12kW Solar Collector


Though not much info is readly available about it on the web, [Joe Carruth] is trying to build publicity (and venture capital) for his home-built solar electric generator. At its essence, it is a Stirling dish system with an adjustable composite mirror surface. This means that instead of having to rotate the entire contraption in order to follow the Sun, [Joe] only has to make  the mirror segments pivot. A Stirling steam engine at the tip converts the energy into the movement used to generate electricity. Solar power plants (or ‘farms’) that are emerging are beginning to consider the advantages of using more efficient Stirling dishes rather than less efficient solar panels. If anyone has an idea as to how [Joe] can automate sun tracking for the mirrors, please post it in the comments. A couple more videos on the topic (in general) are available below:



57 thoughts on “12kW Solar Collector

  1. Tracking the sun is probably easier then anything, you could even do it mechanically. The path is predictible. The angle will change slightly. He can buy off the shelf astronomy tracker for a telescope, already has the software and the hardware done. He can use it to control servos or whatever.

  2. fir the servos he needs to buy 2 screwdriver motors for each row an d column, and put a movable rope between the two, and for the actual adressing of the mirror just use clops to grap the rope, first horizontally move rope, then vertically and move rope, and youre done, i dond know if it is really efficent but its cheap

  3. from previous reading on the topic (and without reading tfa), an important consideration for the mirror would be pointing it down at night so as to help keep it clean and protected from hazards such as dew that could cause the mirror to get dirty and reduce the efficiency. suppose one could also have a housing for the contraption which would open and close at the appropriate times.

  4. Not a very good description. The setup is not stirling, its a slimple boiler reciever. The parabolic mirror support needs to be pointed at the sun (has no heliostat). The parabolic mirror is build up out of square segments that have been pointed individualy. The area is about 3 x 3 meter, or 4 x 4 if you want to be leniant. That creates a max of 16* 800 Watts thermal heat (but the mirror is not pointed accurately and the mirrors aren’t silver. So the max thermal energy is 12,8 kW. A steam engine and generator allow (in optimal circumstances) and efficiency of 12-25% conversion, which means that the power output is 3,5 kW max.

    Stirling engines shown are 50 kW not 50 MW. Stirling engines are prevented from becoming freely commercial (I have tried and tried).

    CSP is wildly advantageous, but please be more accurate..

  5. I was obsessed with heliostats a few years back and wrote down a “Heliostat Vector Calculation” in an old notebook. I know for a fact that this is NOT my original work and if anyone intends to use it you should probably bother to find the originating source. Here’s a link to some scans from my old notebook:

    Heliostat Vector Calcs

  6. I like the idea that each element is ‘smart’ with it’s own sensors, actuators, and micro controllers. That makes the array scalable, must add more more elements. I think the easiest thing would be a set of 4 sensors that sets up a light gradient that could then be used for tracking and provide a measure of when there is enough light to power the system. The absolute cheapest way to control things might be some op-amps, but I would be concerned about the effect of temperature on the whole thing. A micro controller would probably be better, but I think that will increase teh cost of the whole thing.

    My other thought was, as long as the geometry of the system is fixed, then you could have a camera or cameras that track the sun and based on the fixed position of the Stirling engine relative to each mirror, updated the array. It is scalable to a lesser degree so long as the algorithm can handle an array of variable dimensions and there is address space for each new mirror.

    All this assumes that the differences in the angle needed for optimal power differs enough between mirrors to warrant individual mirror tracking. Maybe segments of mirrors could be grouped together. This may not result in max power, but the difference might be small enough that it is not a big deal. Still, if individual mirror tracking can be done economically, why not give it a try?

  7. Back when I was in high school (yeah…20 years ago!), my friend and I build a solar tracker for a science fair project. It was dead simple: two photocells, two op-amps (741 comparators, if I recall), and a handful of resisters. The photcells were placed behind small black tubes in order to provide a rudimentary aiming system; when the sun shone straight down the tubes (one cell was on each side of the dish), and the current was the same on each side, the servo held still. If one sensor showed brighter, it turned the dish until they were even again. We used limiter switches to stop the rotation at the ‘morning’ and ‘night’ locations, and when the ambient light fell to a low enough level, we sent the dish back to the start point until the sun came up again.

    Problems with this hack mainly came from overcast skies causing the CDS cells to not read anything, resulting in the dish sitting still all day; also, a long period of overcast followed by sunshine would make the dish stop following the sun, as it wouldn’t catch up again. Those problems could be fixed by using microprocessors such as an Aurduino. In this system, consisting of many smaller segments needing control, I would link them together in a parabola, and then shift them all at once.

  8. Pointing mirrors could be done using some 2 screw adjustors. Doing it cheaply could be done with N car mirror adjustment assemblies, where N = number of mirrors. If each one needs to point a bit different, put a PIC/ARM/whatever on each one and tell it where to go over a serial network of your choice.

    I remember All Electronics had a collection of the mirror adjusters on surplus years ago.

    Master controller, there are any number of open source bits of code that will tell you where in your sky to find the sun. Crunch the data on a PC, feed to the controller network once a minute.

  9. tracking the sun is very easy.. essentially you input lat/long/elev and time…grind…grind…and you get az/el out (two servos). the grind is done with approximating equations that you can dig up from the naval obs. website which are actually quite accurate.

  10. Does the steam come from a water source or is it taken from regular air humidity? If it is plain water, isn’t this a waste of water?

    Regarding Stirling engines, they are very efficient if constructed correctly. A friend of my constructed one using coke cans and hot glue that worked using the heat from your hand.
    There are plenty of examples on the internet. For building a big one, would be just a matter of scale and materials?

  11. Because the mirrors locations are constant relative to the boiler, you shouldn’t need sensor(s) for each. You just need to know where the sun is relative to the grid, do a little math for each mirror, and adjust. But you would need movement on two axes per mirror, so you’re still talking about an enormous cost in servos alone.

  12. Mount the mirrors on individual pivots with linked parallel control bars for X and Y. Each mirror would need a set-screw for standard deviation relative to the collector.
    Would have thought setting up a cross of 5 photocells in angle-limiting tubes would be able to say when the mirrors need to move. Set a simple voltage trigger to provide direction and step-pulse to a stepper driver and have the stepper geared correctly for a nice smooth movement in whatever axis.

    Additionally, wasn’t there an article not so long ago about water-cooling photovoltaic cells to increase their efficiency? Any reason the mirrors couldn’t focus on a PV cell and the cooling loop for run the Stirling Engine, rather than directly?

  13. *Why* would you want to aim the mirrors individually on a system like this??

    …It adds mechanical complexity, which is bad, and reduces the efficiency of the array, as when the sun is not directly in front of it, the surface area ‘visible’ to the sun is smaller.

    It only makes sense when the cost of moving the entire array around becomes prohibitively expensive, and when you want to be able to dial in the amount of power generated according to demand. Eg. http://en.wikipedia.org/wiki/PS20_solar_power_tower

    And as well, why over-engineer the tracking system with cameras and such? It’s not like the sun’s path is hard to predict; it’s easily calculated from a given latitude/longitude, orientation and time.

  14. Not too long ago my team was tasked with a solar tracking project for a large european solar company. After several trips to the giant facilities in the California deserts running efficiency tests and com tests, we were successfully able to make the tracking aspect play, to within .9 degrees.

    We did this with PLCs and radios. Over 3000 of them.

    The PLCs were doing all the calcs for JDN time and syncing with an accurate time source for very accurate positioning.

  15. @sleeper
    I can think of a reason. Mounting. I could easily mount a flat array of mirrors to my pitched roof, but not a monolithic pivoting dish. It’s probably also a lot easier to get a lot of small mirrors than it is to make an effective single dish.

    The boiler is quite chunky, having it mounted at the center of a dish could block a lot of light. That said though, he does have it set in the middle of the array anyway. Why not just adjust the angle of the mirrors and have the boiler more easily accessible on the ground? He’s using it as a parabolic when he doesn’t need to. Especially if he’s manually adjusting all the mirrors anyway.
    Also as Climatebabes points out, it’s a full-loss steam engine not a Sterling.
    And he says it’s hooked up to a 12Kw generator, not that it’s actually producing 12Kw.

  16. Start with the mirrors focused from the ‘boresight’ (perpendicular to the mirror array) to the target–so that if the Sun were on the boresight, each mirror would illuminate the boiler.

    Now, if you rotate every mirror by the same amount (eg. 23 degrees left, 5 degrees up) then every mirror will focus the same direction (46 degrees left, 10 degrees up) onto the target.

    So all you need is some linkage or other mechanism to rotate all mirrors by the same amount. This can be done with axles, gears, and chains. Only two motors are required.

    (And the mechanical adjustment that allows you to adjust the mirrors for boresight pointing, which you only have to do once. A pivot and two adjustment bolts will suffice.)

    Note that I have only verified that this works in one dimension, but my guess is that it works in two.

    Power output is proportional to the cosine of the distance from the boresight, so you do lose efficiency, but you gain in not having big moving structures. If you lose 20% efficiency, but can build 50% more for the same money, you win.

    For more efficiency, dish the mirror segments. This can be done simply by the stress technique (make them out of thin glass supported around the edge and pulled back in the middle). Precision is not required.

    The control system is simple: just move the ganged mirrors simultaneously left, right, up, down according to where the hot spot falls on the target. If you totally lose pointing, you can hunt for it with larger movements.

    Control it all with an arduino, just so that people on this site will complain.

  17. The easiest and cheapest thing would be to make the entire dish rotate and pivot instead of individual discs, this would also make it less susceptible to harsh weather conditions. That being said, i would probably go all out with a light sensor array at the tip of the dish, and make an arduino calculate and move the dish to point it’s nose into the sun.
    I bet it’s already an expensive project, so why not go all out and make it work as it is supposed to? :P

  18. Great project… I hadn’t watched their youtube channel in about 9 months, but they always cover interesting things.

    The water that system is using is from a storage tank they have there on the unit; to build a recycling condenser would be simple and thus water would not be “wasted”.

    Aiming individual mirrors on an array that small is not reasonable. (the sun is a single source that moves; as long as all of the mirrors are calibrated to reflect from the sun to the steam compartment at the same time [while being on the same platform], there is no need to individually aim the pieces)

    Sun tracking systems can be built very simply or with great complexity; youtube has some great examples, other websites are dedicated to DIY plans.

    Sterling Engines are also simple to build and great examples of these (as well as explanations) can also be found on youtube.

    Great post, thanks; you all could take a look at their channel where they cover some other interesting solar projects as well.

  19. Use a line focus instead of a point focus. If you align a trough on the east-west, and make the length of the trough long relative to the distance between the trough and the focal line, most of the light will land on the focus throught most of the day. Then you only need to adjust one axis. Make the trough out of long parallel slats and adjust the focus as the sun’s path shifts north-south through the year. These would be small adjustments that could be made infrequently, so you could do it manually on a small scale.

    Alternatively, align the trough north-south. In this case, it might be best to use a plane layout instead of an approximate parabola. This arrangement would have to track the sun more extensively, but it would still only be one-axis. You would, of course, gather more light per unit area when the sun is near the horizon, at the expense of additional complexity over an east-west arrangement.

    Remember, of course, that black tubes make very good solar boilers, and are safer at high pressure.

  20. I’m pretty sure that making the entire array and boiler track the sun is mechanically more simple than trying to make each mirror track individually. The mirrors need to turn at different rates so can’t use a simple mechanical linkage, whereas the entire collector/boiler could be set up on a single-axis tracker (adjusted for sun elevation over the course of the year). As the boiler seems to output steam via a flexible hose it shouldn’t present any great engineering challenges.

    I’d be interested to see how much water this uses in the course of a day, and whether the boiler could be replaced by a commercial Stirling. They do exist, but would need modification as they seem to be designed for biomass-burning CHP- I found one intended to output 12kW of heat and 1kW of electricity, with fuel needs stated in terms of cords of wood, but can’t find it now.

  21. rustybadger’s approach is one I’ve used successfully.

    i used 4 optical sensors mounted in each quadrant of a cross made of two bits of opaque material (gash wood, in my case). the “deeper” the cross, the more accurate the tracking.

    two comparators and two dc motors plus drivers means simple and cheap electrical bits. No need for servos.

  22. as stated above the 12KW are thermal power only. in the vid when the saw is powered you can hear that the generator turning frequency drops, this means that approx 300W starts slowing down the generator. also the video is very unscientific, could at least provide some basic information

  23. This is just me thinking out loud, so it’s not fully worked out or anything… But I think this is what I’d try initially.

    I think I’d go with a motor for each row and column. I’d connect up to the individual mirrors with ~1/4″ round stock and some universal joints from a cheap Harbor Freight socket set. Seeing as this is a prototype I’d just epoxy them together. You might be able to use a braided line or cable, like the Dremel-tool cables, which would simplify things a bit.

    The mirrors would be aimed with simple cams that rub up against a bolt set into the frame.

    For now I’d cut them out of hardwood and run a strip of brass or something along the edge. Adjusting the bolts and adjusting the timing on the cams should allow me to adjust the position fairly accurately.

    To account for seasonal adjustment I think I’d use a set of motors and screws attached directly to the frame.

    The whole thing would be driven by a microcontroller.

    This way you’d minimize the expensive parts, the motors. You’d only need one motor per row, one motor per column, say four motors for seasonal adjustment. Stepper motors are cheap and would work well enough, though servos with gear reduction would probable be better. Weatherproofing it would be kind of a pain in the ass, but it should be manageable. If you’re in a wet climate then grease the hell out of everything, dry climate then dry-lube the hell out of it.

    Someone mentioned that it’s important to be able to invert the mirrors to help keep them clean. I think you’d be able to use the seasonal-adjustment motors to flip the whole mirror-frame over. Maybe not all the way, but it should be good enough to get the surface of the mirrors out of the rain, etc…

  24. I think someone already said it- the advantage of a system where the mirrors remain stationary rather than a single large mirror is the spaces you can put such a contraption. One could place many small mirrors on a roof – one could not place one very large parabolic mirror on a roof.

    The cost of having each mirror being controlled by its own system increases cost, but it also adds a great feature of expandability. If you designed the mirror layout around a center point and only had to set up each mirror with dip switches as to it’s location in reference to the center point then as one wanted more kW potential then you just buy a few more mirrors to add to your system place them and set them up with it’s location. Almost a plug and play system.
    Doing so would mean you would have to have solar positioning ability and time ability in each mirror but maybe the center point of the grid could house a box that would provide that via lan interface to each mirror set up via a small wireless network system. Or you could just address each mirror and run it from a single remote computer with a small program that would take into account each mirror location from the center reference point and focal distance.
    Individually controlled mirrors would make this the most versital sytem.

  25. While I applaud every effort toward greener energy, I have to wonder; what about inverting the setup and putting a large fresnel lens on top (like from a large rear-screen projection TV), and focus the light onto the target. The entire structure would be rotated, but that has to be lighter than mirrors. IIRC, the folks as american science surplus (who had a large lens in their catalog) said that on a sunny day they could melt pennies with it. Maybe the fresnel doesn’t transmit as much light as what is reflected from a mirror? Anyway, it would just be one overall rotation, instead of having to rotate multiple mirrors, because the lens already takes care of focusing the overall area onto one point.

  26. If you Google ‘sun tracking software com port’ you will find many sites that discuss or offer sun tracking software. Some offer the code that can be adapted (ie USNO’s NOVAS code could possibly be adapted) and most use the sun’s altitude and azimuth tables produced by the USNO or HMNA on CD. In fact the calculation required to calculate the accurate altitude and azimuth for any position at any time can be programmed quite easily with an input for time and and an output command to a server motor set for any required time interval.

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.