LED based sun simulator

Hello Everyone,
Please take the time to go through this rather long post. We really need your help.
My friends and I have decided to do a mini-project ‘LED sun simulator’ as part of our undergraduate studies. Even if you don't know what that is, I think you can help with some parts. The requirements that we have imposed are:
· We’ll try to use circuits instead of ICs as much as possible.
· We’re trying to keep the cost of the circuitry and LEDs around(preferably much less than) 500$ (or 25000 INR).
· Only the visible light spectrum is being simulated. High-Brightness (High Power?) LEDs with Six to eight wavelengths are used for this purpose. We are thinking of using Luxeon Star LEDs that come in typical wavelengths 447.5 nm, 470 nm, 505 nm, 530 nm, 590 nm, 617 nm, 627 nm and 655 nm.
· Class C specifications by ASTM E927-05 standard (Please see article on Solar Simulator in Wikipedia for details) are expected. AM1.5G spectrum is preferred.
· LEDs will be arranged in a flat panel with colours forming the vertices of a hexagon (for uniform distribution of light, 6 colours). We want to be able to switch on and off LEDs of specific wavelength (Eg: switch on LEDs with wavelength 470 nm) using a laptop and a microcontroller. We intend to send 8-bit data from the computer to the Microcontroller, reserving one bit for a specific LED set of a specific wavelength. The microcontroller would output a PWM waveform that is given to a DC to DC (buck) converter which would output a suitable current to forward bias the LEDs. The current should be constant.
· We intend to use a Rectifier and regulator (or perhaps implement an SMPS) to act as supply to microcontroller and the buck converters.
The block diagram of the system is shown as attachment. I think we'll be able to handle the circuits, but we have some points that we are doubtful of.
Please clear the following doubts we have before we proceed with this project.
· Is the Project possible with the set requirements? Can we really use a microcontroller to receive 8-bit data from a laptop, give a 6 or 8 channel PWM output accordingly? Would you please suggest a suitable microcontroller? Or should a Digital signal processor be used?

• Would this count as a moderately complex electronic system? That's a requirement in the mini-project. If not, how could we make it satisfy the requirement?

· Are LED based sun simulators flashed or continuous? Which ones have better temporal stability? Will cooling pads be enough to provide class C temporal stability?
· Suppose we use about 8 High brightness LEDs of the same color, the distance form similar collared LEDs being 12 mm along the edges of a hexagon with edge length 4 mm. Based on a rough guess, what would be the Irradiance at, say, 300 mm below the LEDs?
· What is the minimum intensity of light required to perform solar cell testing?
· Can we try to get Class B specification on spectral match just by controlling the current through the LEDs, thereby decreasing the radiant intensity? Will this affect the wavelength of the LED output? If so, how significantly?
· Is there a need to output all wavelength lights for an LED Sun Simulator? (Eg: Will the user want random wavelengths like 600nm?). That is, are continuous wavelengths needed?
· How can we check if our simulator meets the standards? What are the testing apparatus required? Would this be affordable? Is there a low cost method to test the accuracy within class C specifications?
· Will we need a feedback from the current supplied at LEDs to check if it is constant? If so, how would we rectify the problem?

Please help us with these doubts. Even if you point us to the correct sources, we’d be grateful.

abhaymv said:

Would this count as a moderately complex electronic system?

Click to expand...

I think you know we can't answer that without a definition of what "moderately complex" means.

abhaymv said:

What is the minimum intensity of light required to perform solar cell testing?

Click to expand...

That of course depends on what you are testing. If I wanted to map out the performance of a panel, I'd want to go from zero to maybe 2X full solar irradiance, since panels can be used with reflectors to increase irradiance on the panel.

abhaymv said:

Can we try to get Class B specification on spectral match just by controlling the current through the LEDs, thereby decreasing the radiant intensity? Will this affect the wavelength of the LED output? If so, how significantly?

Click to expand...

I saw an excellent discussion of this somewhere. I'll see if I can find it. The point is, though, that you need accurate performance (spectral) data on the LED you're using. This will provide spectrum versus current curves, for instance, allowing you to completely simulate the LED in software.

abhaymv said:

Will we need a feedback from the current supplied at LEDs to check if it is constant?

Click to expand...

Feedback is an essential part of a control circuit. So if you want to control current beyond just the PWM parameters, any circuit you use will employ feedback. This is usually done by watching the voltage drop across a low-ohms (like 0.1Ω) shunt resistor in series with the load.

I can't find the the discussion I was reading, but found these similar sources while looking. Maybe they'll interest you.

048360a_PWM_vs_CCR_LED_App_Note.pdf

http://www.lrc.rpi.edu/programs/solidstate/cr_effectsofdimming.asp

http://www.ti.com/lit/an/slyt238/slyt238.pdf

http://streetlightingresearch.org/programs/solidstate/pdf/Gu-SPIE6337-17.pdf

http://vbn.aau.dk/files/39948127/led_spectral_and_power_characteristics_.pdf

http://www.roallivingenergy.com/products/more/spectra_color_mixing_led_driver

Even high power LEDs cannot provide the brightness of sunlight.
Use lasers instead.

That won't stop our government from throwing our grandchildren's money at it.

It seems to be a trendy topic at universities, here and here, even though there are commercial products available, here and here.

Hello,
Thank you for the prompt reply. We were not provided with an exact definition either. However, I think we are to design a circuit that would be moderately complex for Electronics and Communication engineering third year students.
I have the Normalized luminous flux/radiometric power vs Forward current curve in the datasheet of Luxeon Rebel DS68. Does PSpice [Orcad Capture 16.3] already have the library required for this or similar LEDs? And, can any of the programs in the said simulation suite be used to simulate and measure LED output characteristics needed for the sun simulator? When we searched the internet, we found that a spectroradiometer is commonly used for this purpose, but there’s no way we can afford that, and since our college does not have photonics or any other branches related to light applications (We’re doing this project for electronics), they don’t have the said instrument either.
We're thinking of demonstrating how the solar simulator actually works. This is a very small scale implementation. The characteristics of a single solar cell would be enough.
Won’t using the PWM output and the buck converter suffice? Or is the feedback used to decrease the chance of error? Is there a good chance of error?
I’ll definitely check out the references you provided. Thank you!
To Audioguru: We know it cannot provide the brightness of sunlight fully, but I think somewhere near 0.6 sun was implemented. LASERS, unfortunately, are not studied in detail in our course.

I have already seen professional sun simulator by the company Strama-MPS. They are using several coloured high power LED's and a water cooled plate. Very fine equipment! They arrive twice better than class A with this system.
http://www.strama-mps.de/en/product-portfolio/photovoltaic/standards/promosim-evo3.html

Hello again,
I've read in some detail about driving the high power LEDs that are to be used in the solar simulator. I'm at a dilemma.
Should I use a linear regulator to regulate the current supplied to the LEDs, or should I use a switching regulator?
By Linear regulator, I mean using a voltage regulator to provide constant voltage between two terminals (say, an IC 723 voltage regulator) and connect the terminals to a variable resistor or pot. By adjusting thee pot, the current would be adjusted. The high power LEDs used are either 700mA or 350mA.
We intend to drive about four of them in series, and the approximate drop would be about 15 volt(roughly). The 230 V AC supply(as available in India) would be step down to a required level, rectified, regulated using regulator IC 723 and is loaded by the current regulators for LEDs.

I understand this method has some disadvantages, and Switching regulators are used as alternative. But I do not know which of these regulators to use. Please help.
Heating of LEDs won't be a major issue, I think, as this is just a small scale prototype and the maximum time the LEDs would run continuously would be about 10 minutes, and I think the inbuilt heat sinks would handle that. Please correct me if I'm wrong.
The book mentions something about efficiency, but I don't understand. Please help me


Reference book I'm using:
Power Supplies for LED Drivers by Steve Winder

The LM723 voltage regulator is ancient. If you try to limit the current with a series pot then Ohm`s Law shows that the pot must be HUGE and VERY EXPENSIVE.

The datasheet for Luxeon Rebel LEDs shows that all their spec`s are listed when the thermal pad temperature is cooled to 25 degrees C somehow. They are surface-mount style and cannot be bolted to a normal heatsink so they will overheat in a few seconds.
The datasheet shows an odd pcb design full of holes.

Sir,
The basic circuit to provide the current is given in the attachment.

Now, I need 350mA of current. I understand that LM317 cannot provide this. So, I thought I'd go for LM723, with a similar circuit. But is that possible?

As a second step, I'd need to control the current between about 200 mA to 350mA with a reasonably high precision (an error of about 10mA is acceptable) to control the brightness of the light.

Basically, I need to build an adjustable current source, given a constant voltage supply.

Is there a way I can do this using a linear regulator? Which IC would you recommend?

Would high power LEDs overheat even in rated current? How would I cool them? A manufacturer claims:


"The surface mount LED element is pre-soldered to the aluminum star printed circuit board, that simplifies connections and assists with heat sinking. For most applications, the breakout board should be contacted to a larger heat sink to prevent overheating of the LED element."

How would I connect it to a large heat sink? Would that be necessary? Can someone point me to a good source where I can read on this?

abhaymv said:

I need 350mA of current. I understand that LM317 cannot provide this. So, I thought I'd go for LM723, with a similar circuit. But is that possible?

Click to expand...

Look at the datasheet for an LM317. An output current up to 1.5A is guaranteed. It has a metal tab that can be bolted to a heatsink for cooling.
Maybe you were thinking of a little LM317L which has a max output current of 100mA and cannot be cooled.

I need to control the current between about 200mA to 350mA with a reasonably high precision (an error of about 10mA is acceptable) to control the brightness of the light.

Click to expand...

Your vision's response to brightness is logarithmic so 200mA and 350mA will look almost the same. 35mA and 350mA will look a little different. 3.5mA will look dimmed but still visible when compared to 350mA.

A pot with such a wide range of resistance change will be very difficult to adjust. perhaps a switch can be used to switch resistors?

I need to build an adjustable current source, given a constant voltage supply.

Click to expand...

You DO NOT need a constant voltage supply. LEDs set their own voltage. Instead you need an adjustable regulated current.

Is there a way I can do this using a linear regulator? Which IC would you recommend?

Click to expand...

You DO NOT need a voltage regulator (but an LM317 can make a good adjustable current regulator if you can find a suitable pot or use switched resistors).

Would high power LEDs overheat even in rated current? How would I cool them? A manufacturer claims:

"The surface mount LED element is pre-soldered to the aluminum star printed circuit board, that simplifies connections and assists with heat sinking. For most applications, the breakout board should be contacted to a larger heat sink to prevent overheating of the LED element."

Click to expand...

Before you said you have Luxeon Rebels which are tiny. They are only 3mm x 4.5mm.
Now you are talking about Luxeon Rebel Stars that have an aluminum base and must be fastened to a heatsink for cooling.

How would I connect it to a large heat sink? Would that be necessary? Can someone point me to a good source where I can read on this?

Click to expand...

A Luxeon Rebel Star can be bolted to a heatsink with a thin film of thermal grease between or maybe a thermal adhesive can be used.
The Philips Luxeon website probably has an application note about how to cool them.

Sir,
Thank you for the prompt reply. I still have a few doubts on the IC 317. I assume I can use LM317T?

Audioguru said:

Look at the datasheet for an LM317. An output current up to 1.5A is guaranteed. It has a metal tab that can be bolted to a heatsink for cooling.
Maybe you were thinking of a little LM317L which has a max output current of 100mA and cannot be cooled.

Click to expand...

Can I use the same circuit shown in the attachment of my previous post for this purpose? If so, I'd need resistors of very small values and I'd need to adjust them. Switching seems to be the only option there.

Audioguru said:

Your vision's response to brightness is logarithmic so 200mA and 350mA will look almost the same. 35mA and 350mA will look a little different. 3.5mA will look dimmed but still visible when compared to 350mA.

A pot with such a wide range of resistance change will be very difficult to adjust. perhaps a switch can be used to switch resistors?

Click to expand...

Please correct me where I'm wrong:

I'm using the LEDs to implement them on a sun simulator for solar cell testing. So, it's not human vision's response that I'm worried about. I know I cannot get nowhere near the sun's brightness, but we're aiming at matching the relative intensity of the sun spectrum at different times of the day (please see figure attached, which uses LEDs in a wider spectral range )
I understand that I can reduce the normalized luminous flux to a quarter of the desired value by changing thee current from 350 mA to 100 mA. So I think I'd need to adjust resistance between 12.5 Ω to about 3.57 Ω. So using switched resistors would be the cheaper option (I think?)

Audioguru said:

You DO NOT need a constant voltage supply. LEDs set their own voltage. Instead you need an adjustable regulated current.


You DO NOT need a voltage regulator (but an LM317 can make a good adjustable current regulator if you can find a suitable pot or use switched resistors).

Click to expand...

Wouldn't I need a voltage regulator during the conversion of a 230 V, 50 Hz AC to a DC value (say, 12 or 15 V)? Since this is a project for studying electronics, we cannot use DC power supplies available in the market. After we convert AC to DC, it'll be used to drive three separate strings of LEDs with different wavelengths. So, I was thinking I'd need a voltage regulator (as part of the DC supply) following which I'd connect three current regulating circuits using LM 317.
Please correct me if I'm wrong, I don't know much about this topic.

Audioguru said:

Before you said you have Luxeon Rebels which are tiny. They are only 3mm x 4.5mm.
Now you are talking about Luxeon Rebel Stars that have an aluminum base and must be fastened to a heatsink for cooling.


A Luxeon Rebel Star can be bolted to a heatsink with a thin film of thermal grease between or maybe a thermal adhesive can be used.
The Philips Luxeon website probably has an application note about how to cool them.

Click to expand...

Thank you for clearing that up. I'll read on what Philips has to say.

abhaymv said:

Wouldn't I need a voltage regulator during the conversion of a 230 V, 50 Hz AC to a DC value (say, 12 or 15 V)? Since this is a project for studying electronics, we cannot use DC power supplies available in the market. After we convert AC to DC, it'll be used to drive three separate strings of LEDs with different wavelengths. So, I was thinking I'd need a voltage regulator (as part of the DC supply) following which I'd connect three current regulating circuits using LM 317.
Please correct me if I'm wrong, I don't know much about this topic.

Click to expand...

As audioguru said, you need to control current, not voltage. You simply need a suitable transformer, a rectifier, and some filter capacitors. Feed this voltage to the input of a current regulator, and you will get whatever current you need. I wouldn´t bother with smps, since your goal is not efficiency but rather a working design.
Since you need an adjustable current source, look for an opamp-based current source with a pass transistor, these can be simply regulated with voltage, so you don´t need to fuss around with weird-value potentiometers the 317 would need.

Sir,
Thank you for the reply and clearing my doubts on the regulated current source and switched power supply.

kubeek said:

Since you need an adjustable current source, look for an opamp-based current source with a pass transistor, these can be simply regulated with voltage, so you don´t need to fuss around with weird-value potentiometers the 317 would need.

Click to expand...

I don't think I can use opamp based current supply, since opamp is an active device and needs a power supply. Thus which I'll need to construct a dual tracking DC power supply (of +/_ 15 V for 741) which is way more complicated. This is not a problem for voltage regulator, as it does
not need external power to run.
So my doubt still stands in that aspect

Also, I'm driving three separate strings (which would contain about 4 leds in series each), so I guess I'll need to connect three adjustable current supplies in parallel with the rectified and filtered output?

abhaymv said:

Sir, ...

Click to expand...

I am NOT a SIR. A SIR IS KNIGHTED BY THE QUEEN OF ENGLAND.
I never met the Queen so I am just an ordinary guy.

I assume I can use LM317T?

Click to expand...

Yes if it has a big enough heatsink.

Can I use the same circuit shown in the attachment of my previous post for this purpose? If so, I'd need resistors of very small values and I'd need to adjust them. Switching seems to be the only option there.

Click to expand...

Yes.

I'm using the LEDs to implement them on a sun simulator for solar cell testing. So, it's not human vision's response that I'm worried about. I know I cannot get nowhere near the sun's brightness, but we're aiming at matching the relative intensity of the sun spectrum at different times of the day (please see figure attached, which uses LEDs in a wider spectral range )
I understand that I can reduce the normalized luminous flux to a quarter of the desired value by changing thee current from 350 mA to 100 mA. So I think I'd need to adjust resistance between 12.5 Ω to about 3.57 Ω. So using switched resistors would be the cheaper option (I think?)

Click to expand...

I think so.

Wouldn't I need a voltage regulator during the conversion of a 230 V, 50 Hz AC to a DC value (say, 12 or 15 V)? Since this is a project for studying electronics, we cannot use DC power supplies available in the market. After we convert AC to DC, it'll be used to drive three separate strings of LEDs with different wavelengths. So, I was thinking I'd need a voltage regulator (as part of the DC supply) following which I'd connect three current regulating circuits using LM 317.
Please correct me if I'm wrong, I don't know much about this topic.

Click to expand...

A simple cheap AC-DC power supply has a transformer, a bridge rectifier and a big filter capacitor. It is inexpensive so we buy them. If you make one then it costs more. Its DC output voltage varies a little when the load current changes because its voltage is not regulated but in your circuit voltage regulation is not needed.

GOOD GRIEF!

Why are foreign students given a project like this without being taught NOTHING about it by the teacher? Maybe the teacher is lazy?

First the students should be taught the basics (Ohm's Law, etc) of electronics.
Then they should be taught about transistors, opamps and voltage regulators.
Then they should be taught about modifying circuits like making an LM317 a current regulator instead of as a voltage regulator.

Thank you for the replies. I apologize if my questions were stupid. Actually, we were taught about the basics, opamps, voltage regulators etc. We've done most of the basic opamp circuits as practical too. Our academics covers regulators 317 and 723. I am a third year student of engineering, but as you can see, not a very good one in practical matters.

But when we turn to the practical aspect, I'm afraid only voltage regulators were discussed, and we hadn't had an opportunity to implement them using the said ICs.
I think the problem lies in the fact that LEDs are not given sufficient importance in the curriculum.
Using LM317 as a current regulator, I'm afraid, was never discussed. But we were referred to the textbook "Power Supplies for LED Drivers" by Steve Winder to know about it.
Even if we've studied and implemented circuits like rectifier and filter, regulators etc., interconnecting these circuits into a working system was never done. That's where I think I had my doubts, because the LEDs are high power ones and one must be careful in its design.

B.T.W, our teacher just suggested the topic, we chose it because it would be a good application of what we've studied in Linear Integrated Circuits and electronics in general.

Since I think I have the basic outline now, I'll simulate it in Pspice. I hope you guys can help me if I get stuck again. Sorry, but since this project is one that requires precision, I might get stuck anywhere

P.S: I used Sir in the more general sense, I understand it's often used in a teacher-student conversation by the student? That's the case in my country anyway. Sorry if it offended you

Nice and informative thread.

abhaymv said:

Thank you for the replies. I apologize if my questions were stupid. Actually, we were taught about the basics, opamps, voltage regulators etc. We've done most of the basic opamp circuits as practical too. Our academics covers regulators 317 and 723.
I think the problem lies in the fact that LEDs are not given sufficient importance in the curriculum.
Using LM317 as a current regulator, I'm afraid, was never discussed.(

Click to expand...

I am a retired engineer. An old geezer.

You are lucky that you were taught about modern things.
LEDs, opamps and voltage regulator ICs were not invented yet when I studied electronics in university. I learned about them myself BY USING THEM. You should use things yourself if they are not used in school.

I know this would be an extremely stupid question, but I need some diodes to build the rectifier. Since I'm drawing about 3 X 350 mA of current, I believe I should get diodes with rated current more than double of what I require. So, to be on the safe side, I'm using a 4A or 5A diode, and the voltage level can be 30 V. Frequency would be 50 Hz.
From what I understand, I can use a 1N1200 for this. But I understand this overqualifies the requirement. Are there any popular power rectifier diodes better suited for this application?


Also, is there a need for an EMI filter before the rectifier if I'm using the constant current source by LM317?