# Multi-Wavelength Detecting Solar Tracker as Thesis work - A Cascade of Questions

Hi. As a thesis work from my university I will be studying solar tracking devices, and the idea is that I also construct a prototype solar tracker (controlled by Arduino) to prove/disprove my theoretical simulations on tracking methods etc.

Now as I understand it the normal most basic way to track is simply to follow a predefined path for the sun or, as I’ve seen on youtube, using photoresistors to determine where the light is coming from. What I will examine in my thesis is slightly more complicated. Since PV-panels are able to produce electricity better from light with a certain wavelength than others, my idea is to have many different kinds of sensors registering different parts of the spectrum, and based on the collected data a weighting process takes place to determine which tilt/orientation would be most effective to produce electricity. One extreme (and unlikely but rather cool) situation that could occur with a system like this is that the tracking device choses to be oriented towards a place that appears to be completely dark to a human, but since its detecting a strong presence of infrared light it would be able to produce electricity anyway.

This may seem a bit ambitious, especially since I am a beginner with Arduino. But if I just stay positive and ask help from the right people I’m sure I will learn a lot in the process and the outcome will be interesting and cool!

So just to get my design process going a bit and maybe know what stuff I should order from ebay I would like to ask a few questions that you Arduino experts can answer:

• First of all: is it even possible to ask the Arduino to collect data from several types of sensors and calculate a weighting between them and control a motor based on that? I can’t imagine that this would require a lot of computing power, but it is something that the Arduino is capable of?

• If I use photoresistors for this (is that the best option btw?) will I be able to get an absolute value from them for example: “200 W/m² of a certain wavelength is reaching the photoresistor right now”. Or will I only be able to get some strange relative scale that is difficult to interpret like: “On a scale from 1 to 100 this resistor is getting hit by 50 light right now?”

• It would be good if everything the Arduino measures could be logged and exported to a computer so I can assess the data and what it is doing, it that easy do to and what would be the best solution? SD card connected to the Arduino?

• On videos I’ve seen on tracking devices using photoresistors, they only use one or two resistors. In my mind that doesn’t make sense, how does that work? Wouldn’t you really need like 360 photoresistors (one per degree) positioned in a circle to determine where it is brightest? And that’s just for a one-axis track, let alone dual-axis. Or is there a way to detect from which angle light is coming from in an easier and better way?

• I have been thinking about using stepper motors to control the movement. But from what I understand they don’t “know” themselves how they are positioned, so is it impossible to command a stepper motor to for example “go to inclination 20°”, since it doesn’t know where it is right now, is that correct and is there a work around for that in that case?

• From what I have seen a lot of people are using linear actuators instead of normal stepper motors when building tracking devices, is there any benefit to that or is it just a matter of taste?

Obviously I have millions of more questions but I suspect that I will scare people off if I keep writing on this little novel right now.

The sun follows a fairly predictable path through the sky. So you only need to setup your movement to follow this path. The angle of the axis of this movement will be determined by your geographic latitude and the time of year.

So once you have this axis set you only need to ensure that your array moves at the same speed as the sun across the sky. The obvious way to do this is to have one photoresistor aimed slightly in advance of the array and another slightly behind the array. When the array isn't aligned perfectly, one of these photoresistors will recieve more light than the other and therefore give an indication of which direction you need to move.

petters:
• First of all: is it even possible to ask the Arduino to collect data from several types of sensors and calculate a weighting between them and control a motor based on that? I can’t imagine that this would require a lot of computing power, but it is something that the Arduino is capable of?
YES, but you may need a Mega with 16 analog inputs for photoresistors

• If I use photoresistors for this (is that the best option btw?) will I be able to get an absolute value from them for example: “200 W/m² of a certain wavelength is reaching the photoresistor right now”. Or will I only be able to get some strange relative scale that is difficult to interpret like: “On a scale from 1 to 100 this resistor is getting hit by 50 light right now?”
Nothing simple - I don't even know if photo resistors are linear. Maybe you could have a pair that you compare with each other. I don't know anything about photoresistor frequency response. I presume you plan is to use filters.

• It would be good if everything the Arduino measures could be logged and exported to a computer so I can assess the data and what it is doing, it that easy do to and what would be the best solution? SD card connected to the Arduino?
Logging to an SD Card is very practical but it might make more sense to have a cheap laptop permanently connected.

• On videos I’ve seen on tracking devices using photoresistors, they only use one or two resistors. In my mind that doesn’t make sense, how does that work? Wouldn’t you really need like 360 photoresistors (one per degree) positioned in a circle to determine where it is brightest? And that’s just for a one-axis track, let alone dual-axis. Or is there a way to detect from which angle light is coming from in an easier and better way?
Usually the pair of detectors moves with the panel - the logig is to move toward the one with the most light until they both detect the same light level

• I have been thinking about using stepper motors to control the movement. But from what I understand they don’t “know” themselves how they are positioned, so is it impossible to command a stepper motor to for example “go to inclination 20°”, since it doesn’t know where it is right now, is that correct and is there a work around for that in that case?
The Thread stepper motor basics may be useful. Normally you need a process in which the stepper moves to a limit switch to establish the 0 position. After that it can keep position as long as the Arduino is powered up. Stepper motors are very power hungry and for that reason not a ood idea for a solar collector. A DC motor with a non-reversible worm drive may be a better idea.

...R

Robin2:

...R

"YES, but you may need a Mega with 16 analog inputs for photoresistors"

You mean because I will need a lot of input ports, not because of a higher computation capability in the Mega? Right now I have a Uno, I’ve read something about it being possible to extend the number of input ports, for example with this:

Would that work? And is there a way that it could be determined outside of the Arduino somehow which resistor is getting hit by most light and then only one input goes to the Arduino telling it which one that was? Say that I need to go with a solution that involves hundreds of photoresistors, could I avoid the need for hundreds of Arduino input ports somehow?

"I presume you plan is to use filters."

Filters? I’m not sure what you are referring to. What is that?

"Usually the pair of detectors moves with the panel - the logig is to move toward the one with the most light until they both detect the same light level"

But with this type of setup the tracking device would never be able to tell if it’s missing the optimal orientation by an awful lot? right For example if the best place to be is 90° from the current position it would never get there without a bright source leading it there right?

petters:
But with this type of setup the tracking device would never be able to tell if it’s missing the optimal orientation by an awful lot? right For example if the best place to be is 90° from the current position it would never get there without a bright source leading it there right?

Because the sun is not a bright source?

KenF:
Because the sun is not a bright source?

Well imagine that it follows the sun until sundown one day, the next morning it would still be stuck in the position where the sun went down and totally miss that it is shining at the backside of the panel right?

Anyway, that’s not really the point. I’m assuming that you didn’t read my original post, you seem to be under the impression that I want to just follow the suns path – but that is exactly what I want to avoid. The purpose is to have the tracker making an informed decision as to how it should be oriented regardless of where the sun is.

If you live in a sunny cloudless place I can imagine that it sounds quite counterintuitive to you that the best orientation would be anything other than directly at the sun, but this tracker is to be used in a northern climate where diffuse and reflected radiation is dominant over direct sunlight and a lot of the time it’s so overcast that you can’t even tell where the sun is.

where diffuse and reflected radiation is dominant over direct sunlight and a lot of the time it's so overcast that you can't even tell where the sun is

How much energy is theoretically or actually available in these diffuse and reflected sources?

Is it worth the expense to install solar panels if you can't even figure out where the sun is?

petters:
If you live in a sunny cloudless place I can imagine that it sounds quite counterintuitive to you that the best orientation would be anything other than directly at the sun, but this tracker is to be used in a northern climate where diffuse and reflected radiation is dominant over direct sunlight and a lot of the time it’s so overcast that you can’t even tell where the sun is.

The resistance of a photoresistor is affected by a pretty similar process to that which produces energy within the solar panels. If the photoresistors are telling you that there's more energy coming from one direction than another, I doubt that the solar panels will disagree.

Although the peak frequency may vary a bit, you could always use smaller photovoltaic cells as your sensors instead of photoresistors. These will definitely have an output that is directly related to the power that would be available to your main panels.

I'd be interested to see how much this varies in relation to the direct position of the sun. My guess is that it won't. It might be "intuitive" to think there's more energy coming through a break in the clouds (where you can see a small blue patch) but I doubt that's the case. Even the attenuation of the thickest of clouds in front of the sun will still not bring down the energy levels to that of the surrounding sky.

Interesting project.

Don't forget you have mechanical limits on your solar panels. Size and locations that they can be faced. Move them to far and then will shade other panels. If you are working with only one panel things are different.

Detector options are endless. A general detector can be used for wide band area detection and a second can be for fine tuning. You don't need one for each location. Lenses could also be used to give a wider sensor angle. Use a small panel as your fine tuning detector. Tune for maximum voltage/current.

Stepper motors are precise. If you have a zero reference point, you can count the pulses to know the exact location. Read the stepper motor description. An encoder is another option.

You have to do the calculations with the gear ratios, angles and positions to get the motor counts to know where it is pointing. I remember basic analog solar tracking systems from the 80's. Look at astronomy telescope mounts for some tracking ideas. That is done all the time.

i saw a study somewhere that said that tracking the sun path mathematicly based on the known solar geometries is better than scanning for the bright spot.

scanning is time and power consuming. The report said that the update frame rate was once per hour sufficient for maximum return in power invested in scan and position feedback

in other words, i hate to disapoint, but this has been studied before by others, it is not new science and the research proves there is not adiquate return on investment. similar to a perpetual motion machine

Greensprings:
i saw a study somewhere that said that tracking the sun path mathematicly based on the known solar geometries is better than scanning for the bright spot.

He isn't only scanning for the brightest spot. The discussion 'brightest spot vs mathematical calculation' is one thing, but different his second requirement: finding the spot with the best spectrum.
And then comparing if which of those two places two produces a higher amount of power.

Hi

As a thesis work from my university

I hate to ask this what experience have you in electronics programming and hardware.

"I presume you plan is to use filters."

Filters? I'm not sure what you are referring to. What is that?

This worries me, one of your methods of tracking is to look at different parts of the spectrum from the sun to try and calculate solar position, you use filters to isolate the various parts of the spectrum with your sensors.

KenF in reply #1, explained to you how the traditional tracking systems work.

But with this type of setup the tracking device would never be able to tell if it's missing the optimal orientation by an awful lot?

Its because you are measuring directly the solar radiation from two sensors, that are mounted on the moving PV array, and aiming to get their responses equal that you get the optimal position for maximum radiation.
If they were not aimed at the optimal, that is HIGHEST level of radiation, then one sensor will receive more radiation than the other and so the control system would respond and re-align the PV array.

Tom......

You might consider using a webcam and diffraction grating / prism to capture a picture of the spectrum from which you can then use to determine the optimum position. As it appears that electronics is not your forte then you might find it easier shifting all the difficult stuff into a program run on a proper computer. See: http://publiclab.org/wiki/spectrometer

CdS photoresistors are most sensitive to red wavelengths and other types of light sensors will have sensitive regions as well. If you go that route you're I think you're going to ruin your results. For a good "control" in your experiment you should probably be calculating based on date/time/latitude.

petters:
You mean because I will need a lot of input ports, not because of a higher computation capability in the Mega? Right now I have a Uno, I’ve read something about it being possible to extend the number of input ports, for example with this:

Mux Shield II - DEV-11723 - SparkFun Electronics

Would that work?
I reckon it would but I only know what I got from a quick glance at your link
Mega is 16MHz like Uno but has more SRAM

And is there a way that it could be determined outside of the Arduino somehow which resistor is getting hit by most light and then only one input goes to the Arduino telling it which one that was? Say that I need to go with a solution that involves hundreds of photoresistors, could I avoid the need for hundreds of Arduino input ports somehow?
Any alternative is going to involve a lot of design and soldering - up to you to decide which is the most economic option taking account of the value of your time.

"I presume you plan is to use filters."

Filters? I’m not sure what you are referring to. What is that?
I was assuming you were going to use coloured filters to limit the light frequencies that are seen by different sensors

"Usually the pair of detectors moves with the panel - the logig is to move toward the one with the most light until they both detect the same light level"

But with this type of setup the tracking device would never be able to tell if it’s missing the optimal orientation by an awful lot? right For example if the best place to be is 90° from the current position it would never get there without a bright source leading it there right?

I made that comment to explain how a system works with 2 sensors. It is up to you to decide whether that is adequate for your purposes.

Why not make several small moving sensor platforms each dedicated to its own wavelength and then have a means to decide on the best choice among them. These could be very cheap - just a small servo and a couple of LDRs.

That might also simplify the wiring connections - you could perhaps use a small Arduino for each platform and another master Arduino to collect the data. This may also answer your question about determining stuff outside the Arduino - in this case outside the master Arduino

Although, after another 3 minutes thought, a single Uno or Mega could probably control several moving platforms.

...R

jremington:
How much energy is theoretically or actually available in these diffuse and reflected sources?

Is it worth the expense to install solar panels if you can't even figure out where the sun is?

Well of course I can figure out where the sun is even on an overcast day, but the question is if it is still the best place to be directed at. Some previous studies I’ve done on this indicates that ground reflected radiation could be a pretty significant factor in these situations if the ground is of a bright color, for example snow. In the climates I’m looking at about 50% of the annual total radiation is diffuse, the rest is split between direct and ground reflected radiation (so we are generally talking about a few hundred W/m²). Direct radiation has ofc the highest peaks in the summer, but this is also the time when the electricity price is at its lowest, so if selling electricity is the objective this might not be the financially best thing to aim for.

KenF:
Although the peak frequency may vary a bit, you could always use smaller photovoltaic cells as your sensors instead of photoresistors. These will definitely have an output that is directly related to the power that would be available to your main panels.

That is actually a pretty great idea if I can find small PV-cells with the same spectral response as the big one being tracked! Do you know if very small PV-cells at about the same size of a sensor exists? I can imagine a downside of this is that I will never know, or be able to log, the wavelength distribution of the incoming light, it would only tell me the resulting best direction right? (Which of course wouldn’t matter in a normal case, but since this is an academic situation it would be nice to be able to log and analyze things as much as possible)

MeSat:
Detector options are endless. A general detector can be used for wide band area detection and a second can be for fine tuning.

Makes sense, I have been searching a bit to find a broader type of sensor (but haven’t been able to find anything). This general sensor you mention, do you know where I can find it?

Greensprings:
i saw a study somewhere that said that tracking the sun path mathematicly based on the known solar geometries is better than scanning for the bright spot. scanning is time and power consuming. The report said that the update frame rate was once per hour sufficient for maximum return in power invested in scan and position feedback

Ok, that’s very interesting I haven’t seen that. Do you remember where you saw this study or what it was called? Maybe I could use it as a reference in my work. And I would imagine that the conclusion is probably true for most places on the planet, but surely its effectiveness must vary a lot from situation to situation.

TomGeorge:
Hi
I hate to ask this what experience have you in electronics programming and hardware.

Hi! Very little! But I really want to learn and that is a big reason for why I want to do this. It may seem strange that my thesis involves things that are so new to me, but the thesis will be based more on theoretical models and simulations, this Arduino project is more of a secondary objective I have chosen because I really want to learn it and to verify or disprove the theoretical part.

TomGeorge:
This worries me, one of your methods of tracking is to look at different parts of the spectrum from the sun to try and calculate solar position, you use filters to isolate the various parts of the spectrum with your sensors.

I was under the impression that different models of photoresistors are sensitive to different parts of the spectrum, like this:

By using a combination of models it would then be possible to isolate different measurements from each other and weigh them based on their ability to be used for electricity production. Is this totally different from the filter method you are referring to?

Chagrin:
You might consider using a webcam and diffraction grating / prism to capture a picture of the spectrum from which you can then use to determine the optimum position. As it appears that electronics is not your forte then you might find it easier shifting all the difficult stuff into a program run on a proper computer. See: http://publiclab.org/wiki/spectrometer

I looked at your link, and if I would be able to generate a detailed chart of the wavelength distribution like that it would be very good! However: Did I understand correctly that the method described involves shining a laser through a sample such as an oil? Would that be possible when the sample that I want to investigate is the air itself? And if a webcam is used, are they able to pick up on wavelengths outside of the visual field? I’m thinking that they might not be designed for that since people only use them for capturing visible light.

Hi, what is the University course you are doing?
It can't be engineering.
The link you showed only indicates where the relevant frequencies peak for each material, you will need better and tighter responses than just what the sensor response can provide.
Most IR systems have an IR filter on them because the sensor is not totally blind to eveything except IR.
Other sensors are the same.
The only decent system is the twin sensor system, using a small pair of PV's to make sure you are peaking for the right energy is a very good idea.
Tom......

TomGeorge:
The link you showed only indicates where the relevant frequencies peak for each material, you will need better and tighter responses than just what the sensor response can provide.

Yes exactly it’s the peak for each material, but Im assuming photoresistors made out of those materials exist? The image was just an example but its from a page about different photoresistors and had the figure text:

The figure shown here represents the spectral response of photoconductive detectors made of different materials, with the operating temperature expressed in K and written in the parentheses.

And yes there is an interval in which the sensor detects, and in a perfectly accurate situation it would only pick up one wavelength, but then I would also need about a thousand different types of sensors to detect all the incoming light which would be unrealistic. Having a broader range per sensor could be a pretty good idea right? Then by using maybe 4-5 different types of sensors it would still give a general idea of the wavelength types that are predominant.

What you should do is measure the radiation spectrum from various directions. Here is how to do it very cheaply using just one sensor, an IR-sensitive USB camera: http://publiclab.org/wiki/spectrometer

You did not really answer my question about whether it is economically practical to use PV panels "where the sun doesn't shine", but I strongly suspect that it is not. You would be economically better off heating water, with much cheaper collectors.

Given your questionable criteria, it seems that it might be more reasonable to look at the output of the photo-voltaics at various orientations, and start your serch sweep based upon those results. once oriented towards the source giving the greatest output, use the same logic for tracking. photo-cells rather than LDRs for sensors.