Powerful Aquarium Lighting
Posted 17 January 2012 - 11:02 AM
I am working on a few things at the moment all for my aquarium project, however recently whilst building my relay enclosure I started thinking about how I want to power it all. The relay enclosure has a couple of pc fans that I am going to use to aid my heat sinks in thermal management. To be honest like usual I may have gone a little over kill on the thermal management, well on everything but thats my style .
Any way the fans require a 12 volt power supply, the majority of the project runs off the netduinos 3.3 and 5v, a little of the mains that my relays will provide and final I think.... What about the LED lighting?
I was planning on feeding the netduino a 12v supply and then splitting this to feed the LEDs and Fans, however my LED enlosure is starting to look quite powerful all of a sudden...
Despite some opinions I will be controlling my LEDs from my 2 tlc5940s, this provides me with 32 channels of PWM. Each of the LEDs in my aquarium lighting unit are going to be individually controlable (again yes you may be thinking THIS IS COMPLETE OVERKILL but there is method to my maddness...).
As part of the LED lighting I am planning on writing my application so that it retrieves real Lunar cycles from well where ever I want, perhaps where the fish are from to provide times on when to dim and the intensity of moon light (produced from blue LEDS still need to research the wavelengths required). Any way, I plan on trying to simulate a natural environment as much as possible, effects include cloud cover (created by dimming different areas of lighting unit and having this gradually move across the tank), also lightening (a little strobing from single leds again occuring randomly across the lighting unit) and sunrise/sunseet and lunar patterns to affect blue light intensity.
So my questions....
I am going to use powerful LEDs I was thinking either a mixture of 3 watt LEDs and 10 watt LEDs or just a load of 3 watt LEDs.
My first idea was something like:
5x Warm White 10W
5x Blue 10W
5x Red 3W
5x Green 3W
10x Cool White 3W
Or my other idea using just the 3w leds above
14x Cool white 3w
8x Blue 3w
4x red 3w
4x green 3w
I am total open to suggestions, the aim is to do the effects mentioned above, run the most efficient circuit possible and keep costs down. Also thermal management is quite crucial in my book (any UK heat sink suggestions would be much appreciated also?).
What I want to know is can you suggest a driver circuit that I can PWM control the LEDs, I am already thinking each LED is going to have have it's own driver, but I will just make a driver board with a repetition of the circuits(s).
So circuit wise I would like protection for the LEDs as they are current sensitive, thermal protection also and high efficiency, and I would like to keep it as cheap as possible.
I have read this instructable, but I simply just don't know what the best option would be, remember I will be providing PWM from the TLC5940s. And to go back to what brought my thinking of this part of the project forward... What would your recommend power supply wise? Can I get away with 12v for the fans and the LEDs or, will I be looking more 24v and however many amps for the LEDs?
Many thanks in advance for your help,
Posted 17 January 2012 - 12:52 PM
Your project is simply amazing, and I'd keep it as quite hard to manage. Anyway, I'm thinking to the simulation, tons of parameters...but maybe should be funny as well.
Well, first off, I'm wondering whether the fishes could "appreciate" a pwm lightning. I mean that the humans don't feel any vibration of intensity, but is it the same for the fishes? BTW, the dogs are hearing ultrasounds, and we can't...
That should be a question to ask to some biologist, or else...
They're beautiful, but I can say that they're also very powerful. If I don't understand bad, you're going to light the tank with tens of Watts.
Well, I own a couple of hi-power leds (20W and 5W). By powering the 20W led with as little as 1-2W, you really can't keep your eyes on it.
Here's an article I've written a month ago (sorry is Italian), but there's a (raw) video. That's the 20W led, dimmed using the PWM output of the Netduino.
All that to say that probably you won't need a so high power. Instead, many sparse lo-power leds, could be better.
Hope it helps...at least for some of your problem.
Posted 17 January 2012 - 11:47 PM
Thanks! I like the way you get to the point, on reflection and a little more research I have discovered, although the 10, 20, 50, 100 watt led are available and sold for fish tanks, although not useless these higher power ones would be better for very deep tanks, where the par is reduced due to added discipation that occurs as the water is made deeper. See here
Here is some overall info from wikipedia about lighting in aquaria:
The most recent addition to the list of aquarium lighting technologies is LED lighting. These have the potential to be much more efficient than any other technology, but are not fully developed. LEDs have the advantage of point source lighting, but are also adjustable to most power levels. This allows for more advanced lighting schedules, the simulation of cloud cover, or even lightning storms. So far, LEDs have found use mainly as lunar lighting in commercial products.
Reef-keeping enthusiasts have began to build their own LED light fixtures as well. Debate over their effectiveness towards coral is still inconclusive, particularly with respect to their ability to give off UV radiation, critical to obtaining a vibrant array of colors that most people interested in LED lighting are looking for.
LED lighting can be considered one of the most energy efficient and low impact options to lighting a reef tank as well, with a projected life expectancy of seven years.
 Overall lighting considerations
When considering lighting for an aquarium, there are generally two factors to consider: wattage and color temperature. Depending on the type of lighting (i.e. fluorescents, metal halide, etc.) the wattage of light emitted may vary considerably, from tens of watts to several hundred watts in a lighting system. Wattage, while not indicative of color, is equivalent to power and essentially determines how brightly the light will shine. Due to the scattering of light in water, the deeper one's tank is, the more powerful the lighting required. Color temperature, measured in kelvins (albeit slightly unrepresentively) refers to the color of light being emitted by the lamp and is based on the concept of blackbody radiation. Light from the sun has a color temperature of approximately 5900 K and lighting systems with color temperatures >5000 K tend to be best for growing plants in both the marine and freshwater setting. 10,000 K light appears bluish-white and emphasizes coloration in fishes and corals. Higher up on the spectrum there are 14,000 K and 20,000 K bulbs that produce a deep blue tint which mimic the lighting conditions underseas, creating an optimal ambience for invertebrates and livestock present.
Along with this information from http://jimsexoticfish.com
For live plant customers, Eco Exotic has come out with 8000k led aquarium lighting to use in addition to their regular 10000k and colored stunner strips, so you can customize your lighting to your specific needs.
It would appear that, other than providing enough light to allow plants and corals to grow, the color temperature is somewhat your own choice. That being said sunlight it around 5500k and so this should be the minimum with the tank depth taken in to consideration.
If anyone can elaborate or supply further info on colour temp/ intensity for aquariums I would appreciate it.
So back to the project...
I now plan on getting a 750mm x 250mm x 4mm thick aluminium sheet to act as my heat sink, do you think this will be ok heat disapation wise for 3 watt leds spaced 3 inches (80mm) apart?
Next I have pretty much decided to go with 3 watt leds, I have seen a few other projects people have done and alot of people are using the cree 3 watt leds. How do the ebay ones I listed in my first post compare with the cree ones?
The other projects don't seem to have any dimming or colour control really, well not netduino powered any way, that I have stumbled across any way! So an based on the above about colour temp and intensity and the actual colour of the light in general, I want to make this controlable to some extent also so it can be adapted to different situations. Would I be right in saying that colour temp changes with intensity (dimming level) and is there any set rule e.g. a light with a colour temp pf 20k @ 50% pwm cycle would be 10k or would it just be a dimmer 20k?
So the plan numbers wise is 30 LEDs in total (3x10)
----- W ---------- W ---------- B ---------- W ---------- B---------- W ---------- B ---------- W ---------- B ---------- W -----
----- B ---------- G ---------- R ---------- W ---------- G ---------- R ---------- W ---------- R ---------- G ---------- B -----
----- W ---------- B ---------- W ---------- B ---------- W ---------- B ---------- W ---------- B ---------- W ---------- W -----
13 x White
10 x Blue
3 x Green
Total 90 Watts.
To control the leds independantly from the TLC5940s, I guess each will have its own driver, this will be made in to one board.
For the suggested LEDs how do I calculate what power supply I would need and how do I provide the power to the LEDs by using the TLC5940s to supply the pulse width?
I guess a transistor will be used, a resistor and a capacitor maybe for each LED? I am happy to go with an IC for each one, I believe you can use a 555, there were a lot of circuit options on the in instructable link, but I am not sure the best route or if there is a better option?
I quite like the idea of the buck and boost circuits, I think its good to allow for changes in current, yet still supply the LEDs with what they need, also it would be great if the circuit could be tuned to suit the LEDs and be as efficient as possible.
Many thanks again,
EDIT: This link shows someones study on par and depth in an aquarium.
Posted 19 January 2012 - 12:18 AM
This is a particularly good one
I work for Microsoft. Opinions expressed here are my own and do not necessarily reflect those of my employer,our partners or customers.
Posted 19 January 2012 - 04:22 AM
Posted 19 January 2012 - 11:21 PM
Posted 20 January 2012 - 05:16 AM
I guess it should pretty hard trying to drive the high-power led with the TLC.
BTW, I like most of your ideas, but this chip could be more feasible for your purpose, and much simpler to interface.
Please, have a look at page 5, pict 2: that should be the best pattern that you can use for the leds. However, the 78L05 isn't required because you can get the +5V directly from the Netduino.
Since one led requires about 350mA, with a voltage drop of about 4V, you can place up to three leds in series, using a supply of 12V.
At this point the only problem to solve is how to limit the current flowing thru, in a simple yet reliable way.
A resistor isn't surely the best solution.
Let me think about it...
Posted 20 January 2012 - 10:37 AM
Waterproof LED strip?
It's just the reason for suggesting the HL1606: most led strips use that chip.
The problem seems related on the led power, though.
Posted 20 January 2012 - 07:35 PM
Thanks for the suggestion, CW2 the LED strips aren't near powerful enough nor controlable enough for my application in terms of having the cloud and lightening effects sadly, plus they cost a ton and the LEDs are pretty poor.
I wondered if the following may be an option...
The LEDs will need 85.425W, 7.5 Amps, 12v supply
This takes in to account no other components such as resistors etc.
Each LED would have its own driver as below, but based on these ratings can you suggest the compents for r1, r2, q1, q2?
What I was wondering is if the following would work and how I could improve it to be more efficient? Or even add in some current switching etc? Maybe some fets/voltage regulators?
Posted 21 January 2012 - 12:47 PM
Thanks in advance,
Posted 21 January 2012 - 07:03 PM
The resistor connected to CS port is responsable for setting the current over the leds. It is calculated as shown in the application information section as
R=0.25/(Iled+0.5*Iripple) ; Iripple is the peak to peak current on the inductor
You also have to calculate the inductance, Rosc and the capacitor shown in the app notes.
This applies to the buck topology that you would replicate to control leds individually, but I'm not sure if I missed something...Do you plan on controlling all 30 leds individually on a PWM fashion using a single netduino?
Posted 22 January 2012 - 05:26 AM
- mcinnes01 likes this
Posted 22 January 2012 - 09:45 PM
I am going to control the LEDs from 1 netduino using 2 TLC5940s, Nevyn has done some fantastic work with these PWM drivers and they make controlling many LEDs possible. Unfortunatly they aren't really designed for high power LEDs and there is probably a better IC for what I am trying to do, but I quite like the idea of using the TLCs and creating individual drivers for each LED. Thanks for the formulea my electronics knowledge is in its infantcy currently so please bare with me....
As an example my LEDs are each rated with a 750ma current drop, with a voltage drop of 3.5v - 4.0v, I am not sure what this range is for on the voltage drop?
Would this be the right calc for R2: 0.5/0.75 = 0.666ohm so 0.68ohm
Then the power rating would be 0.25/0.68 * 2 = 0.735W so 1Watt like these?
Could R1 be 100k ohm?
In terms of Q1 would this be right?
For Q2 Power = 4v * 0.75a = 3W
In terms of Q2 would this be right?
In terms of the open drain with the mosfets can you show me how I could include a PNP in the circuit? Also do you think the circuit would be better with a diode in like these?
Here is another circuit example I have found that contains a diode, could it be adapted to take in to account your advice about the open drain mostfet issue, and do you think this is a better circuit than the one above? With the diode in R1 would be 22k ohm right?
Thanks again for all your help,
Posted 23 January 2012 - 05:12 AM
The PWM is generated by the TLC.
That's okay, but its outputs are open-drain, thus there's no current coming from: instead, only a sink toward ground.
That is, you must consider the TLC output as a switch to the ground: if closed, the output is shorted to the ground (*), if open, no current at all.
(*) It also provides a constant-current driver, but you won't use it.
Your circuit needs a positive voltage to polarize the mosfet, thus it needs another stage acting as "inverter". Let's say: when the TLC's out is ON (closed to ground), then the mosfet gate must have a positive voltage; when the out is OFF (open), the mosfet must be unpolarized.
You can do it by using any PNP (e.g. BC327):
- emitter, tied to +12V, through a resistor (can be some KOhm);
- base, to the TLC out;
- collector, to the mosfet gate.
R2 isn't useful, and it's replaced by the resistor on the PNP emitter.
The current limitation for the led is important, but the trick shown is a bit rough, and not so effective.
You must bear in mind that it's a PWM driver, so the circuit+led are either on or off. Along this consideration, it's a real waste using a resistor to limit the current:
- it's 750mA, thus the power wasted will be relevant (compared to the driven one);
- By placing a resistor on the source of the mosfet will prevent its saturation, thus the voltage drop drain-source (VDS), won't be close to zero. More power wasted.
- By "wasting" power on resistors and mosfets, you must choose bigger components, and ensure some cooling way.
- The Q2 base-emitter voltage drop isn't much precise, thus it's pretty hard to be sure to ensure a reliable current limitation.
- The PWM is *already* a regulation system: why don't take advantage of it?
Then, consider to power each group by using a 7812 regulator, you must supply the +12 in some way, and it provides a good current limitation inherently. The limitation is 1 A, but since I have some power-led, I don't think they're sucking so much when they're powered as explained.
I'd make some experiment with leds and a supply, on my table first.
The circuit can be refined a lot...after.
Bear in mind that you're going to play with several Amperes of pulsing current (PWM).
That means *at least* the follows:
- the circuit/supply must provide enough capacitors to filter the pulsing energy;
- any wire will become a very good antenna!
Tell me what you think about.
Posted 23 January 2012 - 06:36 AM
Mario is right, your circuit is not very realistic.
I think its a current limiter operating in the linear region of the mosfet, and the whole idea of pwm is to operate in the on/off regions, to save heat and power. I think what is needed is a mosfet driver. discrete mosfet driver
I know it sounds good to use discrete components, but for 16 or 30 channels it would be too much, too much soldering, and too much effort and to hard for someone fairly new to electronics.
I would like to give you some of my suggestions too, as I think there are ways your project could be done and it sounds like you really are keen, which is good.
Would we be able to go back to square one? I feel this is the only way I or anyone could make useful suggestions.
How big is the aquarium?
how much light does it need?
how many channels of light does it need? minimum and desired.
what colours are they?
what other functions are needed out of the netduino?
any budgetory constraints?
how is your electronics expertise?
how is your programming expertise?
could you construct, solder the circuits?
do you have many tools, an oscilloscope perhaps?
ps. For the record I have made a 3 channel hbled dimmer out of an attiny and discrete components. but it took a while to get right.
It is still operational.
ps. you need to be careful with the leds, when you are in development it is easy to blow them. very expensive. use fuses.
Posted 23 January 2012 - 12:35 PM
Posted 23 January 2012 - 09:11 PM
Thanks again for the suggestions, Magpie to answer your questions and back to square one is good with me ...
How big is the aquarium?
My current aquarium is 3 feet long X 1 foot wide and 1.5 feet deep
However this is going to be made on heat sinks in 4 sections to power the new tank I will be getting which is 5 foot x 2 foot x 2 foot
So in my current aquarium the 4 sections will be right next to each other, but in my new aquarium they will be spaced evenly across the tank
how much light does it need?
Some research shows that 2Watts of light per gallon is recommended.
Based on my current aquarium that is:
36 * 12 * 18 = 7776 cubic inches
7776 / 231 = 33.7 gallons round to 34 gallons
34 * 2W = 68 Watts round 70 Watts
As I am building this for my new aquarium it will need
60 * 24 * 23 = 34560 cubic inches
34560 / 231 = 149.6 gallons round to 150 gallons
150 * 2W = 300 W
This is based on T5 lighting I'm not sure if there is T5 Watt to LED Watt conversion table?
My proposed total wattage is 30 * 3 = 90 Watts, my idea is to run the lighting at a lower intensity in the meantime until I get a new tank.
How many channels of light does it need? Minimum and desired.
As I want to provide effects I need independent control so ideally 29 channels, however I could group the blues, group the greens, group the reds and leave the whites separate. So this would be 16 channels minimum.
What colours are they?
Royal Blue x 10
Cold White 7000k x 14
Red x 3
The red and greens are just to allow me to adjust the colour to get the best colour from my tank
what other functions are needed out of the netduino?
I am building an aquarium controller, I already have a lot of work done this will take PH measurements, temperature measurement (1-wire DS18B20s on the relays, LEDs and water so 3x sensors), I am building a relay enclosure which has 6 relays, I will have PWM controlled fans on the LEDs and the relays, an LCD display, Wi-Fi, SD data logging, a web interface, I also have 16 outs and 16 ins provided by some shift registers, RTC for date and time. Here is a link to my blog and here is a thread regarding the control board I have nearly finished and this is the relay enclosure so far.
Any budgetary constraints?
Not massive constraints I don't mind building over some time, but not a massive budget like the link to another project further up cost £750 and I don't really want to spend that much.
How is your electronics expertise?
I am relatively new, a quick learner but I certainly couldn't just ad-lib a circuit off the top of my head but I am good at following advice and instructions and my control board isn't too bad, everything is working so far just need to finish soldering the TLC's which is one reason I wanted to use the TLCs.
How is your programming expertise?
I am OK with programming no genius, but OK, I'm better with windows form applications in vb.net, but I not doing too bad with the netduino and c#. Not great and turning datasheets in to drivers but I don't really know where to begin, I am currently coding some drivers for the serial mux 74hc4052 that I use to switch between LCD, ph. and 1-wire bus.
Could you construct, solder the circuits?
I'm not bad at soldering your opinion is more than welcome there are some pics on the relay enclose link and I will add a couple more pics on my blog shortly.
Do you have many tools, an oscilloscope perhaps?
I don't have a scope, got a digital solder station may be able to get a scope maybe, got a good multi meter.
Mario can you sketch me your recommendation of a driver just by hand if you don't mind with some suggested components, I would be very grateful?
Magpie your ATTINY idea is interesting do you have a schematic or circuit diagram (I'm better with diagrams but am just about getting schematics now)? Mario you also seen keen on this technique and as you know from the other parts of my project I have taken on many suggestions so I am definitely open to other possibilities, but I have built my protoboard with TLCs for the PWM so it would be sad to lose these .
Also if I did want to drive the LEDs separately, and taking in to account minimising power wastage would I be correct in saying a 4v power supply @ 7.5amp 90watt would require less or no resistors and therefore will create less wastage? Then I could use transistors to limit the current?
In terms of PWM does it make a lot of electromagnetic noise and can this be reduced using RF filters on the output lines from the PWM source? The whole driver unit is going to go in a sealed enclosure and it will have a cable going to the LED enclosure which will also have its electronics sealed.
Finally thank you for your help I really appreciate it and it’s nice to find people who have the same enthusiasm for electronics and the patience for beginners!
EDIT: Here is a picture of my control enclosure so far...
Posted 24 January 2012 - 05:32 AM
The PWM is definitely the solution to drive the leds. Now, we should establish *how* the PWM plays this game...
When you say "I'll take a 4V power supply to avoid power wasting", it means that's a switching-mode supply (i.e. PWM like). If not, it will waste power.
BUT...you are going to create your own PWM driver (e.g. TLC)...so, *two* kinds of PWM??? Is nonsense...
Indeed, the ATTiny idea is awesome. Magpie, please, get us some details...
I'd cut the TLC off the circuit, even the board mount it.
The reason is you wish two functions for each led:
- intensity regulation (via PWM)
- current limitation (safe-guard)
Thus your driving circuit should be smart, yet simple at same time. The idea of using a tiny MCU gives you the opportunity to build the "intelligence" to regulate both the intensity and limiting the current. More: you can regulate the current, and the driver will be perfect.
I'd imagine to create X small boards, where X is the number of the led to be driven.
Each board can mount the led directly (which seems easy to fix), and should have just four wires:
- +V (led)
- Ground (led)
- serial data (e.g. UART)
- Ground (data)
All the "data" signal are wired together, to the Netduino controller.
Each small board must be identified differently, so that the Netduino can address the right led.
At that point your C# app can set the led #3 intensity as:
It's just a fictional idea about your final approach.
So, by using the ATTiny, you'll have:
- a very compact circuit;
- high flexibility of operations, because you can program the MCU by yourself;
- easiness of wiring: just power and signal lines, nothing else
- scalability: you can use just one led, or dozen
- purpose: you can change the board a little and drive the fan in a similar way, for instance
- cost effective, because I don't think the parts will be more expensive than discrete parts
Think about it.
0 user(s) are reading this topic
0 members, 0 guests, 0 anonymous users