Tiny, Solar Powered, Light Seeking BEAM Bot (Mini Photopopper)

Today I will show you how I made this tiny little robot!

Since I first mentioned the word in an Instructable and got a couple questions about it, every new BEAM Instructable I feel the need to explain my understanding of the term "BEAM robotics". Wikipedia will tell you that BEAM stands for Biology Electronics Aesthetics and Mechanics, and though I have heard conflicting theories about the name in the past I do feel that BEAM embodies these aspects. It is a style of electronics developed in the late 90s and early 2000s using mostly found components in ways that are quite elegant or clever, in order to create the most basic intelligent robots.

This particular style of robot know as a photovore, phototrope or photopopper, meaning it uses some very simple logic to locomote toward light.

Making tiny photopoppers has been done before, indeed even I have tried (in the video you can see my attempt from 2018 built on a PCB I designed), but I am really happy with how this particular one turned out and am excited to share the process with you all!. I have included links to other examples below, but I wanted to make this new bot for a couple reasons.

1. I knew I could do better than my predecessors just because these (relatively) new IXOLAR monocrystaline solar panels from digikey seemed to have incredible specs and oh boy did they work awesome in the end! Solar panel efficiency is tied to surface area so this means when you shrink a panel in both dimensions, you get much much less power. But these IXOLAR panels are 25% efficient which is crazy! And not only that. they claimed to work fine in partial shade or even indoors which is something that other monocrystaline and polycrystalline cells tend to struggle with. You will notice in the video that I actually have another photopopper using very similar parts but I used an amorphous cell for that one because I wanted it to work indoors (where amorphous cells shine) and I was still unaware of how good these monocrystaline cells had become. Again, you can see for yourself in the video, but the new cell is smaller than the old one, AND it pops around twice as fast indoors! Though this may also have something to do with the lower trigger voltage for this new bot =P.

2. This point also ties into the solar panel power conundrum but I actually wanted this guy to move like bigger photopoppers so I was happy to sacrifice some miniaturisation for good functionality and again, I think this project achieved that. The other examples either wouldn't move too well or are still a little big but I really wanted to try make this guy as similar as I could to bigger photopoppers...just smaller! I could maybe have used smaller parts like some BPW34 cells, and some more clever circuitry/special motors to deal with lower voltage, really Im not sure, but more than anything I really just wanted this guy to be as small as possible and still move "properly"

Photovore here on instructables using a flexible amorphous cell and vibration motors:

https://www.instructables.com/Nano-Photovore-BEAM-...

And a great, albeit short, write up on a much more similar photopopper to this one:

https://www.fetchmodus.org/projects/beam/photopopp...

A super cool example using 8x BPW34 cells and an aluminium cap. Seriously, if you haven't seen Tasi Geri's videos check this out! Such incredible production value... but still a little big for my liking ;).

https://www.youtube.com/watch?v=3RKTKfLhuPE

All of these were great inspirations for if this was even feasible as well as for specific parts (I will touch more on components in the next step).

2x Voltage monitors: MCP112-195

For this particular photopopper I chose to use a voltage monitor that triggers at ~1.9V, with the added voltage drop in the diode and the voltage added by the photodiodes, this means that the circuit triggers when the storage cap reaches around 2.3V. In my previous attempt I used MCP112-270s that trigger at 2.7V which meant the capacitor needed to charge to around 3.1V! This is unnecessarily high for the motors I used but may be right depending on your other component choices.

2x Transistors: BSR14

And decently rated, sot23, NPN transistor should work

2x Capacitors: 470pF 0603

These capacitors provide some hysteresis for the voltage monitors so that they dont just shut off as soon as current starts flowing through the motor. With the MCP112 I found 470pF worked well but you may have to experiment if you use other voltage monitors.

2x Photodiodes: VEMD6010X01

Again, any photodiode would work, but as I was hoping this circuit would still work to some extent indoors, I opted not to use IR specific photodiodes that may track the sun fine but not other light sources. In the end the circuit doesnt work under just indoor lighting anyway so really any photodiode should do.

2x Diodes: BAV70

You can freeform this same cricuit perfectly fine, probably even a little easier, with and 2 SMD diodes but when I first bought parts to make my PCB mini photopopper I thought I would be really clever and buy a pair of diodes in one sot23 package! So thats what I used

Solar Panel: 5.53V, 26.3mW, 23.00mm x 8.00mm (KXOB25-02X8F-TB)

Storage Capacitor: 2200uF, 6.3V tantalum capacitor (592D228X96R3X2T20H)

A 1500uF capacitor may work fine and 2x 1500uF in parallel is another option that would definitely work (as 1500uf SMD caps are easier to source). This ended up being the most expensive single component, much more so even than the panel so we will have to see how it holds up over time.

Motors: 1.5-3V, 4x8mm vibration motors.

I grabbed these motors for super cheap off ebay a while ago but cant find anything else similar. Being flat and with the little nubbins to solder to they worked great for this circuit but you may need to shop around for a suitable replacement if you wanted to make your own.

Brass wire: 0.5mm

And finally you will need some thin brass wire to make some of the connections

Note:

The "tiny photopper" I linked to in the introduction uses much better components (i.e. a voltage monitor with much lower quiescent current and MOSFETs instead of transistors) but I already had these parts from my first photopopper and I wasn't thinking so much about component choices back then. All this to say, if you were to attempt this project, you can definitely find more suitable parts that will make the circuit work better, but unlike in the "tiny photopopper" example, our panel produces enough power that we dont have to worry quite as much about quiescent current consumption. If I were to do this again and buy all new components I would look into using this voltage monitor IC but I would personally not use the dual MOSFET and would try to find a MOSFET with an equivalent pinout to the transistors I used as the freeforming with the components I used turned out great.

We are going to start freeforming by building up a stack consisting of the 2 voltage monitors and 2 transistors. First is a voltage monitor and I bend up pin 1, or the output, to make contact with the gate of the transistor.

Next we must add the transistor. To make the final circuit more robust, but mostly to make soldering easier, I first glue the two packages together with some super glue. I do this for each sot23 part but will just show my set up this once. I use some blu tack to hold the base part in place and some tweezers to place the second part on top. the surfaces are super tiny so instead of trying to put a drop directly on the components, I squeeze glue out onto a piece of tape on my workspace and transfer just a dab of glue over with the tip of a skewer.

Superglue bonds break down with heat which was not a problem with soldering these sot23 packages but is definitely something to keep in mind.

Once the two components are glued together, we can solder the bent up output of the voltage monitor to the gate, pin 1, of the transistor.

Finally we bend up the emitter, pin 2, of the transistor such that it can connect to the emitter of the second transistor.

Continuing as before, I glue on another transistor and solder both pin emitters, pin 2, of each transistor.

Not shown, we also must bend up the gate of the second transistor to make contact with the output of the final voltage monitor.

Finally for the sot23 components, we must glue on the final voltage monitor and solder its output to the gate of the previous transistor, pins 1 on each.

Then we bend up Vdd and Vss, pins 2 and 3, of the final voltage monitor, so that we can solder the capacitors and photodiodes on.

Next we add the capacitors to either side of the voltage monitors. Each capacitor connects to Vcc and Vdd, pins 2 and 3, of their respective voltage monitors.

I used the same super glue then solder trick for these capacitors but because they are much smaller the super glue broke down much more easily and it didn't work as well.

Next we have to add the diodes. As mentioned in the component list, this could be done with and 2 general purpose smd diodes.

The legs of the BAV70 are not quite long enough to reach where they need to connect so I had to use some brass wire to connect them. The brass looks too thick in these photos but it is the thinnest brass wire I have at 0.020" (0.51mm)!

Little bit of a continuity error here in the last photo, the next step should be to connect the emitters of the transistors to the cathodes of the diodes to arrive at out final ground connection but I actually did this after I put on the photodiodes. Would have made more sense to do it here though.

The final part of freeforming the brains is to add the eyes, 2 smd photodiodes. The cathode of each photodiode goes to Vdd, pin 3, of each voltage monitor.

Then I bent some more brass wire to connect the anodes and to form our positive connection point. I forgot to photograph anodes connected but you will see how it looks in the next steps.

Finally we can test the circuit and see if everything is connected properly. To do this we simply connect one side of the motor to the collector, pin 3, of one of the transistors and the other motor lead to our positive point. Then you can connect the positive and negative of the circuit to a large (1000uF+) capacitor slowly charging through a resistor. You can just connect the circuit to a power supply but doing it this way mimics the way our circuit will be connected eventually, to a capacitor charging via a solar panel.

When I first tested nothing was happening and I thought I screwed up! but turned out I just had my power supply set too low and light was hitting the wrong side of the circuit so it was trying to turn on the motor that wasn't there. Phew.

Next we can prepare the motors. Learning from my previous attempts at mini photovores I arrived at this set up for the motors. I cut some 3mm acrylic into a quarter circle shape and glued the motors onto it first to provide some structure.

For these particular vibration motors I also had to remove the minimal extra housing they had and the off-balance weight.

Now that we have all the parts I decided to try laying them out to see how to best continue. Using some blu tack to hold the capacitor I arrived at this layout. It covers the circuit a little but I think the final aesthetic is as balanced as it can be and it ended up being very mechanically solid too! In the end I was very happy with this solution despite it covering the circuit somewhat.

We can now start actually connecting everything together. To start with I cut and bent another section of brass to connect the circuit to the positive of the panel.

Next I bent some more lengths of brass which will end up being the motor mounts. So they have to route from pins 3 of each transistor to some place near the positive of the solar panel.

Then I soldered the motors to these connections and the positive of the panel. This turned out to be a really great solution. It is simple, compact, and very solid.

In the final photo you can see how the brass connections we made have connected to the motor

A nice easy step here, just solder the negative of the circuit to the negative of the panel. At this point I also added the 3mm yellow led at the back to act as the rear "wheel". It doesn't do anything electrically, it is just there for mechanical reasons. This "wheel" only really works on smooth surfaces but it's not like this little guy will be able to handle rough terrain anyway ;).

Final component now! The capacitor! I soldered a length of brass wire to the anode of the capacitor (noted by the small nub), and then soldered it to the photopopper. It took quite a bit of heat to get the negatives connected and by the time it was all done I realised the capacitor had shifted a little so it probably would have been a good idea to set up some kind of jig for this step to keep everything aligned.

I got a little ahead of myself with excitement and did this step without taking a photograph because I was so keen to see if it would run. The problem with using motors like this is that there is not much grip, so to remedy that I stripped some breadboard wire and pushed the insulation onto the motor shafts.

The hard part about this is that every little bump at the tip of the wheel will make the bot jump and not move as well, so I spent some time after initial testing with a lighter and then an exacto blade, smoothing out the tip of the wheel. It still jumps sometimes but I like it, gives him a little extra character (not that it was ever in danger of lacking character!)

Look at that little cutie pie!

Closing thoughts:

I have a spare panel and 2 spare 1500uF tantalum caps so I think at some point I will make him another friend =) but I have many other projects in the pipeline so that will have to wait.

Obviously at this size there is no space for a potentiometer to tune the light tracking of this circuit so you just have to hope (or buy a couple and test) that the photodiodes are well matched.I also tried to use some nail polish to partially cover one "eye" but this was too finicky and didn't seem to do much until it did too much if that makes sense. So I decided to just leave this guy tracking a little wonky =).

Also it only just dawned on me that my username fits this project perfectly.