Hello everyone, this looks like a wonderful forum and I am glad I found it. I would like to build a small LED grow light. I found this previous post on the topic: http://forum.allaboutcircuits.com/showthread.php?t=11687&highlight=led+grow+light

I have built a few decently complex amplifiers and am confident in my skills to actually build the project. My issue is with some of the concepts behind the project. I have the new edition of "Electronics for Dummies" on the way and im sure it will be helpful.

For this project I want to use high powered 1W LEDs. Eventually I plan to try and source parts with the most appropriate wavelengths for plant growth but for now lets assume I will be using a single type of diode; the Luxeon LXHL-MWEC. Spec Sheet: http://www.philipslumileds.com/pdfs/DS23.pdf

The basic specs for this LED are:
Typ. Forward Voltage: 3.42V
Max DC Forward Current: 350mA

Lets say I want to run about 20 of them off of a 12V power supply. If I am understanding thing correctly this is possible by running them is series. Am I correct to think this or do I need a larger power supply?

When I enter these specs into the ledwiz calculator I get a diagram for 7 rows of 3 leds for a total of 21. There is also a warning about the total power dissipated in my resistors. Why is this? It shows all resistors dissipate 7080 mW. Also there are two types used. If I use an input voltage of 18W and 20 LEDs then I dont get a warning and the resistors dissipate 1323 mW. I assume this is a much more sane setup but would like to try understand why this is.

I was reading on insructables about a simple LED driver circuit that powers the exact LED I am thinking about using. Here is the link: http://www.instructables.com/id/Power-LED_s---simplest-light-with-constant-current/

It says that the circuit is more efficient then just resistors and it looks like its possible to drive multiple leds off of the circuit. The diagram on the second page I think shows 3 leds running off of it. If I wanted to use this instead of just resistors would I build one circuit for each row?

Im sure ill have more questions but thats it for now.

Thanks

 

I have one more question. I was looking at a DC LED controller. It has a max input and output of dc voltage of 32 volts. The driver requires the voltage input be at least 2V greater then the forward drop and say that it can drive a max of 6 1W Luxeon LEDs.

The example they gave is 6 1W LEDs with a avg forward voltage of 3.5 volts. The total forward drop for them would be 21V DC and thus they recommend using a 24V DC power supply. They then go on to say that "Parallel stings can be driven directly without any additional circuitry." Does that mean that its possible to wire up multiple strings of 6 LEDs from this one controller?

Here is the spec sheet for the controller:http://www.leddynamics.com/LuxDrive/datasheets/3021-BuckPuck.pdf

 

You can put 3 in series for 12V. 3.6V+3.6V+3.6V=10.8V. The 12VDC needs to be very stable (ie, regulated), and you figure the series resistor via Ohm's Law, 12VDC-10.8VDC= 1.2VDC, 1.2VDC÷360ma=3.3Ω @ 1W, or you can use a linear regulator such as the LM317.

If you have a higher voltage power supply you can put more LEDs in series.

PuckBucks eliminate the need for this math. They can drive LEDs in parallel (I do not recommend it), but the current will be divided among the paths.

Have you read this?

LEDs, 555s, Flashers, and Light Chasers

Bill's Index

 

If you are looking for an inexpensive 24V DC supply, MPJA has one on sale for $10:
http://www.mpja.com/email/11-10-09a.asp?r=%%ref%%&s=4
That would be a good candidate for powering your set-up.

I've ordered many times from MPJA and have always received my orders very quickly, and never had a problem with their items.

If you are using a BuckPuck supply, it will require at least 2v "headroom" from that 24v.
22v / 3.42v = 6.437. Take the integer portion, which leaves 6 LEDs per string.
Six LEDs in series will have a total voltage drop of 20.52v.

I don't recommend powering multiple strings in parallel from one current source, which is in complete agreement with Bill_Marsden's assertion. If an LED in one of the strings happened to open up, the remaining strings would have to share the current; they would be running well over their rated current, and another LED would quickly burn out. The last remaining string would get 300% of it's rated current, and quickly die in a domino effect.

Three BuckPucks rated for 350mA each would power up to 18 of your LEDs.

It would be a reasonably efficient circuit.

 

maybe PWM modulation of multiplexed LED's might cut down on the power requirements (You could alternate which series or rows of LED's are on at one time, but if you do it fast enough all LED's will appear lit,) and also you could use PWM to simulate artificial dusk/dawn.....


My .02

 

BMorse,
PWM works fine for dimming, but not for limiting maximum current through LEDs.

The BuckPucks are actually using a form of medium-frequency PWM via an inductor to limit the current. Without the inductive limiting, the LEDs would be roasted in short order.

 

Wow thanks everyone. You have all really helped me get a firmer understanding of the task at hand. The BuckPucks look like a easy way to efficiently drive the leds. Do they have a large advantage over using 10 cent resistors? The BuckPucks price is not extreme considering the price of the leds but saving cash is always a good thing.

How about the circuit listed in the instructables link? The max power draw is a bit less so I would have to have to have less LEDs attached to each I assume. I guess I would build one circuit for each sting of LED right?

And finally, if I did use the three BuckPucks would I need three power supplies as well?

Thanks!

 

Wow thanks everyone. You have all really helped me get a firmer understanding of the task at hand. The BuckPucks look like a easy way to efficiently drive the leds. Do they have a large advantage over using 10 cent resistors? The BuckPucks price is not extreme considering the price of the leds but saving cash is always a good thing.

The resistors are cheap up front, but you'll pay for the inefficiency over the long run.

6 x 3.42v = 20.52v.
24v - 20.52v = 3.48v.
To get 350mA current across a 3.48v drop would require:
R = E/I = 3.48/350mA = 9.943 Ohms. Let's round it up to 10 Ohms.
3.48v/10 Ohms = 348mA; almost perfect.
Let's check power dissipation.
Since power = voltage x current, dissipation would be 3.48v x 348mA = 1.21 Watts. Total power dissipation for the 3 resistors would be 3.63 Watts.
Power dissipation for the LEDs would be 20.52 x 348mA = 7.141W per string, 21.42W total.

So, 25W total, with 21.42W of it being useful power dissipation; 85.7% efficient. That's really not too bad.

However, the BuckPucks will be around 90% efficient, so instead of 3.63 Watts being burned up in the resistors, you would use 363mW in the BuckPucks, for a total of 21.783 Watts, for 98.3% efficiency.

This doesn't account for the inefficiency in the 24v 4.2A switching power supply; it's just a comparison between using resistors and BuckPucks.

How long does the 3.267W difference in power consumption take to pay off the difference between the BuckPuck and the resistors? You'll have to look at your electric bill and figure it out. 3.267W really isn't much power.

How about the circuit listed in the instructables link? The max power draw is a bit less so I would have to have to have less LEDs attached to each I assume. I guess I would build one circuit for each sting of LED right?

The Instructables project won't really help save any power; it's a simple linear regulator. You might as well just use resistors.

And finally, if I did use the three BuckPucks would I need three power supplies as well?

No, just one power supply that's capable of supplying the necessary current.

The 24v 4.2A supply I provided a link to will be able to power up to 12 parallel strings of 6 LEDs.

 

Over time the puckbucks will pay for themselves. Even with resistors LEDs are cheaper than flourescents though, LEDs are just more efficient.

Hey Wookie, wanna help me design something that does the same thing as a PuckBuck? It can't be that hard, I think I have about half of a design figured out already. I figure a coil, shottky diode, current sense resistor, handfull of small caps and resistors, a MOSFET, two op amps and a 555 would do it.

 

Hey Wookie, wanna help me design something that does the same thing as a PuckBuck? It can't be that hard, I think I have about half of a design figured out already. I figure a coil, Shottky diode, current sense resistor, handfull of small caps and resistors, a MOSFET, two op amps and a 555 would do it.

Well, even though it's directly related - we'll be getting off our OP's topic in a big hurry.

A 555 timer really isn't the best candidate for a switching power supply. There are lots of dedicated switcher IC's out there already which are specifically designed for that purpose, with plenty of example circuits in the datasheets themselves, and many have tools on the manufacturer's websites to custom-tailor the supply to the customer's requirements. Trying to "roll your own" using a bunch of IC's will really wind up being more trouble than it's worth, and not be as efficient nor reliable as a purpose-built switching regulator.

One thing I strongly suspect is that the BuckPuck is a synchronous buck current regulator; instead of a Schottky "flywheel" diode, it uses a synchronously switched MOSFET as an ideal diode for the inductor->load->return path. Without such a scheme, it would be impossible to approach 95% efficiency.

 

Hey Bill and Wookie thanks again. If you guys do end up designing something like the buckpuck please post it.

I am trying to understand how you calculated the amount of LED the power supply can support. I thought it would be related to the total Amps it provided but I cant seem to get the math to work. I am probably doing it incorrectly but here was my thought process:

PSU supplies max 4.2A.
Each LED uses 350mA = 0.35A.
6 LED's per string using a total of 2.1A
Here is where I get lost:
12 stings of 6 LEDs would use 25.2A.
What am I missing?

Also, in the case of this project the efficiency of using resistors I actually very good. I figure this is not always the case which is why people use the controllers. I just kinda got lucky in this instance right?

 

Did you happen to read that article I linked earlier? LEDs are pretty constant in the voltage they drop, there is some variation but it is small once they are turned on. So the forward dropping voltage (Vf) is treated as a constant. Unfortunately, as is true with many semiconductors two identical looking units can have slightly different Vf for each, it messes with our heads and we have to compensate. We use a set number to get us in the ballpark.

Hey Bill and Wookie thanks again. If you guys do end up designing something like the buckpuck please post it.

I am trying to understand how you calculated the amount of LED the power supply can support. I thought it would be related to the total Amps it provided but I cant seem to get the math to work. I am probably doing it incorrectly but here was my thought process:

PSU supplies max 4.2A.
Each LED uses 350mA = 0.35A.
6 LED's per string using a total of 2.1A Nope, each string still uses .35A, the current is shared, the Vf adds. This is why you want something like a 24V power supply. 6 LEDs X 3.6Vf = 21.6V
Here is where I get lost:
12 stings of 6 LEDs would use 25.2A. 12 strings X 0.35A
What am I missing?

Also, in the case of this project the efficiency of using resistors I actually very good. I figure this is not always the case which is why people use the controllers. I just kinda got lucky in this instance right?

Even after you have it build you need to measure the real world number to be sure it is right. Tweaking resistors is easy though.

 

I am trying to understand how you calculated the amount of LED the power supply can support. I thought it would be related to the total Amps it provided but I cant seem to get the math to work. I am probably doing it incorrectly but here was my thought process:

PSU supplies max 4.2A.
Each LED uses 350mA = 0.35A.

This is where you're going wrong.

6 LED's per string using a total of 2.1A
Here is where I get lost:
12 stings of 6 LEDs would use 25.2A.
What am I missing?

Each LED requires 350mA current.
In a series circuit, the exact same amount of current flows through all of the components. Each string of six LEDs will have 350mA flowing through them.

So, since each string takes .35A, and the supply can output 4.2A, 4.2A/0.35A = 12.

Also, in the case of this project the efficiency of using resistors I actually very good. I figure this is not always the case which is why people use the controllers. I just kinda got lucky in this instance right?

One of the reasons you'll get fairly reasonable efficiency is due to the 24v switching supply.

 

Ok thanks for the help. I going to do some more work on the design and im sure will end up with more questions.

On a side note anyone have any good sources for high power leds other then the usual digikey/mouser etc.

So far I have found:
http://www.luxeonstar.com/ (best prices so far)
http://www.futureelectronics.com/en/
http://www.quickar.com/lux.php?session=m1prXB3K

These are the best sources I have found. I might stop by the candlepower fourm and see what that group is up to. I am not going to be using all of one kind of LED for the light. Right now im having a hard time finding lights with the correct nm output to optimize plant growth. This is one of the main advantage of using LEDs.

 

You're usually best off to stick with authorized distributors of a known quality manufacturer.

There are plenty of counterfeit components on the gray market. If you buy a counterfeit part, you're out of luck - and you've just put money in the counterfeiter's pocket.

One thing that hasn't been covered yet is how the Vf of LEDs changes as they grow warm. All LEDs do this. As they get warm, the Vf decreases, which means they get more current. If you're using passive regulation (as in resistors), the resistors will remain the same value, but the Vf of the LEDs will decrease and more current will flow through the resistor, causing the LEDs to heat up even more.

For this reason, you must ensure that your LEDs are mounted to a heat sink that is capable of dissipating the power (heat) that the LEDs will generate. If you do not give them adequate heat sinking, you will have a melt-down.

 

Plants need red and blue wavelengths. Red for flowering and Blue for growing. NASA did some research on this in 2001.

I choose Cree Royal Blue, and Red for mine.. I am still working on the power supply though.

 

I was looking at the royal blues but they said the wavelength was 420?? vs 465 for true blue. Have you tried both? Also what were the results.

 

You two should talk...

http://forum.allaboutcircuits.com/showthread.php?t=32066