I have a slightly movable (torsional (freely for infinite degrees around the z-axis) and translational (+- 40mm in the xy plane)) microcontroller in communication with a fixed one. I'd like to use it without having a dedicated battery. Cables would transmit too much mechanical force (own weight, spring-action), coils would pick up too much noise and require ac/dc conversion in the mobile component, so I was thinking about using a solar cell on the mobile component for power, and either a dedicated TX/RX part, or to solve the RX with the solar cell as well (there won't be much communication, so the (off-)duty cycle won't be much of an issue)
There are multiple papers about the principle, but I am stuck on a practical issue: the mobile component will need about 20mA @ 3.4V (~0.07W) continuously, and i cannot find the values i need to compute whether this is feasible in a small package.
A silicon solar cell has about 10% efficiency, but that is for full spectrum sunlight, and I would be using an LED at 950 nm (because silicon solar cells seem to be more efficient there - only some sources say the reverse, which might be due to NIR not having the most energy in sunlight -), and I’d ballpark an LED at 50% plus the solar cell would only get half of the light at any given moment, to account for movement, so I’d have to power the LED (array?) with 40*20 mA=800 mA, which seems a lot.
Also, I cannot seem to find upper limits for radiation densities for small-spectrum light on silicon cells (amorphous? Crystalline?) - is it the same as for full spectrum, ie 10000W/m2 ?
Alternative power conduits are welcome, but my main focus is on scaling the LED(laser?) and solar cell.
EDIT: there will be plasma cutting and arc welding going on in the near vicinity, which is why i thought coils were out. The welding and cutting equipment is not related to the project i am talking about here, so neither can i use the resultant light, nor can i use the infrastructure (cables). The MCU itself is shielded, and has functioned properly. Currently it is powered by a small lipo unit i got from a minicopter. I experimented with the coils from an electric toothbrush, but got bad results for the translational movement (did not try with welder on ). Movement is +-40mm in xy plane, and torsional movement is constrained to the x-axis. air gap of >50mm would be nice. There is no MCU is 10g, shielding is another 12g (though i'll try to shave that down as soon as i get my hands on thinner metal sheets), lipo is 7g