Hi,

(this is my first post on this site)

Just wanted to share my solar tracker project that I have been tinkering with for a few years now.

This is my 3rd version of the tracker. The first being my final year Electrical Engineering thesis project. The first version worked ok but was a bit of a chunker really. It used open circuit voltage of the solar panel to determine the brightest part of the sky. To measure the open ocv I had to disconnect the load. This was one of its major drawbacks. Its other main issue was that it didnt cope well with cloud cover. On a cloudy day it would find the brightest part of the sky, but this was often well away from the suns direction.

After I finished uni I decided that I would make another tracker. This time I used a sun position algorithm to determine the elevation and azimuth of the sun given time and position data. I used a real time clock coupled to a PIC 18f252 micro to do this. It worked really well. Much better that the first version. Sticks to the sun like glue

Being a tinkerer I decided to improve on my design and use a GPS receiver to calculate all of the time and position data. This works like a treat. Basically I can now just turn the tracker on at any time and at any location and it locates the sun. The tracker must be levelled and face toward true notrh before it turns on. That is all the adjustment that is needed.

The micro calculates the sun position about 10 times a second and when there is a certain amount of discrepency between mesured angle and the angle of the panel a re-adjustment occurs.

I have attached a .ppt presentation that has a few photos.

If anyone would like any more info please get in touch as I would love to share what I have learnt along the way.


Regards,

Paul Hatfield.

Hi Paul,

Welcome to the site its not that active, but hey, its still new.
I built the site myself for those solar enthusiasts like us.

About a year ago i also made a tracker, which tracks the optimum angle using 2 smaller 1/2 watt solar panels that were slightly angled apart (just by about 2degrees), and calculates which angle puts out more power, then moves towards it.

The problems i faced were:
1. The 1/2 watt panels were not matched and their outputs needs calibration in software
2. Over time, if one of them got dirty, tracking efficiency would be extremely poor
3. All over the place on cloudy days
4. affected by heat
5. Too much tinkering involved, had to measure multiple outputs.
6. unit was large, and needed to be set up correctly otherwise it'll fail.

Your design overcomes these problems with added benefits. Also, looks very reliable and accurate, the use of that non-contact encoder and 10bit resolution is a great idea. I seriously think this system will see great commercial success.

You have inspired me to re-commence work on my tracker .

Just a few quick questions, were did you get your actuator from? and would it have enough power to turn 3 225watt panels (they seem roughly around twice the size of your solarex panel)?

Also, what is that fan doing? is it a fan?


did you program in C or assembler? and how hard was it to interface the GPS to the pic.

Have you considered adding a MPPT to the output to further increase your output power?
If you are going to use a MPPT, and you do measure the direct output power of the solar panel, you can try experimenting with this algorithm, its similar to what we use in mppt's.


What do you think about it?

Great work, truly the best homebrew tracker ive seen so far.

Hi Tony,

This site is a great idea. I would love to have a solar setup one day at home. Maybe with my solar tracker running the show

I had pretty much the same problems with my first tracker as you did with yours. On a clear sky day it worked great, but as soon as there was a few clouds about it started to struggle.

I have come to the conclusion that trying to determine the position of the sun using feedback from the panel is just too hard. Thats why I decided to go down the path of determining the position of the Sun mathematically and then just telling the tracker where to go.

The 10 bit encoder is a fantastic thing. It is made by a company called Austria Microelectronics. I just use the SPI port on my micro to read the 10 bit data and then convert it to an angle. The great thing about the AS5040 is that it is totally contactless and its easy to set up and get going. All you need is the special magnet that you can order from Austria micro.

I purchased the Ram from a place in Perth called Linear Bearings-
http://www.linearbearings.com.au/Contacts/tabid/72/Default.aspx

Not sure on the specs of the Ram but its quite powerfull. I am quite sure that it would turn 20 of my panels with ease. Only problem is the holding power of the ram in windy conditions. I havent done too much thinking about it but it would be good if you could lock panels in place when the wind gets over a certain velocity. That way the pressure would be taken off the gears in the linear actuator.

The picture that you have shown is of my original tracker controller. This is the one that needs to disconnect the load every time it wants to scan for the sun. I had a relay between the fan and the micro that I used to disconnect it. Basically the fan was for my own enjoyment and serves no practical purpose. I am easily ammused.

My first tracker was programmed in assembly but now I use C. I totally love C and have become a little addicted. Its just such great fun coming up with new ideas for microcontrollers. Even if they have allready been done by a million other people.

I had thought of using a MPPT for tracking the sun bet decided that it may be a bit much for the final year project. It was my first attempt at microcontrollers and what a learning curve. Took me 3 weeks to get the damn LCD screen going in assembly. That was actually the hardest part of the project

Thanks for the positive comments about my tracker. I have never thought about it as a commercial product as I figured that there would be better ones out there anyway. Maybe I should look into it a bit more.

Regards,

Paul Hatfield.

Wow that looks like an excellent project paulhat,

too bad it runs off a pic not an atmel... let me know if you want me to help you convert it to a more superior microcontroller...

The gps is a really good idea, how does it handle it if you move to a different time zone? Does it need recalibrating or do you have coordinates for the different time zones programmed into the device?

Keep up the good work.

HI,

Time zones are not a problem as the sun position calculation uses GMT time. This is also the time that the GPS spits out so its all good.

Tony I forgot to answer your question on interfacing GPS with the micro. Its real easy to do. The GPS that I am using (EM-406a)
spits out ascii at a default baud rate of 4800. I think there are about 8 strings in total and I use the GPRMC string which has time, date, coordinates etc. Each string starts with a dollar sign then the string name. So for GPRMC its $GPRMC. Each of the values in the string are separated by commas.
All I do look for the start of the string by checking for the dollar sign followed by GPRMC. Then I read the entire string into an array then count commas to find the value that I would like. Maybe there is a much better way to do it but this way seems to work fine.

Regards,


Paul.

Sounds very interesting, im gonna start work on my own GPS tracker now as soon as my holidays come, about 5 weeks. i think im gonna need your help.

Its cool. like a plug and play tracker, except in the northern hemisphere youll need to face it south .

paulhat wrote:Thanks for the positive comments about my tracker. I have never thought about it as a commercial product as I figured that there would be better ones out there anyway. Maybe I should look into it a bit more.

Regards,

Paul Hatfield.

Paul,
At the very least you could sell the schematic, parts list and compiled code to hobbyists. If you set up a website, I bet you could get it listed on sites like: http://www.redrok.com/main.htm and http://www.builditsolar.com/

For stationary installations, in regions where "atomic clock" radio transmissions are available, an RF time receiver like this one: http://search.digikey.com/scripts/DkSea ... 61-1015-ND is about 1/3 of the cost of the GPS unit. Of course the coords would have to be hardcoded uniquely for each install.

I have to say I'd love to see this in a dual-axis application. Since you can compute both elevation and azimuth, it'd just be a matter of adding the control routine for the 2nd axis.

Best of luck,
Mike

Hi Mike,

Thanks for the comments and ideas. You have made me think about re-designing the tracker so that I could sell it as a kit.

Tony, did you start your tracker project? If so, how is it going?


Paul.

Hello Paul.

I have done similar solar tracker controller which works sensorless with special sun agorithm and no photosensor. It is single axis. I am implementing GPS module right now. Before i were using real time clock with I2C communication with microcontroller.
Special algorithm calculates azimuth sun angle and altitude sun angle, sunset time, and sunrise time for single day in the year. The program is written in C and implemented in freescale 8-bit microcontroller.

I have found the idea on this site: http://bluwiki.com/go/Sun_Tracker

Everything work great now for almost one year. I am only afraid that tracking is not optimal. The reason how I found this site is that I noticed that my single axis tracker is not only changing zimuth angle but it is also changing altitude angle while moving form east to west. The table of azimuth and altitude angle which I can track is in the table below.

I am asking if you have any experiences how to consider both angles in agorithm, to achieve optimal positions during the day?

When does your tracker start and stop in these days and what are your GPS coordinates? My tracker will tomorow start at 7:50 and stop at 16:50. My coordinates are 46° east from greenwich and 14° north from equator. The time zone is GMT +1.

Can you please calculate the time when your tracker start moving for the data above. The azimuth agle which I can track is 120° to 240°.

Motor(mm) Azimuth Altitude
0 119,82 27,69
50 127,37 32,6
100 135,29 36,82
150 143,13 40,12
200 151,22 42,72
250 159,72 44,66
300 168,63 45,92
350 177,89 46,47
400 187,34 46,26
450 196,84 45,24
500 206,19 43,39
550 215,28 40,69
600 224,05 37,13
650 232,55 32,64
700 240,08 27,66

Sory for my bad english.

Best regards,

Matej

Hi Matej,

I am not really sure if I understand your problem. It is normal for sun altitude to change throughout the day so it seems that your program is working correctly.

There are many sun position calculators on the web that you could test your values against. I just my tracker when the sun position is within the mechanical limits of the tracker itself. Do you have any pictures of your tracker that you could post?

Sorry if I havent helped you much here.

Regrads,

Paul.

Hi Paul.

Web sun position calculators calculate exactly the same results like mine program and I am sure that the algorithm works just fine.

Yes I know it is not easy to understand my question. I will try to explain again.

My tracker (looking from mechanical point of view) is able to track sun azimuth angle from 120°to 240° and the altitude angle from 27° to 47° and than back to 27°. It is single axis but with the way of moving it changes both angles.

My question is just how to consider both angles in algorithm to achieve optimal position. Because since now I considered just azimuth angle:
I know the time when the azimuth angle is 120° and the time when azimuth angle is 240°. Then I am moving the tracker with linear curve from the time when the angle is 120° to time when the angle is 240° every 15 minutes. I have hall sensor on the actuator which tells me the lenghth of my moving step. It is very simple.

But I am not sure if this is optimal - if you know what I mean. So I would like to consider also altitude angle, but I am not sure about relation. I was just asking if you consider both angles or is your tracker moving just by azimuth angle?

I hope you understand me better now,

Thank you again

M

Hi Matej,

The way that my tracker works is by constantly (every minute or whatever) determining the position of the sun and then moving the panel to this position. I use a 10bit encoder over SPI to calculate the current mechanical position of the panel.
I figure that if you have a GPS or RTC on board you may as well just do what I do. This way you dont have to worry about creating any code that would optimise your linear method.

Any Pics


Paul.

Pictures...

Hi Matej,

That looks like a beautiful part of the world you live in. Where is that?

The panels look awesome. I would love to have an array like that one day.


Paul.

Thank you Paul.

These pictures were taken in may 2008. Now is about 30cm of snow here and it is also very nice. (about -2°C)

If you ae interesting in more pictures I can send you web address of the pictures to your e-mail.

I live in Slovenia: http://www.slovenia.info/

However I have optimised the tracking agorithm now and I hope it will work OK now.

Where are you from? What are the prices of solar plants per watt in your country? What is the price of the energy if you are selling it to public electric network?

Have a nice time,

Matej

Hi Matej,

I live in Tom Price, a small mining town that is 1500k's from Perth in Western Australia.

At the moment electricity costs 13.94 cents per kWh and we can sell it for 12.67 cents per kWh. The problem here is the cost of panels. We pay about $1000 for a 100W panel.


Paul.

Hi Paul, I found your post and would like to talk about your GPS tracking program and the circuit you designed to power the actuator to track the sun, I'm working on a slewing drive bearing to rotote the rack the solar panels are mounted on, the pipe is 6 inch ID and there are 8- 208 w panels on the rack, the drive motor is 24 VDC and has a built in encoder, I look forward to your reply, Thanks

Hi Solarguy,

I would be happy to share info about my tracker. If you could be a bit more specific about the things that you would like to know as it would take me ages to describe the code in detail.

I have just used a H-Bridge (L298) to run my actuator, but from memory its only good for about 10 amps. Solid state relays would be a better option if you have larger currents in mind.

Do you have any pictures of your setup?


Regards,

Paul.

The H Bridge would do just fine, the motor is 24 VDC at 5 amps, Once you program the CPU will it work anywhere in the world ? Just point the panels south in our case and turn it on ? I can built the circuit board if I had a parts list and schematic and would be happy to work with you and get a pc board created using a company AlbertaPrintedCircuits, APC, I have used them before to make boards, I'am waiting to find out more information about the encoder that is built into the motor, I will post that information in the next day or so, I was looking on line and found another product that is like the one you built. The PDF is located here:
http://mono.foundryweb.co.uk/mono/downl ... e/ART180(1)%20-%20GPS%20Tracker.pdf

Does your actuator require limit switches to operate or are they just for safety incase the H drive applies power to prevent the actuator from being damaged ?

Once the panels are setup facing South and power is applied, the unit will position them towards the sun, and then follow it till dark, The CPU tells the actuator how far to get to the West limit, it then stops and then returns to the East limit for the next day sun rise ? Does that sound correct ? Can an input be programmed in so when the input goes low or high it will return to South position ? From your PowerPoint it looks like a simple designed circuit, I can make a board if I had a schematic to follow, If you put your program into the CPU how hard is it to do an upgrade of code , I have included a picture of the Slewing Drive , Thanks, Brad

Hi, Paul I have been trying to contact you to find out more on the building of the GPS system into my slewing drive, can you email me brad@ridgecommunications.ca ? Thanks