Homebrewing a Beast. (2)
What brought me to this point was working with other triggers, and my favourite is the StrikeFinder V2; I’ve a couple and I really like them but one of the greatest frustrations I’ve faced for years is that a lot of the storms in this area especially during the summer are isolated, and occur in daylight.
Might not seem much, but a lot of the triggers that are about are looking for a shift in light optically in the near to full infra-red frequency; and doing it on a passive level so it may need a serious flash to register (or in the case of some sensors, fighting the sensitivity all the way so the detector isn’t overloading). In this, daylight is the problem as when it’s sunny there’s a lot of infra-red being kicked about by the sun and also it reflects a lot off the clouds so it’s either modify the sensor to the hilt to just focus specifically on a small area and hope it sees something or … Build something complimentary to cover the gaps; though this project didn’t start as that, it started as something different.
Quite a few years ago, I was in contact with another photographer who built some triggers, these were actively compared circuits and could detect at quite a range in daylight; but by the design to do anything with the camera they needed disconnecting, as they were only on a single opto-isolator and as the power decreased on the positive rail, so did the sensitivity. On a fresh set of batteries, without direct sunlight hitting the front, they could detect a lightning strike at about 25km, and at night a lot greater distance. However, after a couple of hours the batteries would be low enough to reduce this by half.
Fast forward a few years later, and I’m on passive detection stage triggers, and getting pretty good results in the right conditions, however I was increasingly frustrated by the amount of daylight and not being able to tell if that cloud over there is active, or it’s another one… And when going from one side of the valley to the other takes ten minutes or so between decent vantage points (and having a false knee which sometimes seizes up) in 30ºC + heat, I decided to build a device which I could point in to the clouds and it’d make noise if it detected things.
Thinking back to the device the other guy built, it seemed that it was a lot more sensitive to “invisible”, so I went back to that machine, took it to pieces and then tried to work out how it worked… And then work out the kinks in it that had bugged me from there. Out came the paper, on went the coffee machine and off to work I went. First thing I decided on was a regulated power supply, as whilst the original circuit could run at anything between 4.5v and 36v, having a drifting variable like that makes the calculations a lot harder.
In went an LM7805, and after countless builds on the workbench, quite a few wasted perf-boards and a lot of hours the only single thing that remained of the original design was a 1N4818 signal diodes. Out went the 5% resistors, polyester caps, transistors and in went 1% tolerance with higher resistance, ceramic capacitors, an op-amp instead of just a comparator a few status LEDs and a 368p. Now I’d got a design on the workbench which not only worked, but had consistency in it’s results and had a level of control that wasn’t possible beforehand; and after matching it to a detection circuit that was more easily affected by the pulses a lightning strike makes both before and during the discharge, results started to pick up.
I wasn’t happy with it, though, as being something bolted on top of an Arduino Uno at this stage it was a big, unsightly and ever so fragile device to work with; to the point I’d got a gas powered soldering iron in the kit bag just to keep it together when out of doors. Something had to change there, so I got a few PCB etched, fixed an Arduino Nano on to it, stuck it in a box and started tweaking it up based on what I’d learned with it. Without realising, I’d gone through prototype development and come out the other side with something not as I’d gone in with. I’d started off with an optical detection system, and now I’d got closer to a radiation detection system, added a noise reduction system, removed it, changed the hardware level sensitivity but with the added processor on there, I could do a lot more in firmware than what I could previously.
Two years ago, I’m sat on a rock with the camera, and saw a perfectly formed C/G drop down not 6km away direct centre of the frame; and started cursing to why the sensor connected to the camera didn’t pick it up as it was in the IR wash even though the camera would have made mincemeat out of it even though my machine went “boing”.
That… That was not the end of the journey, but the start of a new rabbit hole.