SOLAR TRACKING SYSTEMS

Won't it get confused by clouds and fog?

A micro, clock, and lookup table would be my first choice. Run it open loop. As long as you know the date, time, and location, the rest should be fairly easy.

I'm not sure what's going on here, but I received the exact same question via email, except that it was from Raza Tauseef instead of Erica Cross. I don't mind students asking questions, but this thing seems to be simply a time burner for those answering the question. Hopefully, I'm wrong about that. Anyway, here's a copy of my reply.

You can buy them on eBay.

There are about 10 approaches possible varying mostly by the accuracy

you're withing about 5% of maximum power output. Read this:

so. You can do as well with a clockwork mechanism (24 hrs = 360 degrees) and a fixed elevation angle (single axis tracker).

The problem with trying to find the sun is that it will also track automobile headlights, airplanes, the moon, police helicopters, clouds, and reflections. I've build a few like this for demonstrations. Watching it track the moon was interesting, but not very useful. In order to make this work correctly, you'll need some kind of sanity check against the ephemeris produced by a clock algorithm to keep it from going astray. you might as well build a clock based mechanism to start with and don't bother with trying to follow the light.

Incidentally, you also have to rate limit the speed of the traverse. I failed to do this in one of my models. A friend approaching my panels pointed a flashlight at the tracker cells, which caused the panels to violently swing in his direction, nearly knocking me off the hilltop.

You also might want to look at non-electronic solutions such as Freon gas powered devices.

Good luck and please note that I answer questions in the forums and newsgroups for free because everyone learns from the questions and the answers. Personal questions and answers, for which only you benefit, is consulting, for which I normally charge money.

Both approaches can work. The first seems to be overkill for a homebrew project, for a couple of reasons:

- The algorithm seems to need floating-point math. Simpler low-cost embedded controllers are fixed-point-only (although newer boards such as the Teensy make this less of an issue).

- The digital approach calculates the angles, and must aim the panels accordingly. In order for this approach to be reliable you would need some sort of feedback from panel to controller, so you can figure out which way the panel is *actually* pointing when the controller is first powered on (and during operation).

The digital approach does make it easier for you to tweak the behavior e.g. switching off the tracker at night so the panels don't bother to follow the moon, etc.

The purely-analog approach has the benefit that the feedback is built into the design... it's pointing the panels based on "light actually hitting the sensors" and not based on the calculated and desired positions of the panels. If (as I guess) you're supposed to mount the two photocells on or by the panel, with a separating barrier which is normal to the plane of the panel, it ought to be a pretty bulletproof method of keeping the panel pointed almost directly at the sun as long as there is direct sunlight on the panel.

It may mis-aim under adverse conditions, I imagine - e.g. a heavy cloud right in front of the sun, and a brightly-illuminated cloud off to one side or the other. Under those conditions, though, this probably won't matter.

You could of course go with a hybrid approach... photocells as analog sensors, feeding into analog-capable inputs of a small fleabite microcontroller which would then run the panel-aiming motors.

I'd vote for the clock method.

I was outvoted and worked with some ham radio guys to build a tracker.

Your solartracker link describes a system that appears to work the same way.

I'd be interested in learning more about the system, the insoalation math, the financial math, magnitude of the system, latitude...

On Wed, 21 Oct 2015 17:02:29 -0700 (PDT), snipped-for-privacy@gmail.com Gave us: snip

Place a single sensor deep inside a tube, and that way incident light will not affect it, and it will stay on track by aligning to only the brightest signal.

You could also use a small solar panel (at the bottom of a tube)as the light sensor and use a voltage sensor as the threshold indication circuitry.

Another good one is to buy an old cheap telescope which had pointing ability and use that controller/positioner to keep the panel pointed at a specific star... the Sun.

The problem is that there are plenty of other sources of light besides sunlight. One of my past headaches was dealing with a solar tracker that preferred to follow automobile headlights. Since it was powered by the same batteries that the panels were charging, it turned off at night based on an RTC (real time clock) and lookup table. However, when it rained, the headlights from nearby automobiles would drive the tracker nuts. It has no provision for limiting rate of traverse, so wild swings in any direction were the norm. I "solved" the problem with a crude light shield, but that was far from a bullet proof and total solution. Watching it lurch its way across the sky on a cloudy day was ummm... interesting.

However, light trackers are really good for classroom demonstrations, where the small panels will follow any kid pointing a flashlight. Such a toy generated more interest in solar power and electronics than anything in my song-n-dance on alternative energy.

It will matter if you're moving a large array. I would guess(tm) about 30 lbs/panel (including the Al frame). A 15 panel array would be about 450 lbs. Swinging that around randomly with every passing cloud, could break or bend things. Slew rate limiting will slow down the damage, but if the motor is powered by the battery pile being charged by the panels, you could run the battery down somewhat due to almost continuous motion.

Why? Once you have an RTC (real time clock), you could control azimuth with a simple starting point (sun rise on the horizon) and then follow the sun at the rate of: 360 degrees per 24 hours = 15 deg/hr = 0.25 deg/min. I believe Stonehenge works on a similar principle. If you need or want elevation control, a simple lookup table based on your latitude should suffice. No need to modify the table after initial installation because solar tracking arrays don't move from where they're initially planted.

Incidentally, I once saw a home made tracker operated by a clockwork and escapement mechanism that lurched forwards 5 degrees every 20 mins. It was fabricated from junk, rust, gears, ropes, pulleys, solenoids, a big spring, and more junk. Every morning, a motorized block and tackle arrangement would tension the big spring, which would then power the contraption for the next 24 hrs. Timing was controlled by a not very stable 555 time running a solenoid that would release the escapement every 20 mins. I got involved when some of the solder joints cracked, the crimp connections fell apart, and one of the panels developed an ominous diagonal crack. It ran for about 10 years before being replaced with something more reliable (and less noisy). I took a few photos, but have no clue where I buried them.

Hi,

Can you direct me to the lookup table link or code example?

erica

It's a fairly simple navigation problem, especially since accuracy isn't needed. It's been almost 45 years since I took the celestial navigation course in college and have forgotten the details but this is fairly simple. Really, all that's needed is the azimuth of sunrise and your latitude to get close. A table copied out of an almanac should work.

Jeff gave the following Chart

for example it says that at 180 degree azimuth the sun elevation will be around 75 degree at 12PM. I am unable to calculate what angle will the solar panel be at.

On Wed, 21 Oct 2015 17:02:29 -0700 (PDT), snipped-for-privacy@gmail.com Gave us:

Buy this cheap computerized telescope and put the panel effector in place of the telescope after programming it to point at the Sun.

Actually, it's even easier than my first thought. If you use an equatorial mount, others have pointed out that a clock gets you the timing for the day (Right Ascension). All you need then is the starting point for each day.

You should be able to steal a table of azimuths (altitude will be zero) of the sunrise from any decent almanac. With a little math, this will give you a declination correction factor to start each day. You'll have to make your clock run in sidereal time (or correct that every day too) but that shouldn't be a problem.

If you start with an alt-azimuth mount, it just takes some conversion from your clock and starting azimuth to RA:DEC format.

I am really a beginner and this is my first every solar project.

I found the following website that can put together the time , altitude and azimuth.

I think this table can give the information but I do not know how to calculate the angle of the solar panel using all the information. I do not know about the terms sidereal time, RA DEC format and declination correction factor.

erica

The definition of the Alt-Azimuth system is degrees from the horizon (altitude) and degrees from North (Azimuth). So you have 180 degrees from North (or due South - which is the definition of "local noon")) and 75 degrees up from the horizon (or 15 degrees from vertical).

With an equatorial mount, that's all the information you need for the day. However, it might be easier to start with the time and azimuth of sunrise.

hour? That's a bad design, not an inherent flaw of sun trackers.

Method 1 always knows where the sun will come up in the morning. But, it may not point the panels in the right direction without some effort to get the mount perfectly aligned with gographical coordinates. For solar panels, though, exact pointing may not be at all critical.

Method 2 should give very exact pointing, but may have trouble swinging toward the sun in the morning. if it just stops when the sun sets in the evening, it will end up pointing almost exactly away from the sunrise. So, it most likely needs some sort of reset circuit to turn it back to the east for the morning. That could be done with a little timer and a switch that stops it when it gets back to pointing east.

Jon

use GPS and it can work out of the box.

you don't need hardware floating point. if you code in C or basic you should be able ti use someone else's float routines (from a library provided with the compiler)

It'll be slow, but should fast enough to track the sun.

Look for GoTo telescope add-ons, which usually be set to track the sky (23:56 h), Moon or the Sun (24:00 h).

If the accuracy requirement is not high, just mount the panels on an equatorial mount (axle parallel with Earths axle), then you need just a single motor running at constant speed and a gearbox to rotate 360 degrees in 24 h. Of course, there must be plenty of space below so that the panels can point downwards during the night. There is no need to worry in which direction the Sun is rising.

Just change manually the panel tilt, a few times each year.