Need Help in Finding Azimuth Sensor for Solar Dish

First of all thanks a lot for your valuable reply dvmdsc.

Ya. The system is big. It is 16m in diameter.

It is a wonderful idea using a light beam directly from the sun. Would this method work in during cloud cover??

When there is light, you can create a beam in a black box, that's how it works.

You detect the voltage on the cell and when it is the highest value, your system stops.

Just put your beam box in the same axis as your collector.

Don't use visible light, use IR to which sensors are usually 2X or more sensitive than to visible light. That way, I think you can use it to position even in times where there is some cloud cover. If the clouds occlude the sun, you will not need to track it.

I'd make the sensor field of view quite wide for those times when the rig is way out of position and the cloud cover finally disappears. And, of course, for those night to day transitions.

I think that clouds are more transparent to UV than they are to IR.

Something like a telescope clock drive might work.

Dunno why you need 0.1 degree???? what are you trying to do? If you just want heat then 0.1 degree is waaaay too tight, unless you are trying to use it to bond gold wires onto silicon wafers.
Del

You are right, 0.1 degree is too tight. We can have it to 0.3-0.4 degree accuracy.

Using an alt-az system, you can't use a clock drive. You'd need a polar system to use a clock drive.

It should be possible to use a table of solar altitude-azimuth positions and encode that into memory, then use that to control the position of the dish. There is probably an on-line calculator that can do that; if not, the equation is fairly straightforward and can be used to generate a table for any location on Earth. (I did it once, years go.)

I don't think it is really necessary to maintain solar alignment to within 0.1 degree. the 'insolation' (flux of light hitting the surface) goes as the square of the cosine, so a difference of 0.1 degree is a change of only 0.999997 from perfect alignment.

... I see while I was typing you'd come to the same conclusion on positioning accuracy.

Thanks a lot Usbport for your valuable reply.

Here the problem is not in finding the sun position but to identify the system position.

Right, that's why I think a look-up table would be best. No worries about overcast skies, you just need a system clock to know the date and time, then use the lookup table to know where to point. Once the system is installed on true north, the position of the sun is just moderately simple mathematics based on your latitude, longitude, and the date, which are all known quantities.

I might still have the equation in a mathcad file somewhere. If I can find it I'll post it. You should also be able to find it online.

This method is really interesting. But I feel in this way, as we are not taking into consideration where the system is positioning, if we blindly follow the lookup table, there are chances that in every step what if the motor has taken a small errors, in the end it would lead to a large error making the system not aligned towards sun.

For this reason we are planning to have position sensors so that we can trace the system position and make sure how much more to move so that the system be aligned towards sun accurately.

Then I suggest you google Sun Tracking Sensor. This problem has been solved many times in the past:

https://www.google.com/search?q=sun+tracking+sensor&tbm=isch&tbo=u&source=univ&sa=X&ei=5_hnUtS6KoytkAf8tYHABA&ved=0CEgQsAQ&biw=1280&bih=553

http://www.suntrackpro.com/

A system of 5 photo diodes in tubes and some clever seeking software to control the 2 motors will find the brightest part of the sky which, on a cloudy day, may not be directly at the sun.

This old thread may be of interest. I've been stashing away solar related theads for a rainy day... or do I mean sunny day?
I think a well designed reflector and collector (we still don't know the use of this thing).
Will be pretty tollerant of missalignment. I still think you're probably a factor of 10 too tight!
Del

Tracking the sun would be much cheaper than predicting it's position with calendars, incredibly accurate equipment and software.

Of course we don't know what the OP is trying to do, but there is always the point, as Oraka said, that on cloudy days the brightest part of the sky may not be on the direct route to the sun.

Perhaps the best compromise would be a combination: motors which keep the dish moving at approximately the correct rate during cloud cover, but are controlled absolutely when the brightest point can be tracked.

We did not get to the moon by steering toward it. We knew where we were, and where it was, at all times. I think you may even get an argument that accuracy is far superior in a predictive system, although I agree that is not a major concern in this app.

You are fully correct.

The first truly common sense post, other than the the one that specified that 0.5° is far too fine. Even 10° won't make any measurable difference......

You are aware that the parabolic device needs to turn about 270 degrees every day and also needs to tilt about 170 degrees? I think you'll end up with a high cost this way.

I have built one with 4 photovoltaic cells (that output voltage) in a Wheatstone bridge configuration. If you take quite big ones (like 1 inch) and make an array, there is a dark area in the middle where the balance is met.

It needs some testing but it works well.

A 5th cell is to confirm when it is dark. This is used to bring the setup to the start position (East oriented)

Parabolic systems are made to get all the sun the can get. Especially the steam systems.

Why does that matter? If anything, your statement simply verifies the physical challanges of sensing the required field, (multiple sensors) The required field of movement of the array is the same, whichever tracking system you choose.

no help , more trouble by googlin'he "gps low power alignment"

I have been experimenting with this same problem....the solution for me has been to couple a 10 turn, high accuracy pot mechanically to the azimuth and elevation so that full travel of the pot is full travel of the axis and then sense the pot position with the analog input to an arduino and send the value out the serial port for further use. The arduino's 10 bit analog gives 4 minutes of resolution.

I have recently found another microcontroller system with 12 bits analog resolution which is just $67 U.S from Mouser and it is basically a direct replacement for the arduino and will run the same code when recompiled. It is called the "gallileo". These devices both use a subset of the 'java' programming environment from Sun microsystems.