Following on from the single axis solar tracker I built recently I would like to make a dual axis solar tracker – but with a twist: I would like one axis to be a set of tank treads. This way one can eventually imagine an autonomous solar powered robot which tracks the sun and maybe can complete other tasks.
After making the single axis solar tracker from logic chips I decided there was a limit to how complex a system could be created in this manner. There were a sufficient number of connections that debugging the system was discouraging. I would like to continue the sleep project, which uses an Attiny2313 and a simple nocturnal solar charging circuit to play a user-defined piezo piano tune. I would like in this project to use an Attiny2313 to control the two axis solar tracking. This will involve a series of mosfet H-bridges (I canāt find commercial H-bridges that are low power enough at the moment), and a capacitor to smooth out the spikes produced by the motors which could threaten the microchipās power supply. It may also require the use of a more sophisticated system for running a chip and motors off a solar charged super capacitor, such as the LTC3105 solar harvesting chip.
The first test is if a solar engine can boot up a microchip (not awaken from sleep as the power supply is zero after a solar engine firing) which can then check a sensor and activate (latch on, really) an H-bridge which will allow a motor to turn until beyond the point the microchip loses power and shuts down. However, this is currently not working and Iām not quite sure exactly why at the momentā¦
Here is the breadboarded circuit with Attiny84 and h-bridge all powered by a solar engine.
The h-bridgeā¦
ā¦and the solar engineā¦
My first step was to figure out how to attach a motor to a solar panel and then to a post which is attached to a tank movement system? Here are some of my attempts below. I am using a robot tank tread platform I developed a few years ago and which I describe at this instructable: https://www.instructables.com/id/Mini-Robot-Platform/ . The 3D model of the chassis is available there, it is printed on a resin printer and pairs nicely with a popular gear motor product. The robot platform has limitations: it doesnāt like slippery surfaces and the gears do fall off the rollers occassionally but itās compact and I already have leftover chassis so it makes sense to reuse it.
This is inelegant and takes up too much space:
This is more compact but destroys itself if you turn the motor too much.
This might be a decent balance but itās still heavy.
This little motor would have been perfect to turn a solar panel but the tiny shaft is difficult to attach to the panel.
Here Iām using an allen key which is super glued to the flat side of a different motor shaft. Works for me!
Moving on to using the LTC3105 board now. The goal is to get a microchip running with the LDO output by the chip, and then to have the microchip activate the motors with an H-bridge which is connected to the Vout of the LTC3105.
Iām combining several typical application circuits which are presented in the datasheet and have made a small PCB which will serve as a demoboard to do some tests.
Here is the demo set up which is currently not working:
I 3D printed some nice thingiverse 2 axis pan-tilts which use mini-servos:
https://www.thingiverse.com/thing:1401116
https://www.thingiverse.com/thing:708819
Iām looking in to which voltage and current rating I should chose for the solar panels.
Part Number : KXOB25-05X3F Open Circuit Voltage [V]: 2.07 Short Circuit Current [mA]: 19.5 Typ. Voltage @ Pmpp [V]: 1.67 Typ. Current @ Pmpp [mA]: 18.4
For the LTC3105 the highest efficiency is around 2.25V (between 2-2.5V) and drops sharply off at 2.5V. 2.07V is at 85% efficiency which I think it good enough. I have found that solar panels sometimes go above their OCV too.
Quoting from the datasheet: http://ixapps.ixys.com/DataSheet/KXOB25-05X3F.pdf
āMonocrystalline cells, such as the IXYS SolarMD, have a spectral sensitivity range from 300 nm (near-ultraviolet) to 1100 nm (near-infrared), which includes visible light (400 to 700 nm). Due to this wide spectral range, they can be used in both indoor and outdoor applications. Monocrystalline or single-crystalline material is the most expensive but it does not contain impurities, and as such the power conversion efficiency does not degrade over operating time. The power conversion efficiency of commercially available monocrystalline cells ranges from 15 to 25%. The surface of these cells is a homogenous dark blue or dark grey.ā
Apparently there is about 1,360 watts per square meter of sunlight. However, āAveraged over the entire planet, the amount of sunlight arriving at the top of Earthās atmosphere is only one-fourth of the total solar irradiance, or approximately 340 watts per square meter.ā (https://earthobservatory.nasa.gov/features/EnergyBalance/page2.php). Solar panels are 15-20% efficient so, on average, 50-70W ish.