a controllable LEGO BB-8 droid; we are well over 800 now.  Please keep on sharing and let's get to 1000 supporters soon.  I can only do it with your help.  Many thanks.

I bought some more magnets today, for BB-8's head-holding function.  With the full set of magnets in the head and the sphere cluster, the head is better at following the cluster in translation and rotation movements through a thin tray and also the thicker tray made of beams.  It also works through both at once.

The magnets seem to have a longer repulsion strength than their attraction strength.  I can experiment with different configurations to see what works best.

I'm working on a motor and brake mechanism for BB-8's spinning function.  It has to ramp up slowly and stop quickly in order to spin the whole droid.  I have to ensure it's safe because it has quite a lot of energy when spinning fast.  It hasn't melted any more parts or fallen off!

For the brake I thought of using a Servo motor in parallel with the drive gearmotor.  When the gearmotor is running, the Servo motor would be off-centre and would push the brake pads away from the brake wheel.  Stopping the drive motor would also move the Servo motor back to centre and apply the brake.  When I tested this at low speeds it was fine but at high speeds the electrical power generated by the drive motor, back-driven from the counterweight, caused the Servo motor to move about, sometimes to the fully-off-centre position, rather than returning to centre straight away.  I got the same effect when I used a pole reverser switch instead of the IR receiver.  Moving the switch to the middle position open-circuits the drive from the battery box and the drive motor dumps the back-drive power into the Servo motor signal input.  I decided this would not be good for the Servo motor, even if it has protection diodes on its inputs.

I fixed it by using the other output of the IR receiver for the Servo motor.  The motor driver circuits in the IR receiver include flywheel diodes to short any overvoltage from motor back-drive onto the power rails.  The Servo motor was more sensitive than the flywheel diodes on the same output but on the other output the flywheel diodes take effect before the back-drive power can get to the Servo motor signal lines and it gets rid of the unwanted moving about.

The brake will need to absorb a lot of energy quickly.  I remembered the mechanism in the old Technic ideas books.  Using a wheel and tyre with 24mm overall diameter and up to 8mm wide should work for the first brake experiment.  I'm continuing to work towards that.

A big Thank you to everyone who has supported the idea of a controllable LEGO BB-8 droid; we are well over 800 now.  Please keep on sharing and let's get to 1000 supporters soon.  I can only do it with your help.  Many thanks.

A big Thank you to everyone who has supported.  Please keep on sharing.
Here is some progress on BB-8's spinning function.

After gathering some more counterweights I succeeded in making a robust weight from eight of them for the purpose of spinning up and stopping suddenly, to make the spinning function.

It is essential that the weight blocks should stay as one block and not fly off in all directions, particularly not towards people.  I have learnt from previous experience of a large-scale LEGO train crashing into my hand as I repaired the track.  The previous static counterweight showed me that the weight blocks were prone to falling off if their strength of connection was not designed-in as a high priority.

For the whole counterweight I decided to use wheel bearings as the strongest spinning connection available from existing parts, able to withstand quite a high torque.

Early spinning tests with a single Power Functions L-Motor went well.  I ramped up the speed from the LiPo battery and then turned it off.  The torque on turn-off would provide the spinning function's power but this was not as much as I thought I would need to turn BB-8 as a whole droid.  The starting torque was greater when turning-on the battery, having already set the output to full speed.  The LiPo battery is very good with starting torque.  If the current trip limit is exceeded then it will try to restart the model.  I have found the limits of its restart capability with trains on a hill and would not exceed them with BB-8.

I keep in mind that the electronics have to soak up the stress of these sudden starting and stopping operations and it is only a speed control chip inside the battery or IR receiver, plus I have read about LiPo battery overheating problems in laptops, cars and aircraft, so I need to ensure safe use of the LEGO battery for many repeated operations.  The current trip limit of 800mA was not exceeded with these eight weight blocks but the heating of the driver chip is proportional to the square of the area under the current curve if the internal transistors are MOSFETs that have a resistive channel in the ON state.  I still have an open mind about using more weight blocks, whether those would be static or rotating whilst still allowing the turning function to move the counterweight closer to the drive motors.

To get a faster speed I added the 8-block counterweight to the previous internal mechanism that included a pair of Power Functions Train Motors.  Previously these had been able to spin a motorbike wheel with a 5:1 gear-up ratio from 40:18 cogs.  The 8-block counterweight was too great a load for this and was also unable to maintain speed after hand-spinning.  The same was true with a 3:1 gear-up ratio from 36:12 cogs.  With a 1:1 ratio from 24:24 cogs the motors were able to spin up the counterweight from rest to a high speed.

Unfortunately I then found that the wheel bearing melted!  Although the wheel bearings are good at handling radial loads at modest speed and axial loads at rest, a 400g axial load with radial vibration at high speed is too much for them.  The spinning counterweight was caught on the sofa as it fell out of the mechanism so no harm was done beyond the brick melting.  Before I ran the spinning counterweight I recognised the potential for injury from high-energy moving parts, so I took care to make sure my body was not in the way of any spinning parts that might drop.  Just like using power tools - always have an escape plan.

I also recognise that, from a young age, spinning LEGO models fast is fun!  If in doubt, stick to spinning wheels with rubber tyres that don't have significant potential for injury.  For some of my other fast-spinning models, such as jet engines and propellers, I have used a clear polycarbonate sheet between myself and the model.  For BB-8 the spinning counterweight will need a more secure attachment to the central mechanism to prevent damage to any shell parts.

One factor contributing to the event is that the metal weights inside the weight blocks are not fixed so they are able to vibrate, which causes an imbalance and vibration in the whole counterweight.  This would limit the reliable rotational speed but I think a speed lower than a 1:1 ratio of the train motors would be sufficient for the spinning function.

I'm happy that the maximum speed was not more than a 1:1 ratio because a higher ratio gives the reverse drive from the counterweight to the motors a lower load when the motors are turned off, meaning that there is less braking effect from the motors and less power for the spinning function.

Often more progress is made from an apparently failed test result than one that is immediately successful.  There is certainly more opportunity to learn and that's the whole point of LEGO.  This project is more exciting for finding the limits of what the bricks will do.

Ultimately the spinning function needs a mass at a speed to get the energy to spin the sphere as it stops.  A larger mass at a lower speed might be safer as long as it fits.  The size sets the limit of the mass of weight blocks and the speed will have to make up the difference.

As well as the counterweight spinning experiment, I have taken a couple of pictures of the head with the sphere magnet cluster using the thin, clear plastic tray.  I still have to get some more magnets to make the sphere cluster as strong as it can be, and try it through a thicker tray for greater distance.

I'll keep going with the spinning function for next time.

Please join over 800 people in supporting this project.  Share it with your friends and help us reach 1000 supporters soon.

Please share also the magnetic solution and let's petition the LEGO company for this functionality in its products.  Magnets made previous products more fun.  If this project gets enough support then I hope the magnet solution would be developed for the benefit of many themes.  I can only do it with your help.  Many thanks.

A big Thank you to everyone who has supported.  Please keep on sharing.
Here is more about BB-8's magnetic personality!

I have replaced the previous head self-levelling mechanism, which included some of the old-style LEGO magnets, with stronger magnets encased in existing LEGO pieces.

Any moving BB-8 droid needs magnets of some kind if the head is to stay on whilst the sphere rotates.  At least one other LEGO BB-8 model of a smaller scale has used the previous magnets that are no longer available.  All LEGO Ideas projects using magnets, whether old-style LEGO ones or others, will need them to be encased properly for safety.

There is currently no strong magnet piece in the LEGO parts range; the latest buffer beams with magnetic couplings are not designed for a role like BB-8 head-holding and their magnets are too weak.  Even for their purpose in LEGO trains, LEGO train fans have to align the magnets between two buffer beams to get the strongest connection.  Often the stud-based joining facility has to be used so that trains don't leave vehicles behind.  The piece has to allow the internal magnet to rotate whilst encasing it safely in a size no larger than the previous piece.  The increased strength of newer magnet types may be enough to overcome the increased distance from having a safe encasement but it is not really enough to overcome the extra distance needed for the rotation.  Another factor against using these for BB-8 is that the rotation would not allow the individual magnetic polarities to be controlled if BB-8 needed some to attract the head and some to levitate it.

For BB-8's head holding function to work, it is necessary to encase some stronger magnets in some existing LEGO pieces in a safe way so that:

1. they could not be swallowed; 
2. they are in a piece smaller than a buffer beam that would fit inside BB-8 as well as in other applications; and
3. they have good attachment to other LEGO pieces to enable them to be built into the head-holding function and to make it reliable.

My current solution is to use the LEGO dustbin pieces to encase some magnets of about 12mm diameter, with some Technic parts to complete the enclosure and provide attachment.  As with the 'real' BB-8, the magnets are arranged so that other magnets can interact with them, keeping the head in the right place relative to a movement-controlled magnet cluster in the sphere.  I've started with a similar magnet cluster shape for the sphere to the one in the head.  The criteria are strength, shape and fit.

Developing a new, safe, versatile magnet solution for the LEGO system is even more valuable to the LEGO fan community than a kit of BB-8 on his own.  I hope the LEGO company will take this forward for many products.  Quite a few people would like to see a return of functions that were used in the M-Tron Space theme, where items could be picked up from the planet surface as the space ship swooshed over them.  Even though the scope of a LEGO Ideas project has to stick with existing LEGO pieces, it would not be a big stretch for my solution to become a permanent pair of pieces; a modified dustbin and an insertable lid with axle that locked together safely so that they could not be opened by a child.  The same sort of scheme is used to encase the Power Functions motors.  The outer parts would retain the dustbin handles as a means of attachment to clips and as a means of preventing the parts from being swallowed.  The outside of the dustbins might have the underside attachment bits removed as that would reduce the distance between magnets and help with strength.  A rounder underside, more like a boat stud shape, might work well as it would allow a suspended magnet to continue to slide across a surface if it made contact and it would allow other models to use them to slide on a surface without them being suspended.

I have doubled-up on the magnets in each dustbin as it was stronger than single ones.  There was no extra benefit from using more than two.  A test showed that the attraction strength between two dustbins with these magnets is equal to that of the old train coupling magnets when the dustbin lid parts are also attached to the bottom of the bins, so these magnets gain about 8mm of attraction distance over the coupling magnets for the same strength.  Unfortunately the dustbins don't hold 2x2 round boat studs very well on their undersides so I've arranged three of those between the dustbins, again with strong attachment.

The first test of the head and the sphere cluster shows that the head can be controlled from the sphere cluster through a thin plastic tray, even without the full complement of magnets in the sphere cluster.  Magnetic field strength reduces with the square of the distance so there may be a long way to go to get the field strong enough for a 2M-thick shell and air gaps.

The second test, with a set of beams put together as a 1M-thick tray, shows that the head can still move with just a few magnets in the sphere cluster but that's the limit of reliable movement so I should get some more magnets before I try it with a 2M-thick tray.

The arrangement of magnetic poles might extend the field a bit more.  Note that the Disney patent on the design of the 'real' BB-8 uses a bespoke arrangement of magnetic poles.  The magnets I've used so far fit quite well in the dustbin pieces but are too large for barrel pieces.  There are other sizes of magnets available so there is scope to widen the experiments if the first solution needs improving.  I'll experiment at controlling the head through some other parts with more magnets, to see how many it needs.

I hope to make some progress on the spinning function for next time.

Please join almost 800 people in supporting this project.  Share it with your friends and help us reach 1000 supporters soon.

Please share also the magnetic solution and let's petition the LEGO company for this functionality in its products.  It made previous products more fun.  The solution might also work as part of a MagLev train system.  I did a few experiments using the old-style magnets in 2x2 holders but the new solution would work just as well as it also has a 2x2 format.

If this project gets enough support then I hope the magnet solution would be developed for the benefit of many themes.  I can only do it with your help.  Many thanks.

Thank you everyone who has supported. Please keep on sharing.
There are quite a few changes this time. BB-8 has become more attractive in two ways, both in appearance and in magnetic personality!

I have replaced the white tubes with orange ones to make the orange circles more prominent. This helps them to look rounder too. There might be a benefit in some 3M liftarms in orange but those are not in a current set.

I have replaced the previous head self-levelling mechanism with stronger magnets. A few of you asked about how BB-8 works, so I'll say a bit about how previous mechanisms have worked. The self-levelling mechanism uses the 36-tooth cogs, which have bevelled edges, as pads. Whilst one pad stays still in the middle, opposite pairs of pads around it rise and fall together to match the contour of the sphere shell. The size of the pads was designed to be large enough for them not to fall into recesses too easily. The rise and fall range is more than sufficient for the shell features. The mechanism uses bevel gears in U-frames to make the "opposite" function. One pair has to have the drive cross over the other, so it is raised with some 16-tooth cogs. Each pad has a representative train coupling magnet attached but those are no longer available because toy safety regulations have been tightened since the last sets with loose magnets were released. I knew the magnets would need more magnetic strength.

More about the magnets that replace this mechanism next time.  The size of Updates is limited!

From the latest parts order, some more weight blocks will enable me to swap out the battery box I was using and make a more regular shaped counterweight. This might enable a spinning function based on spinning up the counterweight slowly and stopping it quickly, so I will experiment to see how well that would work. I expect the spinning function would work better on a hard surface where the rolling and steering work better on a soft surface. These pictures show the previous counterweight, with a battery box and 8 weight blocks. The 2x2 plates with side hole are a good interface but more are needed to make sure the counterweight is secure. Even more so if they will spin up! 4 blocks are about the same weight as a full AA battery box.

Here is a view inside the sphere showing where the battery box was attached (mechanism upside-down) and also a right-way-up view. With the orange tiles added to the circles it limits the opening of the panels before detaching the #6 axle joints that hold them. Attaching them in the closed position was difficult but it's not so bad in the open position.

I have jig to test the clearances of the internal mechanism against the shell. The next task is to rebuild it to be as tight as the sphere so that I don't get too optimistic about how much space there is for the mechanism! The previous mechanisms have had some clashes with parts of the sphere. Building to a spherical space envelope is a new experience that is not taught in other sets. If this gets enough support and becomes a set then that's an extra learning point to add to the fulfilment of LEGO company values.

Please join well over 700 people in supporting this project. Share it with your friends and help us reach 1000 supporters soon. I couldn't do it without you. Many thanks.