Solar BEAM Marble Machine

Ladies and gentlemen, I present to you, my marvellous "Solar BEAM Marble Machine"! A brass marble machine with a solar powered, solenoid actuated, lever lifting mechanism!

This machine uses capacitors to store energy from the sun and, once charged, dumps all that energy into a coil, pushing away a magnet, and lifting a small steel ball through the use of a lever.

In previous BEAM related Instructables there was some confusion surrounding my use of the term BEAM. I don't think it is such a common concept anymore. How I understand it is that BEAM is a hobby electronics movement from the late 90s to early 00s that focused on turning found or common components into analogue circuits and robots. https://en.wikipedia.org/wiki/BEAM_robotics

The circuit used here is a type of very common BEAM circuit called a solar engine which stores charge from a solar cell (or other low current power source) in a capacitor until there is enough usable energy (i.e. the capacitor voltage is high enough) to do some meaningful work. This is normally used to run a small motor in short bursts even when the sun is not strong enough to power the motor directly. It is especially helpful in this particular machine as there is no way a solar panel of this size would be able to provide the current necessary to drive the coil while capacitors have no trouble. Furthermore, even if we had the necessary power, we would not want current running through the coil at all times.

For this project, the word BEAM works on two levels, one being for the electronics style, and the second being that the main lifting mechanism is a lever or... a beam =D

Note: In the past I have tried taking all photos on a white background but I found, with the work I do, it often gets dirty too quick. I have tried a black background here, let me know what you think.

One of the most important parts of this design is the lever as the coil and magnet creates a rather strong force but only when the magnet is close. Hence, we can use a first order lever to translate a strong force over a short distance into a smaller force but over a much larger distance

The other important part of this design is the solar engine which, as mentioned in the introduction, stores energy from the solar panel in some large capacitors until there is enough to do some useful work. I used a SunEater I solar engine which you can read more about here http://faq.solarbotics.net/suneater/suneater.html with some slight tweaks to work with my coil. In the past I have only used a simple FLED based solar engine however they are always finicky to get working correctly and especially with a coil dumping so much power. So I opted for the slightly more complicated, but much more robust SunEater!

You will see on the left of my schematic we have 2x 4700uF caps in parallel, these form the bulk of our energy storage. I used low ESR, 16V capacitors. The low ESR lets the capacitors discharge quicker and I would have preferred 25V as if something does fail and the solar panel is in full sun it can get quite close to 16V and potentially damage the capacitors however I could not find 25V low ESR capacitors this large at my local electronics store. You will also notice a 100uF capacitor which I use to help provide the initial burst of energy as its ESR will be much lower than that of the 4700uF. I'm not sure if this capacitor even helps in this particular build however it has helped with similar circuits in the past so I keep including it. I also have a small, 10nF capacitor across the power supply as without it, sometimes the circuit would lock up and oscillate.

I also swapped out the green LED which is used as a voltage reference for when the circuit should trigger and replaced it with an 8.1V zener diode. This means the circuit doesn't trigger until the capacitors get to around 8.3V which I found is the right amount for my circuit. If you were to build something similar you may have to experiment to get the right value. In fact, in my final design, I left this diode replaceable for future adjustment.

I also expected high current to flow through the final PNP/coil path and although in testing the final design, it seems I could have in fact gone with a smaller transistor, I opted to use the TIP32 power transistor as it should be able to better handle large currents.

We also need a flyback diode across the coil to drain any excess power created by the coil's magnetic field collapsing on itself which could potentially damage components. In my testing I found that enough voltage was being produced to light a yellow LED quite brightly and consistently however in my final design the LED does not light and I am not at all sure why. I think I will try replace it with a red one next chance I get.

Before building anything, I wanted to make sure this project would actually work. I have attempted BEAM projects with a coil and magnet in the past with very disappointing results as it is a much more inefficient way to convert power into motion than say a motor or even a purpose build solenoid. Alas, I had already wound a suitable coil previously for one of the aforementioned projects and really wanted to use it for this one.

I built a crude lever arm and breadboarded the SunEater I solar engine.

In testing I found that I needed much larger storage capacitors than I thought (2x 4700uF caps or a total of 9400uF) charged to higher than I would have liked (around 8.2V). As well, I needed much stronger magnets so I went out and bought some n45 magnets that were as wide as my coil. Even still I was not able to lift the ball bearing very high but the results were promising enough for me to go forward.

Despite me still not knowing if this would work or not I decided to start. Luckily it did work however, as you will see, I test as often as possible while making this to ensure everything is still going as I expect.

Instead of going through the size of brass rods and tubes I use in every step, I will list the general sizes for each application here. All brass rods/tubes are K&S brass as it is easy to find, I like the hardness, and all the rods and tubes are designed to fit well inside each other which I use a couple of times throughout this build

Lever mounts: 4mm square brass tube

Lever arms: 2mm brass rod in 3mm square brass tube

Other lever parts: Various 2 or 3mm brass sheet

Large supports: 3/32 inch brass rod

All other supports: 1/16 inch brass rod

All track: 1/16 inch brass rod

Track switching lever tube: 3/32 brass tube

Freeform circuit connections: 0.81mm or 1/16 inch brass rod

As mentioned, I had built the coil with the hopes of using it in another project. I actually intended it to be used in the brass dragonfly that I posted on here a couple years ago so luckily I have some photos of winding it.

As you can see, I started with a form built up of cardboard, a short length of aluminium tube and some M3 hardware. I then chucked it up in a drill and started winding.

I'm not sure how many turns it is but you can see in the pictures the final dimensions of the coil (I didn't even have callipers back then!) as well as the inductance. Luckily I am studying electrical engineering so I could measure the coils inductance with my universities LCR bridge but it isn't important anyway.

Diameter: 20mm

Height: ~7mm

Inductance: 1.2mH

Resistance (no photo): 8 ohm

Next step was to start building all the brass parts. I started with the lever arm as that is what everything else will be built around.

I drew up a quick design on illustrator and transferred it onto some brass to cut out (I do this with nail polish and a laser cutter but there are definitely better ways).

Next I soldered some 3mm square tube onto the brass pieces. These will hold 2mm round bars which will ultimately connect to the ball holder. I did it this way to make adjustments easier.

Next I pressed in some 8x4x3mm bearings so that I could solder both halves together. Unfortunately I guess the heat from the next soldering steps made them seize up a little so what I ended up doing is just sanding down the brass rod they ran on so that the bearing doesn't even need to spin anymore and really I don't even need the bearings at all anymore! I think I should replace the bearings with bronze bushings at some point so that the steel bearings don't not wear down the brass axle.

In any case, I soldered the halves together with that brass piece with the funny protrusion. The protrusion is there because I didn't yet know how the lever would balance so I didn't know if I would need to add weight to this side or the other one and I was planning to build weight off that protrusion if needed. It ended up slightly balanced too heavy on this side so I ended up cutting the entire protrusion off later, once the whole lever was built.

In trying to free up the bearings I chucked a 4mm brass rod into my drill and spun it as fast as possible. At some point it caught, and the whole thing went flying, breaking completely apart. I spent maybe half an hour solid looking for those parts in the rain (as I spun it up outside) and ended up finding 2 of them in my roof gutters by sifting through the wet mud and leaves.

... lesson learned

The next step was building the lever mounting posts as, once again, everything else will be based on this lever so I need to know how it sits before doing anything else.

To make the posts I used 4mm square tubing and soldered a small section across the top of a longer one. Then I sanded everything flush.

My plan was to have the axel fit into the posts securely but still be removable so I took a 4mm round bar and filed a 3mm square section on each end to fit inside the 4mm square tube. This worked really quite well.

I ended up using 2 small rubber o rings on each side of the lever to keep it centred on the axel.

Next I finished the lever by adding the ball holding rails. I mounted a small section of bent, 1.6mm brass rod (which will be used to make all of the tracks) onto two, long 2mm rods which, as mentioned previously, will fit into the 3mm brass square tube. You will see this also later but I really like using blu tack to hold parts in place for soldering though it does get sticky and messy when very hot.

In order to keep a snug fit in the base I ended up laser cutting the square holes in the base for the posts in order to get them really tight and accurate. I actually designed the hole to be 3.8mm in my drawing so that it would be a super tight fit. At this point I am just guessing on the size of the base and placement of the lever. The base ended up being too small and needed to be changed very soon after this step but it was a great reference for drawing up the final base so I am very happy I did it this way.

As you can see from my Illustrator drawing, I originally envisioned the lever to sit the other way. The problem with this is that the centre of gravity sits above the axle so after lifting the ball it never wanted to come back down to the position I drew it in, it wanted to keep rotating so that it was hanging upside-down. Obvious once the lever is in your hands, not so obvious in a 2D, digital drawing.

Luckily by flipping the lever and mounting the magnets the other way around I got it so that the centre of gravity is just below the axle when set up as I have it. As I found in testing, a lowered centre of gravity is necessary to always pull the lever arm back to its balanced resting state rather than trying to keep rotating. Having the centre of gravity still somewhat close to the axle means that not much extra energy is used lifting the lever, you can imagine if the centre of gravity was much lower, as if there were a large, stiff weight hanging down, lots of extra energy is needed to lift the weight which doesn't help lift the steel ball.

This was the first marble track I ever built so I got a couple helpful tips from this video:

I ended up using 1/16 inch (or 1.6mm) brass rod for the track and an 8mm steel ball. By recommendation of the video, I formed my coils around a 1/4 inch (or 6.35mm) as 4/5 of 8mm is 6.4mm. In the end this actually turned out to be just a little large and I reckon a 6mm rod would have been right for 8mm balls and 1.6mm rod.

Unlike in the video, I found it necessary to tin both parts whenever I needed to solder anything, otherwise my process was very similar to that in the video.

One other thing I found is that the jig is not always the best as is can only be used for straight tracks so I measured the inside of a working track with my digital callipers (at about 6.5mm) and used my callipers to properly space the tracks for turns and my spiral.

Finally now we can start on the track. I begun by bending some of the 1.6mm rod around a 10mm form such that the ball will fall down through the hole. Then I straightened out the ends slightly narrower than the width of the ball for the rest of this section of track. Finally we add the half round supports

I then started putting the track on the base and realised the base was way too small! It was still a great time to test and see if this would work as at this point I was still unsure. With confirmation that the lever could lift the ball and, just as importantly, go back down by itself, I moved onto making a final base.

So I designed up and laser cut a larger base. There are 4 horizontal layers of 3mm plywood surrounded by 2 layers of wood. I did it like this as I wanted all my brass supports going through as much wood as possible and to be super solid.

But I also wanted it to look nice so the top 2 layers of wood are glued like a box lid so they can slot in without tabs. The outside, side panels also have minimal tabs. And the whole thing was designed to have a slight lip so that any stray balls stay on the base and don't get lost.

In the first photo you can see all the parts. In the second you can see the tabbed parts of the base glued together without top and sides (ignore the extra slots). And in the last 2 photos you can see the lid-like top I made.

Next I spray painted the outer parts of the base black. Remember to start with light coats and don't forget to flip over the sides and spray the top of the inside of the sides as they will be visible as the inside of the lip. I knew I wanted a black base but didn't think flat black would cut it so I went back with sandpaper and did a couple layers of paint and sanding back to get to the final look.

Then finally, I pre drilled some holes for brass nails. These were always supposed to be decorative but I decided to use them as real nails, to hold the work while my glue set. I also imagined the black wood "skin" would slide off after gluing and not be glued to the base as I like everything to be removable. Unfortunately, because I am really bad at wood working, the nails were too close to the inside surface of the outer wood pieces and ended up pushing the entire outer skin apart and away from the inside box. So I quickly decided to remove all the nails and glue the outer panels directly to the base such that I can put the nails back in properly later once the glue had dried.

Later on in the project I decided I didn't like the bright white wood against the black paint as much as I did when the wood was slightly darker (whenever I wiped it down with rubbing alcohol it looked so much better) so I oiled the entire base with 2 coats of boiled linseed oil. It also gave it a nice, natural sheen which I was very happy with. I added back in the brass nails once everything was dry.

I was really happy with how this base turned out and I will likely do the same worn, spray paint look on future projects!

To mount the coil in the correct position I was originally planning on using laser cut wood supports but I felt that would cover too much of the mechanism. Instead I came up with what I think was a really elegant "claw" design inspired by stone setting techniques. I filed the square tube end with a round needle file so that it would fit better over the support. By bending the arm that slides into the square tube slightly there is enough friction to keep it fixed in there really securely.

The next step was building the lever to switch between the two tracks. I wanted this marble track to be as dynamic as possible for such a small track so always planned to have 2 paths.

This was quite a tough part to solder up as there are so many joints in close proximity so I had to set it all up with a helping hand and blu tack so that I could solder it in one shot. I held the top part on with pliers but I probably should have used another helping hand to hold that in place also.

I also drilled a small hole in the 3/32 brass tube which was actually rather difficult as the drill bit was only 1.2mm and made of carbide (for drilling PCBs with a CNC). However I felt it was worth doing to more easily oil the lever once fully assembled.

Important: More fiddly lever things. I found the opposite for this lever as compared to the marble lifting lever. You never want it to find its balance. As such, the side arms should be short enough that the centre of gravity is above the pivot so that when the lever falls to one side it wants to keep falling to that side. if the side arms are too long and the centre of gravity is low it will not stay to each side when pushed by the marble.

Next up is mounting the track switching lever and supporting bits and bobs. A short section of 1.6mm brass was soldered to the mounting post for the lever to pivot on.

I also made a brass guard rail in case the balls come off the lever with too much velocity (not a big problem in my testing but it did happen). As well, this section of brass bends down and underacts as the stops for the switching lever.

I also cut a small section of brass with a hole in it to solder onto the rail and hold onto the other end of the lever axel without having to solder to the lever axel. This is because I knew I would have to clear coat the lever disassembled in order not to gum up the mechanism and I wanted a way to secure everything once it was clear-coated without heating the axel and in turn heating the already coated lever.

I do eventually cut everything down to the correct length I just see little reason to just yet =)

Now it is time to start on the two tracks! First I found a bunch of round objects to wrap my wire around. This takes trial and error more than precise measurements as the brass wire is elastic and wont form completely to the round form. It's also quite easy to slightly tweak the radius of the curves after bending. Shown are the 4 round objects I used for this, and the spiral track.

I had to make the track in two sections. I build each section by making the appropriate bends, lining the rods up on my jig, and then soldering on the half round supports.

I then sanded the joint between the two sections flush, broke all the edges of the tracks to be joined with a needle file, and soldered them together. I found I needed lots of solder for this step if I wanted to make anything close to a solid joint however I also found that you really want to be careful so as not to let solder build up on the inside of the track and inhibit the balls path.

Next I laid out the track directly on the base and marked where I wanted holes. Of course the track will be slanted so these hole positions are not perfect however I didn't feel like taking the whole track off every time I wanted to drill a new hole so I just made my best guessed on hole locations and drilled them all at once. As mentioned previously, I wanted a tight fit on all my supports so I drilled the holes out to 1.5mm and forced the 1.6mm rods in.

Then it was just a matter of cutting the support rods to size and soldering them. I cut them all slightly shorter than measured so that I not only could raise, but also lower the track once everything was together for fine tuning. This turned out to be very helpful and the holes in the base are tight enough that it is all still incredibly solid.

Very similar deal with making the spiral track except it is much harder to use the jig for laying out support spacing or for keeping the track width consistent so I relied on my digital callipers. Again the track was made in two parts and soldered together

Mounting the spiral track to the base as I did the zig zag track, the track is now finished! perfect time to test everything still works

It was at this point, where after so much work, I could finally see the finished product in my head and everything was working exactly as I had hoped. I was ecstatic.

Added a couple final things before clear coating the brass.

1. Added some supports to the ends of each track, not for structural reasons but because I noticed that with these extra supports, the ball will not be able overshoot the lever arm and will instead hit the support and fall into the lever. Though my final track was slow enough that this was never a concern, it occurred sometimes in testing.

2. Added extra support to the coil arm. This was always the plan as there is a lot of stress on the coil and lever arm.

3. You may have notices some blu tack on the lever arm while testing. This was to perfectly balance the arm. Obviously I did not want blu tack in the final design so after weighing the blu tack with some precise scales I soldered on a length of brass and sanded it until I hit the perfect balance.

4. Not pictured, but after clear coating in the next steps I put some black heatshrink over the post where the lever hits to stop its momentum. This was added to protect the clear coat, soften the blow and lessen the noise.

I always wanted to clear coat everything I could as I knew I may be leaving this marble machine outdoors. I found this brass and copper specific clear coat in my local hardware store and used that.

As mentioned at the end of the track switching lever step, I knew I had to clear coat the lever first and then solder on the final support and mask it all up to clear coat the rest. Otherwise the lever and axel would have definitely gummed up from the clear coat.

I prepared all the brass and masked off the bearings just with some blu tack then sprayed all the brass parts. The spray worked very well for the lever, however it left a very bumpy surface on the tracks and other parts. I think maybe this was because it was too hot on the day I sprayed the remaining parts and the coating flashed off before evening out.

In any case, it left a matte surface on the parts which was also very rough however it smoothed out very quick with some wet sanding with 1500 grit sandpaper. I then went over the track portion with 2000 grit to make sure it was as frictionless as possible for the ball.

I am happy with the clear coat as it seems super durable and worked great on the track switching lever however, even after sanding smooth, the rest of the parts are noticeably duller than the lever or unsprayed brass parts because of the way it dried. If anyone has any ideas how to fix this for next time, let me know.

Now we will move onto the electronics. I start by mounting the solar panel as that is what the rest of the electronics will be built off.

I laid out the supports in place to make sure I would be happy with how everything sat in the end. Then I roughed up the solar panel with a file in the places I knew I would be putting epoxy. I will need to epoxy the supports to the panel as it will be quite easy to bump once assembled and it on a tall support, both things that will threaten to rip the panel off the supports. I also took away some of the solder mask in places to make a couple more solder points for even more support.

Unfortunately, if I went into depth, free-forming the circuit would be a massive Instructable on its own (and I think I will make one for free-forming at some point). On top of that, it is more of an art than a set of instructions so I think it would be best to just describe my process rather than individual steps.

For this particular circuit I knew it was a tricky one to troubleshoot as it is hard to test sections at a time which is how I usually like to build. As such I decided to try lay it out as close to the way I breadboarded it as possible to minimise risk of mistakes.

I start by making my positive and negative rail. Often this is a very good place to start. Although it may not always be the most elegant and aesthetically pleasing way to lay out simpler circuits, it is very useful for slightly more complicated ones. Each rail is supported on the opposite side through a small, 10nF, axial capacitor. This too is something I often do as not only is it a great way to massively boost the rigidity of the circuit, I also found in testing that I actually needed to put a 10nF capacitor across the power rails of my prototype otherwise sometimes the circuit would not properly trigger and would oscillate.

Then I started soldering up the components similar to how I would connect them on a breadboard. You can track how I went about this through the photos. One strange thing I did is use a socket for the zener diode so that if, in the future, the caps or coil become weaker, or friction builds up in the lifting lever, and the lever can no longer reliably lift the ball, I can change out the zener diode for one with a higher breakdown and hopefully keep the machine working without massive changes.

Important: You may notice I have a resistor across the positive and negative rail. This is a small, 100 ohm resistor that ensures the solar panel does not start activating the circuit before it is finished, potentially ruining components. Just shorting the positive and negative of the solar panel would work also.

The final decorative touch, and one I think ties this whole thing together, is the brass name plate. I first drew my design in illustrator and then cleaned and painted an appropriately sized brass off-cut with nail polish. I then laser etched away the nail polish where I wanted the final part to be black and left the polish where I wanted the brass to show through. I then mounted the brass to some foam and left it in a mix of 1 part 30% hydrochloric acid to 2 parts 3% hydrogen peroxide. I forgot about it and left it in too long, about 2 hours, so some of the letters and edges are wonky but I think it adds to the charm.

Finally I stamped my name and the year, filled everything in with nail polish (proper enamel is what should be used but I knew I wanted to clear coat everything anyway), and sanded back to the raised brass sections once the nail polish was completely dry.

I am super happy with how this one came out however I do have some future improvements I may make. I think I will add a switch that simply shorts the solar panel through a small resistor so that I can effectively turn off the device as that is a feature lacking on most of my BEAM circuits.

I also noticed that in full sunlight, the circuit charges faster than it takes for the ball to roll back down the track. If I had known this could be the case, perhaps I could have designed for a way to use multiple balls however for the time being I am trying to think of a simple timing circuit that won't allow the circuit to trigger again until enough time has passed. The problem there is that the extra energy still being produced by the panel has to go somewhere. Maybe I will make it spin a motor for a second, we will see =)

In any case I have the circuit sitting on a shaded balcony so for now it is working great.

I also still definitely need to clear coat the circuit however I will wait to see if I add these other improvements first.

Thanks for reading!