Hi, for my current digital modelling class i have been given free choice for my final assessment so i decided to kill two birds with one stone and use the opportunity to model out a blueprint for a perfect infinity dodecahedron. But this one would be different, all the controllers, batteries and circuitry will be inside the frame, no addons, no leads/wires for power, no stands, no blacked out fake-faces, just a perfect normal self contained dodecahedron!

An infinity-shape is essentially any geometric shape with tinted mirroring and lights on the inside, they almost endlessly reflect around and around creating trippy light patterns. Normally though the frame doesnt contain anything more than the wires, and portable ones with built in controller and power are never dodecahedrons, either they have a box hanging off, a stand, or one of the faces is extruded 1-2 inches thicker to hide a beefy controller, which also ends up blacking out that face so you kinda have to ignore it.

My friends and i decided to build one too, however, when i was being roped in, i realized that the frame would have a lot of free space, i had recently become acquainted with WS2812B RBG lights, controlling them, audio reactivity, microcontroller PCB design (or rather, how easy it *can* be if it only has to do 1 thing), i had also recently learned about different sorts of batteries, and concluded, that if we were to do this thing, making a commercially viable infinity dodecahedron, it was entirely plausible to fit the entire circuitry inside the thing, rather than hiding a power supply and microcontroller off somewhere. That is the fundamental objective of this project, internalizing EVERYTHING.

The requirements for the portable infinity dodecahedron would have to be as follows:

• frame must house STM32F4 chip
• must contain batteries in each edge
• must contain battery charge/regulator/boost converter (doesnt need boost if 3v LEDs can be sourced)
• Must still be structurally rigid with a frame no larger than is typically seen in 3D printed frames.
• edge length of the windows must be 250mm
• must be USB rechargeable

These constraints are fairly plausible.

The STM32F4 chip; specifically looking at the F411 comes in a maximum form factor of LQFP100 which is 14x14mm, odds are on the pcb it will have to sit diagonally, though, only a handful of the 100 pins will actually be required. this sets the constraint at sqrt(14^2 + 14^2) = 19.8mm assuming its a fully 45 degree diagonal angle. If working with the LQFP64, which is 10×10, then it would be 14.3mm. from my breif glance i think the 64 has 64 pins, which is acceptable since this thing will only have 1-2 input and output devices. the PCB on a diagonal chip can be as wide as the chip, so, 19 or 14mm.

Batteries; the batteries i have selected are standard form factor cells, though i suspect if i look hard enough or am willing to have them made to order (or standard sizes from a made to order source) i can probably get something pretty good, 10mm wide cells of varying lengths exist for example. In any case, what im going with is the 10440, a 10mm wide cylindrical battery, or, the 08540, a 7mm wide cylindrical battery, but that assumes i can actually find them, they are used for mini-vape pens but are oddly difficult to actually source for some reason. either way these are pretty standard lithium ion cells, ignoring their size. a benefit of the 10440 is if it is in a steel canister it can be glued into a hole cut in the frame to take on the structural role of the plastic it replaced.

Battery regulation and charging; will be handled with internal and external circuitry. First, the battery protection boards, these shut off in the event of a short circuit, or over or under voltage. these come pre-fabricated and the power demands are low, so the ones normally so tiny they get embedded inside the cells themselves will be adequate, they are usually no more than 4mm wide, they are of negligible size and dont need customization. Power regulation however is another matter, i have found a 5v in 5v out battery charger/booster, the IP5306, it has a simple enough circuit that i could custom make one, plus pre-fab boards are small enough i may be able to make do with just them, however they seem to lack a stepdown converter as there is only 1 inductor, which would be used for boosting the 3v battery back to 5v, meaning they are just a linear input regulator, meaning they convert excess power into heat.

The need for a 5v line however is rendered moot if i can find ws2812s which can run fine on 3v, technically they already can, i have confirmed this, but not effectively, and certain colors would also be skewed as a result. alternatively i could also just find some other RBG LED entirely but then i would have to make a port for the FastLED library that it will use.
The entire frame will be filled with batteries, and if 5v is needed, will be filled with numerous weak 5v boosters that add up to more than enough for a stable system. It may also end up working out that using 3v directly is acceptably dim and is an ok tradeoff for running in portable mode. plugging into a USB port for when you need to run it at full brightness.

The frame; ultimately i will want to commission something made of metal, or injection moulded, or possibly even carbon fiber, or plastic reinforced with carbon fiber or metal rods for rigidity. Early designs will be 3D printed. after a prototype is finished i will look into sourcing existing compatible materials like certain standardized extrusions for the frame.

Charging; can be handled in numerous ways, i may not end up using the IP5306 at all, truth be told it seems pretty dodgy plus no english datasheets exist. The first method involves having a diode separated 5v rail, using diodes so the batteries dont constantly try to recharge themselves using their output.
2nd method involves getting a different more reliable charger iC for each cell, and shifting the boost converter work to some dedicated boosters.
Another way is simply hook up every battery in parallel and include a built in safe charger/5v supply for running off external power AND adding a direct line to the batteries to use an external charger that can charge far quicker than the safe charger could, using a proper high quality lithium charger.
Any combination of these is viable, but i want to see how the prototype works with the 5v in 5v out so that running LEDs directly off the batteries is never an option.

Future implementable ideas include lasers, sound reactivity and physical interactivity.

• Miniaturized ultrasonic sensors can be implemented and their signal input can seed responses in the animation patterns.
• INMP441 i2s microphones are ridiculously small and can be used for adding audio reactivity, hell, even full synchronization and frequency response
• infrared sensor/emitter arrays can be added into the frame in a similar fasion to infrared touch screen kits, like this https://www.aliexpress.com/wholesale?catId=0&initiative_id=SB_20200917170252&SearchText=monitor+IR+touch+frame
  The result would be that, using a very basic/simple array it could approximate something like 30 different locations you might touching on the panel based on which beams are being interrupted
• Lasers could be neat to add, a collimation lens can be glued onto a fairly strong red diode to improvise a beam, though it wouldnt be visible, and using a beam strong enough to be seen would also be very irresponsible, nontheless i will try and see what happens if you shine lasers in an infinity dodecahedron
• accelerometer; an incredibly trippy effect would be using an accelerometer to play a full 3D pattern (with respect to the the 3D geometry of the dodecahedron) and having it oppose any movement, meaning your room would have a projected pattern that can only be seen inside the dodecahedron and youd have to move it around to see, i.e, requiring 1M of movement to the left to find a specific blinking green dot, move left, its gone, move right, its back again. It would be like a viewport into another dimension except the pattern is static, or at least changes slowly, so, it wouldnt look auto generated at all. it would actually look like you were peering into another dimension.
• electronic compass; similar usage as the accelerometer, i say compass but really its any chip or chip combination capable of discerning if its upside down or not, this could be used so that only the bottom 3rd of the lights illuminate, ignoring the reflections it would be like the dodecahedrons lights behaved like water, constantly either always only occupying the bottom, or, pouring into the bottom. a combination of reed switches and good management of accelerometers would accomplish this, using the multiple reed switches to determine absolute orientation.
• disco ball; a disco ball suspended inside is something i want to see what the effects for might be, though i suspect theres a good chance it might overwhelm the patterns inside. If it has a positive effect i will try and make one panel openable to place the disco ball inside, or otherwise offer variations, or at least make one panel not rely on glues to stay shut so that it could eventually be reopened safely.
• wifi, bluetooth, 2.4ghz radio and IR control options. wifi would be accomplished through an additional ESP8266 chip, bluetooth chips are tiny, whole chips are as big as the stm32f4 so no worries there. 2.4ghz transceiver boards are the same, so plug-n-play is plausible allowing a VERY easily constructed serial transmitter to be used to connect the dodecahedron to a computer or phone via special dongle. this could potentially also be abused to let you communicate with your mice or keyboards, or game controllers too though not in a meaningful way.
  lastly IR, this would work using a common TV, air conditioner, sound system, or etc, remote. you could program the dodecahedron to play certain patterns when it sees certain IR transmissions. any device capable of reading IR can also operate in a sort of learning mode, basically all devices with remotes see 100% of transmissions, they just choose to ignore ones they dont recognize as their own. this means every single button on your TV remote can be programmed in as a trigger for whatever you want, you would set the system to start recording, press your remote button a few times, and bam, now if it sees this, it will know to execute a particular function, or apply a modifier, change a variable, etc. if you collect a bunch of old TV remotes each button on all of them could be mapped out to do whatever you want. most commonly would be making the animation speed faster or slower, or turning the brightness up or down. you could also build your own remote controller too using an arduino with a little LCD to pick out patterns but at that point the 2.4ghz transceiver would be a better choice. IR would simply be a matter of convenience since literally any remote control would work.

Ultimately we are trying to make a commercial product that performs vastly better than any of the free DIY designs around, or manufactured versions of those designs. Or just releasing a kit with the frames to self-assemble.
And now for my class im taking the first steps by modelling all the geometry in solidworks. Ultimately i dont think it will be able to simulate it, this is moreso just for creating 3D printable frames and making more of a technical model than a virtual prototype. everything will be to scale and ill be able to see in what ways everything can fit in the frames without getting too thin or weak.