Mini-Aquarium Lighting - electronics question

[Magic Matt]

Hi everyone!
I'm having some trouble working out what I need to be doing with LEDs - been away from using them too long I think! I've done some researching and planning, but I need somebody smarter than me to check what I'm proposing makes sense. Could you please help?


Brief background

I want to light my aquarium, which has no standard lighting or fixtures. There are two small recesses either end, 50mm wide, that I think would each happily occupy a 48mm diam x 20mm tall heat sink. I want light over most of the aquarium, so I figure one light each end.

I have a spare 15V 800mA power supply I would like to use.

I calculate the lighting needed overall to be around 300 to 600 lumens (very approximate, as it varies depending on how planted the aquarium is), which means andwhere from 150-300 lumens each end.


Proposed solution

One 48mm diam x 20mm heat sink mounted in the recess each end, with LEDs on stars glued to the heat sink. To protect the units from the water, a clear perspex disc will be glued over the front of the recesses (leaving them open backed for cabling and airflow).

I plan to mount wither one or two emitters on stars to the heat sinks (depending on answers to the questions below!) and wire in series.

I have found some "Cree XR-E Q5 Emitter on Premium Star (228LM at 1A)" which I think would be ideal. Rather than run them flat out, I think I'd rather run them cooler, and at maybe 500mA, so they last longer. I think that gives me around 130lm per emitter on the chart, which if I ran them in pairs per heat sink would give me 260lm per side - seems ideal!


Questions

Does it make sense to run them in pairs as I'm suggesting (2 per unit, 4 in total) rather than running them at higher current and 1 per unit (2 in total) ? Am I right to believe that will benefit me in terms of longevity of the LEDs, or is it negated by the fact they're sharing the same heat sink?

At 500mA I think that gives me Vf of 3.4V? So 4 x 3.4V = 13.6V - would it makes more sense to run them direct drive with a single resistor (I thought 3.3ohm 5watt), rather than rig up a constant current supply with an LM350?

I also considered adding a power transistor to the circuit at a later date, to build PWM (driven from a PIC). Can I again just run the output of the transistor through a resistor then out to the LEDs, and use a signal from the PIC to the transistor to switch the lighting on/off at speed?

The heat sink says 9.5C/W - does this mean that with about 3.5W of LEDs attached it will increase in temp by 33.25C or am I misunderstanding how they work? That seems to me to mean the heat sinks will get up to about 55C - is that ok, or should I be trying to add active cooling?


Many thanks for any help.

 

[AnAppleSnail]

Hi everyone!
I'm having some trouble working out what I need to be doing with LEDs - been away from using them too long I think! I've done some researching and planning, but I need somebody smarter than me to check what I'm proposing makes sense. Could you please help?


Brief background

I want to light my aquarium, which has no standard lighting or fixtures. There are two small recesses either end, 50mm wide, that I think would each happily occupy a 48mm diam x 20mm tall heat sink. I want light over most of the aquarium, so I figure one light each end.
I have a spare 15V 800mA power supply I would like to use.
I calculate the lighting needed overall to be around 300 to 600 lumens (very approximate, as it varies depending on how planted the aquarium is), which means andwhere from 150-300 lumens each end.
Proposed solution
One 48mm diam x 20mm heat sink mounted in the recess each end, with LEDs on stars glued to the heat sink. To protect the units from the water, a clear perspex disc will be glued over the front of the recesses (leaving them open backed for cabling and airflow).
I plan to mount wither one or two emitters on stars to the heat sinks (depending on answers to the questions below!) and wire in series.
I have found some "Cree XR-E Q5 Emitter on Premium Star (228LM at 1A)" which I think would be ideal. Rather than run them flat out, I think I'd rather run them cooler, and at maybe 500mA, so they last longer. I think that gives me around 130lm per emitter on the chart, which if I ran them in pairs per heat sink would give me 260lm per side - seems ideal!
Questions
Does it make sense to run them in pairs as I'm suggesting (2 per unit, 4 in total) rather than running them at higher current and 1 per unit (2 in total) ? Am I right to believe that will benefit me in terms of longevity of the LEDs, or is it negated by the fact they're sharing the same heat sink?
At 500mA I think that gives me Vf of 3.4V? So 4 x 3.4V = 13.6V - would it makes more sense to run them direct drive with a single resistor (I thought 3.3ohm 5watt), rather than rig up a constant current supply with an LM350?
I also considered adding a power transistor to the circuit at a later date, to build PWM (driven from a PIC). Can I again just run the output of the transistor through a resistor then out to the LEDs, and use a signal from the PIC to the transistor to switch the lighting on/off at speed?
The heat sink says 9.5C/W - does this mean that with about 3.5W of LEDs attached it will increase in temp by 33.25C or am I misunderstanding how they work? That seems to me to mean the heat sinks will get up to about 55C - is that ok, or should I be trying to add active cooling?
Many thanks for any help.

Putting high-power LEDs in a tiny aquarium can cause trouble because they are usually enclosed, heated, and have little airflow. Incidentally, if your heatsink touches the water then you'll relieve many thermal worries, since a metal-to-water interface is quite good at moving heat.

Let's talk electronics, then LED use, then thermal stuff.

1. LEDs are low-voltage, high-current things. If you put LEDs in series, you keep the same current and add the voltage. In parallel, keep the voltage and add current. The best option for long-term, reliable use, is to get a current-controlled power supply. If you just want to hack things together, a wall-wart unregulated power supply will work, but its output will creep over time. You've chosen the XR-E on star, so that makes soldering easy. Consider your wire runs. To put the LEDs in series will require decent wire (14-18 gauge would be aplenty) inside the tank. This wire should be neat and anchored to something so that it doesn't peel your LED stars off.

Note: You have an 800 mA rated supply. Resistors only change voltages in a circuit. So, broadly speaking, you'd need complex electronics to cut this to another value. You can run your LEDs as follows:

"4s1p" 4 LEDs in a row, each getting 800 mA. Total voltage around 13.6v; LED power is 10.88W
"2s2p" 2 sets of 2 LEDs in a row, each getting 400 mA. Voltage about 6.8v; LED power is 5.44W
"1s4p" 4 LEDs in parallel, each getting 200 mA. Total voltage is 3.4V; LED power is 2.72W

I suggest the first configuration. Why? Because if you adjust the resistance (Vtotal - Vresistor - Vled = 0), the voltage for the LEDs drops. If the voltage for the LEDs drops, then the forward current must change. So you can fiddle with your resistor value until you get just the output you want. It's not great efficiency (60% ballpark wall-wart efficiency, cut further by the resistor burning heat), so about 50% electrical efficiency. But it's cheap and doable. With adequate heatsinking, you shouldn't be able to burn your LEDs out with a resistor value above 2 ohms. This resistor will have to dissipate at least 2W.

2. LEDs are nearly waterproof. Salt-water tanks can put gunk on the wires, and the optical-grade silicones used can pick up any voltatile organics, but low-voltage current won't hurt your fishes. Probably. Research this to be sure. The wires and resistors won't take particular harm from the water aside from rusting. Note: One failure mode of wires is to wick water under the insulation and rust the wires internally.

3. Thermal! You're planning to run (13.6v * 0.8A) = 10.9W (or less) of LED in an enclosed space with poor airflow. About 1/3 of this energy becomes light, and 2/3 heat. That heat sink rating is going to be based on heat transfer from the heatsink to air according to some standard - quite possibly for computer case airflow (Small box with many fans). One way to improve heatsinking is a big heatsink. This is expensive. Another is active cooling. This requires pelletier junctions or fans, both of which are fidgety. I think you could run water over your heatsink. Note: Salt water eats aluminum. This may do sad things to your fish. Don't forget that your resistor would burn (15V-13.6V * 0.8A = 1.12W) of power and heat.

You can use a PWM-driving transistor, but I do not know where or if to use a resistor in such a circuit. For more complex projects, there are plenty of DIY electronics things here, and you can buy ready-made LED drivers for almost any use case.

So: Cut something handy to fit. Maybe a metal plate. Attach the heatsinks to it and mark where LEDs will go. I suggest using Arctic Adhesive thermal adhesive (Not compound goo, you want firm attachment) between the stars and the heatsink. Solder wires before you attach to the heatsink.

 

[Magic Matt]

Thanks for such detailed help!

The tank is an unusual design, in that the lid (of sorts) looks more like a miniature vegetable basket than any aquarium lid I've seen. It consists of a kind of plastic mesh - suffice to say it does nothing to stop water evaporating, so I plan to address this seperately.

The heat sinks will be poking out of the top of the lid, so airflow will be whatever air is flowing around the room, and by putting the clear perspex over the front I'm creating what looks like a flashlight front if you were to look from inside the aquarium. This provides two things - protection from water for the electronics, and also diffusion of the lighting (I may use slightly frosted perspex to give more of a flood effect). The heat sinks are open at the top, so this makes routing cables outside of the aquarium very easy (most probably from their location to the back, then along the back). Fortunately the mesh type design also makes this easy as I can use cable ties and make the whole thing removable. I don't want any metal or plastics in contact with the water that are not rated food safe - it's just not worth the risk - hence I don't want to run water over the heat sinks.


Would I be better running 2S1P (2 emitters, 1 per heat sink) @ 1A (obviously different PSU) or 4S1P @ 500mA (4 emitters, 2 per heat sink), in terms of making the LEDs useful life longer ?
4S1P @ 800mA is going to be just too bright for the fish.


The reason I'm considering a PWM modification later, is possibly linking it to a system that is driven by time of day. The electronics / PIC section I have no problem with, it's just interfacing it reliably with the LED string! I did have some RGB MC-E emitters on a different project, but I managed to fry them in a mad moment of miscalculation. :(

 

[AnAppleSnail]

With a design like that, it'd be easier to add a 20mm 'computer heat sink' fan. Just feed it less than 12 volts (Use the fan's internal resistance to calculate the resistor needed) to keep it at a reasonable speed. 7-9v should be nice and quiet.

XR-Es in long-term service at 500-1000mA last for tens of thousands of hours, ASSUMING:
1. Thermal relief is okay
2. No volatile organic compounds heated nearby: This clouds the optical silicone
3. Limited physical abuse. If glass shards sever the bond wires, the LED fails

Go 4s1p, it makes the circuit simpler (And more efficient).

If we assume you're going to use that 15v power supply, then here are the steps you'll have to follow to dial in the resistor. You have done some already.

1. Find XR-E datasheet, Q-5 brightness bin:
107 '+/-7% lumens at 350 mA
Estimate goal of 150 lumens per LED, use the Relative Luminous Flux for 150% flux: 600 mA or so.
Estimate forward voltage at 600 mA: 3.45 '+/-0.2v or so
2. Assume things are average, that all 4 specimens of yours are at spec. The Cree spec distributions are probably a bell curve, so this is an okay approximation since we're not at the bleeding edge of tolerances.
So 3.45*4 = highest Vf we can get: 13.8v
Question: What if you ran the 4s string at 15v? Each LED would see 3.75V. This is just over the expected forward voltage at 1000mA, so even with no resistor you are unlikely to toast your LEDs quickly. (Thermally, you may manage it though).
Q2: How do we turn 15V at 800 mA into 500mA at 13.8v?
A: The LEDs can handle this small change. If the (idealized) LEDs only see 13.8v, then they can only ever pass 500 mA. This means that if we use a resistor to limit the voltage they see, then we limit the current (And therefore the brightness).
Vloop=0
Vsource - Vled - Vresistor = 0
15v - 13.8v - Vresistor = 0
1.2v = Vresistor
V=IR
1.2V = 0.5A * R
2.4ohm = R
Note: This is all in theory, and should be carefully tested (Check some things before you solder it together)

1. Check open-circuit voltage of your "15v" source when it is unplugged. Some will give up to 30V, which tells you: "The darn voltage creeps up when it's throttled!" High quality sources do not creep voltage very much, which is ideal. If the voltage creeps up, start with a bigger resistor.
2. Get your LEDs wired and clamped onto a heatsink. Wire this up with the resistor for a moment. Did anything go blue? Are you blinded? Now try for a whole second. Is anything too hot? Is it bright enough? Way too bright? Now use a multimeter to measure voltage (Across the LEDs from 1 to 4) and current. Is it what you expect? Adjust accordingly.
3. Make it 'permanent' except for the resistor. Do an hour-long test, see how the heat and light are. You can adjust resistor values using parallel resistors. Ex: 3 8-ohm resistors gives 2.66 ohms, each resistor dissipating 1/3 the total heat. Poof, you're done. A throttled "800 mA" power supply cut down to 500 mA (Or whatever). Ask away with questions, I may have skipped steps.

 

[Magic Matt]

Wow, thanks, that makes it all crystal clear! :)

I have some spare 40mm fans that will probably do the job fine. They're 12V, but I know will run off of 5V as I have two of them running from a USB port to cool the hard drive area of my laptop. Seems to me that if I ran the fans as 2S1P configuration that's 7.5V each (ish) and they'd both be happy.

Just got to wait for the bits to arrive to put it all together now - I have everything except heat proof wire and heat sink adhesive - I only had the normal PC CPU stuff.

Thanks again!

 

[qwertyydude]

One other led to consider is the Cree MC-E. With 4 dies you can put them all in series which means a vF of 14.8v. It's so close to the power supplies rated voltage you'd probably be safe direct driving it or you'd only need a very small resistor like .5 ohm or even just put a regular 1 amp diode in series to drop a little voltage. This also makes your lighting very efficient as very little energy will be dissipated by the resistor. With two of them you can run them 2s2p and still have the same resistance although you'd need a higher power power supply or simply a bigger resistor to prevent drawing too much current from the load.

You'll easily get over 1000 lumens or more if you needed to and also if you down regulate them to lower currents and say a 600 lumen (300 each) you'd increase overall efficiency of the system and prolong led life and generate less heat to dissipate rather than pushing a smaller output led to its limits.

 

[Magic Matt]

I'm struggling to understand where the idea it's less heat comes from. Surely I'm using the same overall number of watts, so the heat to dissipate would be the same?

 

[Mr Happy]

To put the LEDs in series will require decent wire (14-18 gauge would be aplenty) inside the tank.

Whoah! 14 gauge wire can handle 20 A easily. With only 500 mA and short wire runs you can use much thinner and more flexible wire. Check wire sizing charts for voltage drop at the drive current and wire length you are considering. If you have a voltage drop under half a volt in your wires you will be doing fine, especially if you are using a series resistor or regulated current supply.

 

[Mr Happy]

I'm struggling to understand where the idea it's less heat comes from. Surely I'm using the same overall number of watts, so the heat to dissipate would be the same?

The less heat comes if you run the LEDs at lower currents. Otherwise watts in equals heat out, to a first approximation. (Some of the input power leaves as illumination, but for a conservative approach to heat sink requirements you can neglect this.)