Given the majority of the power consumption of a blinker is leakage current across the voltage-limiting zener, I thought I would try reducing the number of cells to bring the maximum voltage down under the voltage limit for the supercaps, and leave out the zener.
The 5-cell arrays I build generate just under 3.1v in direct morning sun. The voltage quickly drops as the cells warm. That's 0.62v per cell. The supercaps are rated for 3.0V, anything above this will degrade them. I will put together a 4-cell array, remove the zener and run some tests.
I hooked up a multimeter across the capacitors and logged the voltage over time. The modified blinker charged to around 2.2v in the sun. Once the voltage plateaued I moved it inside and continued the logging. From previous testing I know that the LED will continue to blink until the capacitors discharge to around 1.45v. About 20 hours after bringing the blinker inside the voltage had reduced to around 1.8v, and the LED was still blinking.
This is a massive improvement over previous versions. The reduction in peak voltage by removing one of the solar cells was more than offset by eliminating the leakage current through the voltage-limiting zener diode. The supercapacitors are rated for a maximum voltage of 3.0V, and a five-cell array reaches around 2.9V in circuit in full sun. However I am very happy with a >24 hour runtime with only a few minutes of charging in the sun, and the lifespan of the supercaps will be improved by staying further from their maximum rated voltage.
I have designed a new, smaller, version of the PCB with space for two supercapacitors and no zener diode. This will be a bit cheaper to manufacture, with fewer components to fail. I haven't analysed the logged voltage data yet, but I expect to get 30-40 hours of dark runtime with this configuration.
Geocaching is an outdoor activity in which one player hides a small cache containing a log book and, space permitting, some toys or tokens. That player then posts the coordinates online, usually on geocaching.com. Other players look up the coordinates and find the caches, signing a logbook and logging the visit on the website. Geocaching is popular world-wide, and there is a fierce competition to be the “First To Find” (FTF) a new cache. I have been tracking cache activity within 12km of the Brisbane CBD and there are around 200 cache finds per week in the area.
Some caches have puzzle components where riddles must be solved or clues must be found to complete a part of the coordinate. For example one of my caches is not found at the posted coordinates, the location is offset by a few hundred meters in a random direction. In the fiction of the puzzle our GPS was faulty. It was precise, but not accurate; it could reliably reproduce coordinates for a given physical location, but those coordinates didn't align with anyone else's. To make the puzzle soluable we provided calibration coordinates for a reference point you could find with your eyes. Once the player had found the calibration object (a specfic flagpole outside a stadium), they could calculate the offset for our 'faulty' GPS and work out the actual location of the cache.
I want to use a blinker as a part of a geocache. The coordinates posted on the website would be the location of the blinker, hanging high in a tree. A cryptic hint would instruct the finder to look upwards at dusk to see the blinking LED. The morse coded message could be a relative offset, like 'TWO HUNDRED METERS EAST' or absolute coordinates. This information would lead the player to the actual location of the cache. The main challenges of this idea are ensuring the blinker gets enough sunlight to charge, and securing it high in the tree without harming the tree.
One way to solve the sunlight problem would be to equip the blinker with two sets of solar cells. Conecting two parallel arrays of four series cells, sandwiching the electronics between them, or in a triagular arrangement with the two arrays steepled to catch the most sun. There will be limited ability to control the rotation of the blinker around the Z (up/down) axis once the blinker is hanging in the tree. However the solar cells generate decent voltage with indirect sunlight, so exact alignment isn't super important.
Getting the blinker up the tree without climbing up could be done with a length of stainless wire with a barb on one end a hoop on the other:
O——>
The blinker would be mounted with an eyelet cast into the resin. Some fishing line could be tied to the barb, cast over a suitable branch, and then threaded through the stainless hoop at ground level. The blinker could then be raised into the tree by pulling on the line until the barb passes through the hoop, locking the blinker in place. The fishing line could then be pulled until it snaps. It would be a good idea to weaken the line where it meets the barb, so we don't end up with a trailing end flapping in the wind.
It is important that the blinker hangs down, and doesn't end up perched atop the branch, or caught in a crook. This could be done by using lightweight stainless wire, adding weight to the blinker, and/or making the stainless wire longer. The design of the barb needs to be simple, robust any made completely from stainless steel. Potentially it could be made from some single-core stainless wire formed into an arrowhead shape, with the 'haft' ends crimped to the flexible wire.