Sustainability Hacks: Solar Panels Built From Old Windows And Factory Rejects


The crew over at The Tech Junkies recently took another look at solar power and thought that the pricing had come down enough for them to consider powering their entire shop via the sun. Cheaper or not, they still didn’t want to pay retail for solar panels, so they decided to build their own instead.

They scoured eBay for a bit and scored a nice batch of “production error” solar cells for about $0.25/watt, which is a great deal. After unpacking and sorting the cells, they began fitting them into a set of old window panels they had sitting around their shop. The cells were wired together using tabbing wire, and after a quick test to ensure everything was working correctly, the panel was permanently set using epoxy.

In its current state, they estimate that their panel can generate 35 watts of power, though they have a few design changes in mind to raise that number a bit. The total cost was roughly $375 for enough materials to build 9 panels, which is pretty reasonable.

Be sure to check out their blog for a quick overview of what it takes to build a solar panel if you’re thinking of putting together one of your own.

37 thoughts on “Sustainability Hacks: Solar Panels Built From Old Windows And Factory Rejects

  1. Very nice. I used some huge windows I salvaged by a hospital dumpster to make a solar water preheater. You can see it on my blog. Sorry there isn’t more info. It’s not electricity but it sure saves the gas bill–the meter reader wastes his time ;)

  2. I think your maths is out, HaD. You say they ended up with 9 panels at 35W each – so whilst the cells where $0.25/W, the panels actually ended up costing $375/(35W/panel * 9 panels)=$1.19 per watt. Sadly, that’s at the lower end of what you can buy new panels for – so whilst this is a good hack, the economics don’t really make sense.

  3. $1/watt because they’re not including their labor costs. Also:

    a)nine panels is absurdly low density for 300W. Usually that’s 1-2 panels. Panels cost money to mount and take up roof space.

    b)There’s no way in hell panels with wood frames made out of windows will last very long. Here in New England, they’d be lucky for them to last one year out on a roof. Commercial panels are warrantied for a quarter of a decade.

    So, basically, they flushed $300 down the toilet for a system they’ll be lucky to have offset one or two lightbulbs, and it’ll need repairs or replacement in a few years, which means it would have cost them the same as a commercial panel. Bravo!

  4. How you wire up the cells is paramount. If you wire them in series for higher voltage, you increase the series resistance of the panel which makes it act more like an ideal voltage source. I.e. if you draw the V-I curve of the source, it looks like a straight line going down from maximum current on one axis, to maximum voltage on the other axis.

    The maximum power point of the V-I curve is at the current and voltage that maximizes the area of a rectangle drawn from (0,0) to touch the curve at the other corner. That tells you what your load should be to achieve the most power output.

    If you connect the cells in parallel, you get less voltage, but the V-I curve becomes completely different. It starts out vertical and only dips down very close to the panel’s nominal voltage. The maximum power point under this curve can yield much much more power for the same panels at the expense of output voltage.

    Which is why it would be wise to device a MPP-tracker that knows the characteristic curve of the panel and boosts up the voltage to charge batteries at a rate that is most suitable for the cells.

      1. The typical cell voltage is 0.5 volts, so you have to put some of them in series or you’ll lose most of your energy to the forward voltage drop of the semiconductors you use.

        You can’t efficiently boost up voltages from less than a volt. The Joule thief works, but once the source voltage reaches 1 volt, the efficiency is no more than roughly 30% unless you’re using special low drop semiconductors.

    1. Sorry. I meant to say the V-I curve starts out horizontal and drops down only very close to the maximum voltage when the cells are in parallel.

      Also, the maximum voltage doesn’t change much. The current will go from 100% to nought while the maximum voltage will dip roughly 30% from full sunlight to the point where the panel stops producing meaningful amounts of power. Knowing that, it should be rather simple to design a boost converter that draws just the right amount of power to get the most of it.

      1. Can you point us to a guide which shows us how to wire them up for maximum efficiency?

        Preferably one with that shows the calculations needed to make the most effective matrix given different solar cell specs.


      2. @Cageybee
        I can’t really point to any source, it’s all from the top of my head. Google would give you better answers to that.

        And it really depends on what you want to do with the panels. How much voltage you want out? 220 volts, 120 volts, 12 volts?

        At some point it becomes less efficient to run a converter than it is to simply wire more cells in series. However, the rule of thumb is that the more cells you can get in parallel, the better.

        The fundamental model of a solar cell is a constant current source in parallel to a diode. This diode represents the recombination current path of the charge separation due to photons hitting the cell (represented by the CCS), and the forward voltage drop over the diode due to the current represents the source voltage of the cell. Then you add a resistor in parallel to that to represent other internal leakage, and another resistor in series to the path to your load to represent the series resistance of the cell.

        And that’s about it. Once you figure out what the characteristic curve is for your single cell (or a piece of paneling), you can figure out roughly what it’s doing internally and then start to optimize, mostly by applying the Ohm’s law.

      3. Or what I would do is to measure one small panel, draw an equivalent circuit for it, and then simulate it on a computer.

        Then I would connect up a bunch of them virtually, in different combinations, and see which one gives the most favorable characteristic curve for the intended purpose. You don’t have to do many variations because you can interpolate from the results what the most optimum combination should be.

    2. You are emphatically wrong. Power output is the same regardless of series/parallel if all cells are lit with the same light. Solar cells are not magical devices that act differently if other devices are in parallel instead of series.

      The advantage of parallel is the panel will continue generating a proportional amount of electricity if part of the panel is shrouded in shadow. Series will pretty much stop. However, for practical reasons you must have some in series because 0.5V is too low to efficiently extract power.

  5. lets not get too wrapped up in the economics of the project. remember a lot of these hacks are more about stimulating creativity and innovation, and like myself, have lead to bigger things, with actual commercial value. cheers to building that setup! i hope it inspires others to experiment as well

  6. It is not very sustainable to permanently set the panels in epoxy. One guiding principle of sustainable design is to make things easy to repair and recycle at their end of life.

    What if one panel fails? Or they ever want to break the panels out for use somewhere else?

    1. @[alan] – (assuming you meant cells and not panels) how else would you “fix” the cells so they do not move? Given that they are fragile enough to break if you breathe on them too hard, what else to do?

  7. :O Are they dumb??? They broke some of the cells in half to fill out the full surface area of the windows they were using??? A series of 33 cells with 11 of them only being half. Do they even understand basic electronics??????
    Having half cells in series will halve the current output of the whole panel.
    I am simply aghast at what I’m seeing here.

    1. Whoops, that is definitely an issue. When building this first panel we were most concerned about getting the voltage high enough to test with the inverter, but the next set of panels we build will be using the full cells. I’ve updated the post with this info…thanks!

      1. i agree with joeinbend

        see? he is learning just like the rest of us have!

        dont forget that if u throw a ton of scientific info at someone they might not remember it all at once. kudos to R&D, if u skip R&D then we would “all just know everything” and we would not have this great website and hobby.

        PS: those cells were rejects and might have been destined for the landfill, instead theyr making power so people can have light(soldering iron) their shop to build a SUPER EFFICIENT power inverter or portable battery box WHILE THE POWER IS OUT! ;)

  8. Is that grid tie inverter UL listed?
    When I do a solar job for a customer, we have to get approval from the utility and they come and inspect it and watch it do a proper anti-islanding cycle before they approve the system to be grid tied. This thing just plugs in… Hardly seems safe to plug into a grid supplied power circuit.
    Also, even if you spin your meter backwards you don’t roll the kwh back, they still clock forward to charge you money.

  9. Wattage means little in a “real world” design application.

    Why try to run 120-volt/100watt lighting off solar panels? Dumbest. Thing. Ever. Thats the mistake most people make when thinking about solar.

    Most, yes MOST if not ALL of these panels on eBay are junk. I’ve purchased many from the “main four” sellers that call them “A-run” or “B-run” panels, or seconds, or rejects; I think they make these terms up, as I cant find a solar manufacturer that says they sell ‘reject’ panels.

    So, that means they are all from China. Not that that makes them bad, but there is some trolling financial going on here.

    The best shipment I’ve received from an eBayer was a pack of 108 panels, 10% were broken all the way through, two were snapped in half (can still hack them to make them work) and all but two were cracked.

    They are fragile as heck, near impossible to solder (use the expensive conductive tape or that 3M copper stuff posted by a user in the “glass PCB” hack, burglar alarm tape from Radio Shack works too) but they do seem to put out what they claim on SSV.

    After all is completed, you’re FAR better off buying a commercial panel when they go on sale, like I have a few times – Newegg and Northern Tool often have huge discounts on panels and kits from 15 to 500 watts.

    Source: I’m an electrician and solar-certified for surveys and installs (And I come here! :) )

  10. I’d forgot to mention that I believe newer homes will be designed with solar and 12-volt or 24-volt lighting, hence my “dumbest” comment about 120vac lighting from solar.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.