Automatic Photovoltaic Tracker

On an area moved by the motor construct a shade pointing outward. On the east and west side of the shade place a photosensor, each of which is wired to a single input. Establish the relationship such that when the shade is pointed precisely at the sun both photosensors are lit. When the sun moves (yes, I know the earth is moving but it's easier to describe this way) towards west the eastern sensor will become shaded. Use this condition to move the array toward the west until the 'both illuminated' relationship is established again. This will continue until your 'west limit' is made. After a while both sensors will be dark. When this is true for a certain amount of time begin the nightime traverse back to the east limit.

This may be somewhat simplistic but it should work.

I Googled concerning this and it seems a time delay before performing a movement is desirable so that the motor is not constantly hunting. You may have to experiment.

bernie_carlton said:

On an area moved by the motor construct a shade pointing outward. On the east and west side of the shade place a photosensor, each of which is wired to a single input. Establish the relationship such that when the shade is pointed precisely at the sun both photosensors are lit. When the sun moves (yes, I know the earth is moving but it's easier to describe this way) towards west the eastern sensor will become shaded. Use this condition to move the array toward the west until the 'both illuminated' relationship is established again. This will continue until your 'west limit' is made. After a while both sensors will be dark. When this is true for a certain amount of time begin the nightime traverse back to the east limit.

This may be somewhat simplistic but it should work.

I Googled concerning this and it seems a time delay before performing a movement is desirable so that the motor is not constantly hunting. You may have to experiment.

Click to expand...

Dude, that's genius.

I have thought about this exact application a lot, and your method is the way to go. I had thought about two sensors feeding a self balancing bridge (used aerial antenna rotator) but I hadn't thought about just using a shade perpendicular to the sensors.

You could probably hardwire that circuit such that the geometry of the shade and sensor positions would end up being the interval and run time.

I would add a backup mode in which you follow a data table every 30 minutes or so. Just populate the data table with ideal position info you can calculate based on the direction of the pivot. You can find web sites that can tell you the exact path of the sun over your area. It takes a little more homework than Bernie's brilliant mode, but the advantage it might would be dealing with overcast weather.

Very cool Bernie. A shade! I Love It.

OkiePC said:

I had thought about two sensors feeding a self balancing bridge

Click to expand...

Okie, were you considering ordinary cadmium sulfide photo detectors as part of a Wheatstone bridge? Then using the resulting imbalance as a bipolar output?

but in case of a cloudy weather or a reflection ( a lake or the humidity of the plants), the system became unstable

Click to expand...

Oh, that pesky 'real world' stuff. How dare it mess up an easy idea.

The problem exists when a cloud for example, reflects the sun's rays directly to the sensor. That condition is translated to the PLC like: the tracker is exactly where it should be.
I could program the controller, not to make reverse movements until the west limit. What is your opinion about combining this method with the real time clock method?

is the tracking system motor powered by the battery system that is being charged with the solar panels? if so I would just use a lookup table. you dont want this thing hunting around, especially on a day with little sunlight.

I'm pretty naive about this (solar tracking) so let me ask. Assuming a polar mount (axis pointing toward - at least in the Northern Hemisphere - the North Star) and the array is pointed at the solstice elevation for the location, how much does the azimuth position vary from a pure timeclock position (one degree of movement every 4 minutes)?

And does the deviation really matter?

Late correction -

and the array is pointed at the solstice elevation

Click to expand...

should of course be the equinox elevation.

The elevation of the tracker would be standard. The deviation matters but it would be considered in the whole study. So i had to do only with, the rotation of the tracker. I have to follow the azimuth. I am waiting for that proportion that you mentioned, the degrees for every minute. When they told me, i will let you know.
With this proportion i could program the PLC. And at every f.e. 4 minutes give a command to inverter to move the tracker. But how the PLC would "know" the dawn?

I was presuming a Real Time Clock set according to the precise local noon. The table of positions would be set, perhaps as I said naively, to one degree every 4 minutes before and after that. The would set full East at 6:00 AM, whether your limits would allow that is another matter.

To what extent does that deviate from the true position of the sun (azimuth only)?

It's a little late to be asking if your PLC has a real time clock, don't you think?

On the assumption that you do have a real time clock with a calendar, you shoulod be able to get data on sunrise and sunset as a function of day of the year and your latitude. Then you can use the time of day to track east to west orientation based on that. (As it happens, these relationships were developed by the ancient Greeks.) If you have the data you can create an equation from Excell or another curve fitting program using regression analysis.

Similarly, the elevation corresponding to sun position will vary as a function of time of year - lowest in the winter, highest in the summer.

pontiosf1 said:

But how the PLC would "know" the dawn?

Click to expand...

A chart that has dawn and dusk times for your locale can be had at this site
http://aa.usno.navy.mil/data/docs/RS_OneYear.php

Of course your PLC would need to have a "real timeclock". As a substitute for a internal real time clock, a input connected to a real timeclock set to turn on the input at dawn on the summer solstice would work. All of the other dawn times would be later in the day.

I tried to copy and paste the chart but the formatting was wrong.

PS. The chart is based on Greenwich mean time.

Just for curiosity
1. Why are you doing this when this equipment is already on the market?
2. I have difficulty understanding how you are going to do this without a real time clock - either integral in PLC or external and then triggers PLC to start.
3. What is your allowable or desired error in tracking the sun movement.
4. Why not control for both axes to get best performance? Granted you can adjust vertical weekly or so for seasonal variation?

Seems to me the best is to have both axes controlled.

You can get tables from US Naval Observatory like Mildrone did, or from US Dept of Energy EREN group, or from local astrologer society or local library.

For simplicity I like Bernies idea of using photo detectors. Not only would they have to be shaded but they would need either to sight down a long tube to get best accuracy. Start and stop times could be done with these detectors or by an independent pair (one East for sunrise and one West for sunset). This just might work provided you are willing to readjust and calibrate once a week say.

Dan Bentler