Backtracking algorithm for solar tracker

Support forum for the Energy Matters Solar Analyser; a FREE solar power site evaluation tool for the iPhone and iPod Touch that provides full solar site analysis (Australia only)!

Re: Backtracking algorithm for solar tracker

Postby Tracker » Mon Jan 27, 2014 8:01 am

.........Several (or many) degrees, error in directly pointing at the sun is not going to make any measurable difference to the collection of incident solar radiation....


KISS.... comes to mind..
Dynamic biaxial tracking is ideal,, but surely, not necessary.

I have an array that I adjust for season..
My main Grid Connect system is seasonally adjusted, for angle-of-dangle..
I can't see any incremental difference from one step to the next.
I do see the difference from one extreme to another..

........Simplest possible system would have to be a propane mass weight transfer tracker.
Next step up would be a SINGLE AXIS dc motor driven optical self seeking tracker...


So, were I experimenting with a tracker, I would start with a seasonably adjusted elevattion and linear position control..
KISS..... we know what time the sun will rise and set..
So a simple direct drive control for time of day... no cloud issues, no edge effect causing false bright spots..

So, with KISS in mind, what percentage difference would you get by using a super accurate biaxial controller.???
..
.
Retired Engineer and keen PV experimenter - Always ready to learn and share.
2 x CMS2000 (fan cooled) GCI and SE 170W panels
1.7kW First Solar/Outback Island circuit - Peak Replacement Power
Governments won't save the world :-) They will just TAX it :-(
Tracker
Solar Crusader
Solar Crusader
 
Posts: 5067
Joined: Wed Sep 23, 2009 6:54 am
Location: SYDNEY --- EA - Network, Retailer - EA

Re: Backtracking algorithm for solar tracker

Postby Warpspeed » Mon Jan 27, 2014 8:25 am

Tracker wrote:So, were I experimenting with a tracker, I would start with a seasonably adjusted elevattion and linear position control..
KISS..... we know what time the sun will rise and set..
So a simple direct drive control for time of day... no cloud issues, no edge effect causing false bright spots..
.

Definitely agree with this.
Self correcting "smart" systems sound great in theory, but they tend to panic and go nuts when they momentarily lose whatever inputs they need to self correct.

Doing some serious thinking about MPPT right now, and am coming to some very similar conclusions.

When the sun suddenly goes behind a dark cloud for a few seconds, then suddenly emerges again, a software tracking MPPT goes absolutely frantic.... trying to figure out why the power is dropping off in both directions of correction.
Warpspeed
Solar Crusader
Solar Crusader
 
Posts: 345
Joined: Fri Jan 03, 2014 4:18 pm

Re: Backtracking algorithm for solar tracker

Postby Tracker » Mon Jan 27, 2014 9:30 am

Warpspeed wrote:.....When the sun suddenly goes behind a dark cloud for a few seconds, then suddenly emerges again, a software tracking MPPT goes absolutely frantic.... trying to figure out why the power is dropping off in both directions of correction....

But, that is not the issue in my mind.. My CC likely has a panic attack, but it only takes seconds to readjust, (depending on the algorithm used)..
I CAN see a sophisticated tracker "Looking" at the bright fringe effect of a cloud and going crazy, from R to L and up and down.. (not that you would alow such dumb changes..

No, I go back to my un-experimented question... "What percentage difference between full biaxial tracking and a fixed single-Axis linear control"..
Could not be simpler... a heavily geared drive and a DC motor.. Change the voltage to vary the speed

My strongest of gut feelings is that you would not see the value of all the tracking technology..

Don't get me wrong - Gordon will do it with ease.... but I get tired of trying to keep all the modern technology working right, as it is now..
I've seen trackers, on farms, where they stopped once too often, and now sit forlornly facing NORTH..
..
.
Retired Engineer and keen PV experimenter - Always ready to learn and share.
2 x CMS2000 (fan cooled) GCI and SE 170W panels
1.7kW First Solar/Outback Island circuit - Peak Replacement Power
Governments won't save the world :-) They will just TAX it :-(
Tracker
Solar Crusader
Solar Crusader
 
Posts: 5067
Joined: Wed Sep 23, 2009 6:54 am
Location: SYDNEY --- EA - Network, Retailer - EA

Re: Backtracking algorithm for solar tracker

Postby Warpspeed » Mon Jan 27, 2014 9:45 am

Tracker wrote: No, I go back to my un-experimented question... "What percentage difference between full biaxial tracking and a fixed single-Axis linear control"..
Could not be simpler... a heavily geared drive and a DC motor.. Change the voltage to vary the speed
..
.

Should not be any difference, if a single axis tracker is aligned exactly with the earths rotation.
Its not only simpler, it is potentially much more precise.

The earth is not mounted on gimbals !
it cannot suddenly start rotating north/south, it ALWAYS rotates exactly easty/west in exactly 24 hours.
And your tracker only has to do likewise and follow that same exact never changing motion.

If you have grid power, a small mains synchronous motor will work as well in your tracker as it does in a mains powered wall clock, and keep absolutely perfect time, without needing any correction at all.
The power utility actually correct the 50 Hz grid frequency if it starts to run ahead or behind time.

Even a really tiny motor geared down to one revolution in 24 hours will generate immense torque with a suitably strong gearbox (or sprocket) in the final stage.
You just need to flick it into reverse rotation every 12 hours with a precise time clock, and you are set to go.

Even if you stuff up the gearing calculation of your constant synchronous speed tracker, and it only travels (say) 171 degrees instead of the ideal 180 degrees in 12 hours, it will still track without adding any cumulative error if it reverses at EXACTLY 12 hour intervals.
Warpspeed
Solar Crusader
Solar Crusader
 
Posts: 345
Joined: Fri Jan 03, 2014 4:18 pm

Re: Backtracking algorithm for solar tracker

Postby Warpspeed » Wed Jan 29, 2014 8:48 am

Some further ideas on building a reliable single axis sun tracker.

My solution (I am actually in the process of putting all this together right now) is to digitally divide the 50 Hz mains frequency down by 32 to generate 1.56 Hz.
That powers a common 5 watt 200 steps per rev stepper motor, which drives first through a small 25:1 gearbox, which I have yet to source, then through a much larger 54:1 gearbox that I already have here.
That little stepper motor works out to 750+ ft/lb of available torque after final gear reduction !

That will produce an exact one revolution in 24 hours with that gearing, with zero long term time drift, as the power grid frequency is constantly being corrected.

I plan to try a homemade mercury tilt switch made from a long glass tube with wire whisker contacts at one end, and a blob of mercury.
At solar noon the glass tube will be vertical with the contacts pointing upward.
At solar 6pm or 6am the glass tube will go exactly horizontal, and the mercury blob will run down hill and short the wire whiskers together.
This will toggle a flip flop and reverse the stepper motor direction.
This has the potential for accuracy without requiring any adjustment to keep it correctly timed.

It should be able to run continuously forwards for 12 hours and reverse for 12 hours and hopefully continue to track faithfully over several days of total cloud cover and zero sun.

It will also have a pair of sun sensors consisting of two ordinary green LEDs to sun track in the normal way, but only if at least one of the sensors detect full bright direct sunlight.
When there is strong direct sun to track, the stepper motor alternates between either twice normal speed, or half normal speed to stay pretty much locked onto the sun. This is easy to do by dividing the 50 Hz mains frequency down by either 16, 32, or 64 to get twice, normal, or half speed.

The way I have arranged this, my mains frequency digital counter chain steps through a short 16 counts long lookup table located in EPROM.
This lookup table contains the bit pattern to operate the stepper motor windings in correct sequence.
There are different lookup tables that contain different bit patterns for fast, normal, and slow speeds, as well as forward and reverse, the mains frequency counter itself does not change count direction.

These different lookup tables are selected by EPROM address lines which are controlled by the two LED sun detectors and some voltage comparators.
The electronics have already been built and tested in prototype form, and run the stepper motor exactly as expected.
I just need to buy and adapt a small 25:1 geabox which should not be difficult.
Warpspeed
Solar Crusader
Solar Crusader
 
Posts: 345
Joined: Fri Jan 03, 2014 4:18 pm

Previous

Return to Solar Analyser Support

Who is online

Users browsing this forum: Alex Morrow and 1 guest

new solar power specials