Maplin to run from the solar panel in the garden to my first-floor study (where the rest of the equipment is too) for ~£5; too low a rating would waste power as heat in the cable and might be dangerous, but at ~17V and 20W peak the maximum current that the solar panel will generate is a little over 1A. I also bought a cheap £10 multimeter from Maplin to help me monitor the state the of battery and power consumption. This brings the total to about £250. I already have simple tools such as wire-strippers, screwdrivers, etc.
The BrightLightSolar 20W kit comes with the option of a larger (100Ah) battery, but given that my 16W light should use a little over 1A and thus the 40Ah battery should be able to power it for nearly 20 hours (to a safe 50% discharge level), ie about a day or two of heavy use, there does not currently seem to be any benefit in using the more expensive and higher-capacity battery.
I asked BrightLightSolar how much time my lamp might be powered each day given the 20Wp (20 Watts peak) panel in their kit. Mark Hilton very kindly replied:
If used every day for the same amount of time [and with] the panel mounted optimally (i.e. due south at 35-40 degrees without shading), you could expect a 20 Watt light to be powered up to about 2.5--3 hours in mid-summer, and up to 1 hour in mid-winter. Two 20 Watt panels would double this performance, and so on. The kit is suitable for use with either the 40Ah or 100Ah battery - if you are going to use the system periodically (i.e. enabling the system to recharge itself on the unused days) a larger battery would enable you to capture more of this energy, giving you far greater performance/operation running times on the days that you use the system.
This suggested to me that battery capacity was not really going to be the critical issue for me, and that I might have to buy an extra solar panel in the winter to bolster the performance of the system if it proves inadequate, ie if I can't run the lamp as long as I need to each day.
Given the orientation of my house (with my study overlooking a very small west-facing garden), I'm unlikely to be able to get anything like 'optimal' power generation, but the initial setup has the solar panel hanging on a south-facing wall free of shadow from around noon until dusk. Further experimentation will be needed, I'm sure, to maximise generation when light is feeble in winter.
The charge controller (the Morningstar SHS-6 which is optimised for cost and simplicity/robustness), probably just about optional in very small systems, helps protect against overcharging or overdischarging the battery, which can seriously damage it, as well as protecting against short-circuits, reverse polarity and other damaging/dangerous nasties.
The solar panel needs not just to be pointed in the right compass direction, but it needs to be installed at the correct slant (ie not just horizontal or vertical) to maximise electricity production. It is possible to use mechanical systems to track the sun, but they add weight and cost and complexity and consume power themselves. In most cases it is best to face solar panels due south if possible, then at a good year-round angle to the horizontal, possibly adjusted a couple of times per year. Mark Hilton of BrightLightSolar very kindly advised again:
The panel should be orientated 35-40 degrees from the horizontal - this is a good angle for year-round optimisation. In mid winter, the panel will perform better at angles of up to 55 degrees from the horizontal, while in mid summer you will get slightly more power at the 30-35 degree level. Due to our northerly latitude there is a significant difference in the amount of light (and therefore performance of a solar system) between mid winter & mid summer - the UK gets 80% of its solar resource from March to September.
From my observation of the Stellarium application set up for my home location, peak (noon) sun angle above the horizon is more than 60° in June, and as little as 15° in December, so if I wish to to maximise December output for most balance year-round (to be less hugely over-specified in summer relative to winter) then I should possibly mount some or all panels at as much as 75° to the horizontal, and even a near-vertical mount might not be disastrous. I do have obstructions (trees) to December direct sun at that low angle, but conveniently they will have shed their leaves by then, letting more light through.
It turns out that there are sensible minimum and maximum battery capacities for a given system (assuming lead-acid throughout here).
The minimum capacity is determined by the need not to discharge the batteries below about 50% to avoid shortening their life, so if there is the need to last through a very cloudy few days, for example, a minimum Ah to average A current drawn of 200:1 is suggested (48:1 is 24 hours' use discharging a full battery to about 50% capacity, so 200:1 is about 4 days' use). That is, minimum capacity is determined by the load.
The maximum capacity is determined by the charging source and battery charge leakage. All batteries self-discharge to some extent: for lead-acid this is said to be between about 1% per day and 40% per year depending on type, condition, etc. If your charging source is not providing more energy/current on average than this self-discharge, then the battery will eventually discharge completely and be badly damaged. So it would seem sensible simply to ensure that the batteries are protected by arranging that even in mid-winter the batteries are receiving more than 1% of their rated capacity each day (plus any used by the load). In the UK, an optimally-tilted fixed solar panel should generate about 1Wh per day for each 1Wp of rated capacity in mid-winter. Given, say, a 100Ah 12V battery, we should be supplying it with over 1% (1Ah) of energy each day to overcome self-discharge, which therefore requires 12Wp of solar panel (12Wp = 12Wh = 12V@1Ah approx). (My initial system with 20W of panel and 40Ah of 12V battery should be fine.) So each 100Ah of battery capacity at 12V requires at least 12Wp of well-sited solar panel to see it though a UK winter. (At 24V, 24Wp is needed, etc.)
Actually wiring up the solar power system was very simple and probably took an hour all told; the controller has specific clearly-marked connectors for the solar panel, battery and load, and the instruction leaflet explains what the various LEDs indicate including the battery state and whether it is currently charging. (On the day that I set the system up, the battery was still clearly charging almost until dusk, which I otherwise would not have known nor suspected, and the next morning it was charging again even though the panel was in deep shade and the Sun is behind the house from it.)
As I write this paragraph, I am using the new light above my desk on the 8W (one tube) setting and it seems nice and bright. I'll have to fiddle around to find the best position and to avoid excess glare. (Note that the battery is shipped partly charged to avoid damage, so I'm not running entirely on solar power on this first evening...)
Side note: switching to lead-free solder has required buying a more powerful (50W) temperature-controlled iron else even getting the solder to melt other than on the smallest of joins can be a pain. I'm also trying more expensive AgSnCu (ie silver-containing) solder rather than just SnCu (tin/copper).
I may have to shift my working time earlier to make better use of natural light and reduce my dependency on the solar light, though of course I still have my mains light if I need it. And the light only represents a small part of my energy consumption while at my desk: my laptop probably uses 30W to 50W+, never mind Web servers and connectivity, etc, but I can't solve all of those issues at once.
In mid-July I added a 62W amorphous UniSolar ES62 panel in parallel with the 20W panel using Schottky blocking diodes. This should give me far more hours of lighting in the winter, and some excess for further extensions/experiments.
See how I extended the project to charge my mobile phone and power my home Internet server.
Basically, it worked!
I got through the winter without mains electricity for lighting, and I managed to take my Internet server off-grid with the excess energy for a significant chunk of time as a bonus.
This convinced me to take a step up and make my office "zero-carbon" with a small grid-tied solar PV system.