DIY LEDs - The write-up

[terahz]

Yes, as I quoted in my post, they think the cost can be reduced by 75%. Of course, I'm assuming that the savings are passed to the customers.

On a side note, this is a US company, not a random Chinese factory that is replicating LEDs, so at least it sounds a bit legit.

[brassmonkyballs]

Originally Posted by terahz

BTW, just wanted to share something I read about yesterday. A company called Bridgelux has managed to make a 135lm/w LED (at 350mA with 2.9Vf) with a different technology that, according to them, "... can deliver a 75% improvement in cost over current approaches. ... The company anticipates the delivery of its first commercially available GaN-on-Silicon products over the course of the next two to three years."

The point here is, an LED with better than CREE XPG specs at 1/4 of its price.

http://bridgelux.com/media-center/pr...tate-lighting/

great news..glad to see LED technology expanding. Wonder what CCTs we'll see on debut...4700k is a bit low.

[TheFishMan65]

If you go back to the beginning there are some pictures of SoundWave's who started the thread. There are not too many pictures in here, but lots of other threads have pictures. This is mostly discussion on theory IMHO.

Click my home page and go to summaries. The beginning of this sumary has some link that will probably have more pitures. Exactly what are you looking for pictures of?

[chris023]

Do these fuses/resistors/blocks seem okay for a parallel build? Digikey was out of resistors, so I just decided to go with everything from mouser. Thanks.

http://www.mouser.com/ProductDetail/...nfrQNvQWCSQ%3d

http://www.mouser.com/ProductDetail/...mikiHj9qMe4%3d

http://www.mouser.com/ProductDetail/...13SVsWOw%3d%3d

[kcress]

chris;

Yes!

Yes!

Yes!

Those are all the ticket.

[chris023]

Sweet... Thanks!

Originally Posted by kcress

chris;

Yes!

Yes!

Yes!

Those are all the ticket.

[S2minute]

OK...so, not meaning to post a dumb question. I would like to drill the screw holes to mount the XP-Gs, XP-Es and XM-Ls. So..looking on verious retail and wholesale sites as well as trying to find posts, i can`t seem to be able to find info on the spacing of the holes. Any help on this?

[brassmonkyballs]

Originally Posted by S2minute

OK...so, not meaning to post a dumb question. I would like to drill the screw holes to mount the XP-Gs, XP-Es and XM-Ls. So..looking on verious retail and wholesale sites as well as trying to find posts, i can`t seem to be able to find info on the spacing of the holes. Any help on this?

Not sure of the XM's but the XP's are 19mm center to center. What I do though is lay out my centerlines for every LED. Then I use an LED as a sort of jig and I have a small spring loaded punch. I line up the LED and use the punch to mark each hole position. I do this for every LED. Then I take it to my drill press and drill them all.

As an aside...I used to tap every hole for 4-40 but someone gave me an idea and this time I used a #41 drill and found that a #4 3/8" sheet metal screw works perfect and screws easily into the aluminum. I used small #4 nylon washers under each head.

[S2minute]

Thanks , all i need then is 1 led to use as a jig lol unless there is a sketch of one somewhere i can use.

[brassmonkyballs]

Originally Posted by S2minute

Thanks , all i need then is 1 led to use as a jig lol unless there is a sketch of one somewhere i can use.

scroll down to the 2nd page in below link

http://media.digikey.com/pdf/Data%20...0Series%20.pdf

[reefmusic]

Originally Posted by brassmonkyballs

As an aside...I used to tap every hole for 4-40 but someone gave me an idea and this time I used a #41 drill and found that a #4 3/8" sheet metal screw works perfect and screws easily into the aluminum. I used small #4 nylon washers under each head.

I use #4 sheet metal screws too.

[TheFishMan65]

I used a 3/32 (easier to find) drill and sheet metal screw.

[EMUReef]

couple questions... thinking about doing a LED build for my NC12.

From what i assume 12leds is enough. I'd be mounting it in the Stock hood. With the fixture being so close to the water, would there be a need for optics?

Combo wise i was thinking 4 XP-G Neutral Whites 8 XR-E Royal Blues.

If i Got 1 Mean Well dimmable driver.. from my understanding if i put all 12 LED's on the driver im can only dim all 12 right?

What are my options for dimming each string individually?

[TheFishMan65]

12 Should be plenty. No optic, but add a splash guard.

Might make 13 depending on current, but better with 12.

Only option for individual dimming that I that hits me fast is another driver.

[TheFishMan65]

Well I am having a bad day. I seem to have lost the summary. I tried quoting it but it looses all the quotes I had accumulated. Does any one know how to get the html text from old posts so I can edit and repost it?

[brassmonkyballs]

I put 6 XPGs and 6 XRE's on a 14g bio. No optics....PAR at surface is 1100, at bottom forward corners 200. Run it down at 3/4. Will be changing out 2 white to RBs. Only option for separate dimming is 2 drivers and the Meanwell ELN- 60-48 is too much. A Meanwell ELN 30- 27 would do the trick though.

[TheFishMan65]

Well I think I recreated it and then some. Please let me know if I missed something.

I have read this thread and several others. I will try and summarize, because there is a lot of reading and a lot of repeated questions. Please realize a lot of this is from memory and my opinions so it could be wrong or now outdated. As this goes on I am posting more and more quotes. I tried to get the original post; this means that you can click the blue greater than symbol and view the original post and then the posts around it if you need additional information.

And thank you to all of the people (too numerous to count) that have supplied this information. I have added some information from notes that CJO took (and he updated the info – Thanks). I am sorry, but I do not know if they are direct quotes or just notes. I am not trying to plagiarize.

So I can keep track of it this is version 3.0
These are some only seminars that if you really want to know more might help. Thanks Electronic Design and Digi-Key and a bunch of companies.

• Designing for LED Luminaire Reliability
• LED Color-Temperature Control
• Controlling LEDs with Microcontrollers
• Using Analog Semiconductor Technologies to Power Solid State Lighting Applications
• Characteristics of HB LED

INFORMATION COLLECTED FROM HERE (and probably some others)
So if you are still interested in doing LEDs I recommend looking at them as well:

• This thread soundwave’s– just incase this gets posted else where
  This is the first post in the original split. For you those that are new there used to be a 40 page limit to a thread – a new thread was then started until it reached for pages. Etc.
• LED build and PAR readings (pic intensive) - Santoki
• Taqpol's DIY LED prototype build, detailed pictures
• LED lighting on a budget! - kcress
• Widmer's LED Projection System

Related threads that might be useful:

• DIY LED driver for reef lighting

There are numerous other threads relating to builds, but I think these cover most of it. If you think I missed an important thread (highly possible there are a lot of threads out there), please let me know.

LED Specs
Cree XR-E Royal Blue
Color: Royal Blue
Dominant Wavelength Range (nm): 450 – 465
Max Current (ma): 1000
Viewing Angle (°): 100
Standard Min. Flux @ 350 ma: 425 mw, 350 mw

Cree XP-G Cool White
White: Cool
CCT (K): 8,300 – 5,000
Max Current (mA): 1500
Viewing Angle (°): 125
Standard Min. Flux @ 350 ma: 139 lm, 130 lm, 122 lm, 114 lm

WHERE TO GET THE STUFF
Here are the most common location where people are getting supplies.
LEDs

• Rapid LED
• LED Supply
• ETG Tech I find this site confusing so you may have to contact [email protected]. Also last time I checked the minimum order was $75, which is doable with a 12 LED configuration with lenses.
• Deal Extreme The cheapest place for Cool White Q5 XR-E (~$4.50), but slow to ship.
• Cutter They are out of Australia, good prices. I have read they are slow to ship, and at one point had real hard time getting you what you ordered.

Power Supplies

• MPJA For small builds they have a
• 24V 6.5 amp for a good price.

TOOLS
You will need the following tools (and some others probably) to complete this project.
Soldering Iron
Most of the recommendation is for a 40 watt. There was a nice discussion in one of the threads. If I could just remember where I would quote it. A post by der_wille_zur_macht says 40-50 watts.
Here is a decent, low-priced soldering iron that CJ found.
How to Solder a Star

Originally Posted by kcress

In the LED thread I detailed proper soldering technique. Here's it is lifted in its entirety:

Soldering... It's a bit of an art form. It you're an expert you could probably solder the stars with anything but a torch. But if you aren't accomplished a 'just right' soldering iron is required. Something around 25 to 40watts. The biggest problem would be one too small as then you damage what you are soldering by sitting there with the heat on, waiting, and waiting for the solder temperature to be reached. Meanwhile the device is being roasted.

You need "rosin core solder." Make absolutely sure it's not 'ACID' core solder. Acid core is solder for soldering radiators and copper pipe. The acid remains and will dissolve your electronics in short order. No-core or plain solder is also very bad as there is no cleaning action which is vital to an acceptable solder joint.

You want lead solder too - NOT "lead free".

Let your iron heat up fully. Then wet it with your rosin core solder until it's blobbed up in a wad. This lets the rosin act on the tip to clean it. Wipe this blob off on a moist sponge,(not sopping wet, just barely damp). If you have no sponge use a damp folded up paper towel.

Wipe with a very fast twirling dragging motion as you are trying to clear all the solder off the tip before you cool the tip to the point that the solder freezes again.

Apply a weee bit of solder to the tip and then immediately apply the tip to the star's pad. As soon as you apply the tip to the pad take your solder and touch it to the exact point where the tip meets the pad. You are trying to maximize heat transfer from the tip to the pad. You do this by filling all the voids with a molten metal thermal bridge. Pause for, perhaps, one or two seconds move the solder to a distant place on the pad and hold it there. When the pad heats up enough to melt the solder feed the solder in at a good clip until the pad has a large mound of solder on it.

Do the same thing to your wire if it is un-tinned. (actually do this before you do the pad!) Better(superior) would be to use only pre-tinned wire in the first place.

Now once you have your mound of solder on the star pad bring over your wire, hold it on the Star's pad. Place the tip on top of the wire. You want to reheat the pad thru the wire. When the wire sinks into the blob hold the position and watch the blob. You want the entire blob to melt again. The instant the blob has remelted remove the tip while holding the wire EXACTLY where it is. No wiggling allowed! Coffee shakes need not apply. Once the solder has refrozen move on to the next connection.

Originally Posted by kcress

Hi hllywd.
Pre-tinning has its place but.. The physics of soldering are kind of complex. When you pre-tin you leave a layer of already heated flux on all the various surfaces. This can cause subsequent soldering to be sub-par. Also using the 'heat up the two and just melt them together' method means you get almost no flux action on this second solder-less heating.

If you ever watch the flux action under a microscope you will see the flux push everything foreign away allowing only the two items to be connected with solder. When you use the tack them together method none of this happens.

In most cases for our LED fixtures it's still good enough. But I would rather not. I realize that to do it the right way you need three hands. I have one of those little stands with alligator clips sticking off of it. I just grab the wire with one of them and have it hold the wire to the pad. Then I can show up with the iron and the solder.

Jay1982; You would want to re-heat that joint. You always want a hot solder joint, not a cold solder joint. 'Bulbous' is often a sign of a cold solder joint. The other sign is a dull un-shiny result.

This thread has a lot of information on soldering
Soldering Technique

Multi Meter
You will need to be able to measure voltage, current and you will probably want resistance. Almost all meters have these so it should not be a problem. However, when measuring current you will need a meter that goes to at least 1 amp I would recommend 10 amp.

And a few tips from kcress:

Originally Posted by kcress

You can change the settings with the driver on. Be very careful you touch nothing when that driver is plugged in. Anything could be live, like those pieces of metal.

On your meter, if you change things over to current measure absolutely always power down. Remember if you interrupt the chain while the driver is powered - you will fry the chain.

Also, always unplug the meter current lead from the meter and replug it into the voltage position IMMEDIATELY when done measuring current. Not doing that will surly result in you picking up the probes and trying to measure a voltage while the meter is in current mode. Result - blown meter.

LEDs
Which LED
The whole reason most of got into this light scheme was for efficiency. Having said that there have been a lot of question like will this LED work. The thing to look for is efficiency and spectrum. If the spectrum is what you want (matches the bulbs in there now) then you are fine. Most people have picked the CREE XR-E and XP-G LEDs because of their efficiency of over 100 lumens per watt. If the LED you are looking at is not over 100 lumens per watt you probably don’t want it.

How Many
The current recommendation is 1 LED for every 10-20 square inches of tank surface. Fish only could get by with the lower count, and a coral tank would need to be near the higher end. However if you take advantage of LEDs ability to focus corals could be spot lighted and cut down on the number of LEDs required.

More recently I have seen to set of numbers based on square inches:
The guideline (Ok this is one coming from memory):
12-14=Hard Coral
14-16=Soft Corals
16-18=Fish Only

The guideline:
10=Hard Core (Too Much) Light. Very Deep Tank
15=SPS (most people use)
20=LPS, Softies
25=Softies

Spacing
Currently the general recommendation is 2 to 3 inches between LEDs.

Color Mix
Currently the general recommendation is:

• 50/50 for white XR-E and royal blue XR-E
• 40/60 for white XP-G and royal blue XR-E

If I remember correctly this is 14k. If you want a bluer tank you more royal blue. Before you ask, there is no XP-G in royal blue.

To find more information on color, see der_wille_zur_macht’s post .
Here are some more pictures that may help:
From lpsouth1978

And I can't imagine where dur_wille_zur_macht found these
XR-G spectra:


XR-E spectra:


Ultra Violet LEDs
Since a corals UV protection is clear and metal halide lights have UV shields, it is believed that UV LEDs are not needed. But if you do here is a source that hlsooner found: UV LED

Forward Voltage
This always comes up and I found this real nice post that was talking about meanwells, but applies in general.

Originally Posted by der_wille_zur_macht

Yep, that's more or less correct. In the datasheet for each respective LED, there's a V/I curve that lets you look up the voltage the LED will drop at a specific current. With this information in hand, you can calculate the voltage expected across a certain number of LEDs in series.

The numbers do vary a bit from model to model and even color to color, so it's important to look up the actual figure if you're interested in doing the math. And of course, clearly, it varies as the current varies.

Plus, there is a fair amount of variation from LED to LED, around the number quoted in the datasheet. So, LED XYZ might be quoted at 3.5v at your desired current, but you might find some that drop 3.55 or 3.48 or whatever. So, it's important to leave a margin of safety when planning how many LEDs can be run on a given driver.

In most cases, with the 48v meanwell drivers, you can just assume 12 LEDs and be safe. But if you're using some of the lower-drop LEDs (i.e. the XP-G or Rebels) and lower currents, and you want to really load things up, you might be able to get away with 13 or even 14 per string.

POWER SUPPLIES
If you are doing a really big fixture multiple power supplies can be a problem. See this post by kcress:[*]Summary of multiple power supply issues)

SPLASH GUARD
I think the recommendation is anything less than two feet should have a splash shield to keep the spray away from the LEDs.

Wire
To quote der_wille_zur_macht, “Wire gauge between LEDs should be 20 - 26 gauge, pretinned, with good insulation.”

Kcress found a great place for wire:
Stranded wire

HEAT SINK
Types
True aluminum heat sink. Most have used this so far but it is expensive and heavy. The advantage is that will a fan to get good air flow you won’t have to worry about heat.
People are beginning to experiment with some the the aluminum channel found at local hardware store. Not a lot of reports, but I have not heard anything bad. The advantage are that is lighter and cheaper. I also think it would work better for convection cooling. If air can rise through your fixture with smaller pieces of aluminum you should get fewer areas where air flow is limited – also yet to be proven.

Mounting
There are three main mounting options (that I can remember).

• Screw and thermal paste
  Screw make for the easiest changes later, but a lot of time to drill and tap. I personally did the screw method, but did not tap. If you follow me lead DO NOT tap with the same screw twice – the head will twist off. Oh what fun .
• Thermal Adhesive pads
  These may be hard to take off. Some have done it, but it was early and the longer they sit and the hotter they get The better they sick if I remember correctly.
• Thermal epoxy
  

  Originally Posted by Jay1982

  To the OP: I just spoke with a rep at Arctic Silver about their Arctic Alumnia thermal adhesive. If you use this product, its on for life. He said its basically made with the same compounds as used in cement and if someone were to try and remove an LED from the heatsink that the LED itself would fragment apart before the adhesive gives way.

  However, people have been able to remove LEDs that were affixed with the thermal adhesive without too many issues in the short term (a couple of weeks).

How hot is too hot
Kcress said this very clearly:

Originally Posted by kcress

If you can keep your finger on it you're OK.

And I mean the bulk area of the heat sink not the edge of a fin.

After doing some research kcress changed his mind:

Originally Posted by kcress

Temperature..

While some effort is required in this direction almost all the builds we're seeing, with the exception of Stu's, are massive overkill. Note my recent build. That's just 0.063" Al with some simple fins. Even that is more than necessary! We should step back and recognize that trying to get 10 years out of these fixtures is folly.

Look how far the state-of-the-art HBLEDs have gone in the last 3 years. We are at 100 lumens per watt now. It is expected that we will be at 150 lumens a watt in something like 2 years. Are you really going to want your same old fixture running over your same old tank in 10 years, when your fixture is costing you perhaps three times more to run than a newer fixture? That's way after the time you should've recycled your present one just to save energy costs.

Spending a fortune on aluminum and running 6 fans for a wind tunnel sound effect is not the way to go.

I recently researched this for my build.

How hot?

This is how hot:

One hundred fifty eight degrees F directly behind the center of the star on your heatsink's base metal - not fins - is where we should be aiming.

That puts the LED's operating temperature on the far far left of the recommended operating curve for lifetime. The curve goes far to the right from there. How far? The curve goes to 302F!!!

So, even though in the past I've suggested if you can touch it you're certainly cool enough I've changed my opinion. With my new understanding I'm upping that recommendation to a value that actually makes more sense. 158F

158F is instantly painful to the touch. That's a reasonable temperature - as far as the HBLEDs are concerned. At that temperature the L70 is still seventy thousand hours that's 19 years!!! LEDs will probably be 1000 lum/W by then and our fixtures will be long dead, economically and aesthetically.

Save some money and start aiming for too hot to touch directly behind the LEDs and really, really warm heatsinks.

Heatsink Size
CJO found this information and I think he got it from lynxvs.
[QUOTE[I've posted this before I think but it might be helpful. I use just a flat plate of aluminum as my heat sink. I did some calculations below to justify heat sink size. I attach a PCB directly to the plate using screws.

Max Junction Temp = 150° C
Power of Single LED = 3.4 Forward Voltage X 700 mA = 2.38 W
Ambient Temp = 70° C ( A SWAG)
Thermal Resistance between Junction – Case (From Data Sheet) = 10° C/W
Thermal Resistance between Junction and PCB (From Rebel application note) = 7° C/W
Total Thermal Resistance = 10 + 7 = 17° C/W
Total Thermal Resistance between Junction and ambient air = (150 – 70)/ 2.38 = 33.61 ° C/W
Thermal resistance between Case and Ambient air = 33.61 – 17 = 16.61 ° C/W

The amount of heat dissipation that can be achieved with a flat plate of aluminum is indicated below.


Using a 3mm plate looks about 20 cm^2 per LED converting to inches is equal to 3.1 in^2 * 50 LEDs = 155 in^2 The plate I am using is 24 X 7.25 = 174 in^2 not sure if you can count both sides of plate as surface area… I also have two cooling fans to help [/QUOTE]

[TheFishMan65]

DRIVERS
There are two main type of drivers. I did not use either one so I did not pay a whole lot of attention to these. But before we get there, there has been discussion about running string that involved terms like series, parallel, matrix, and others (thanks CJO for reminding me about this fine kettle of fish ). Generally it is agreed that each driver should drive one string as kcress says (I think this was all him):

You should only have one driver for each string. Period.

Any other scheme risks all the LEDs as soon as one fails shorted.

Two stings in parallel will toast them all. Why?

If you are running two stings in parallel and each string is, for example, 700mA, your driver would need to put out 1400mA. Now if one LED shorts the driver will continue to drive 1400mA into the two stings. But the string with the shorted LED will have a different voltage requirement than the good remaining string. This causes what is termed as "current hogging". The good sting will either go dim or OFF completely while the bad string may have 1200mA running thru it. The remaining LEDS will fail in seconds.

Once the entire string with a short in it has blown or one of the LEDs fails OPEN the driver will then focus on driving the 1400mA thru the remaining good string. Every LED in that string will also fail within seconds in a domino effect.

One driver per string!

OK, CJO was really interested in this and has a good reason, I don't recommend (and I think most will agree) this, but if you really want more information check the bottom for some notes CJO collected.

Meanwell
These are nice because the run off of 120 so no extra power supply is needed. There are several different type. The differences (I think) are how much power they can supply and how they are dimmed. This thread seems to answer a lot of questions: How to dim a Meanwell ELN-60-48D...
D Version
I have seen several question on wiring so since Stugray said it so clearly:

Originally Posted by stugray

The AC wires that go into a plug dont matter which way.

The V+ goes to a + on the LED the - of that LED goes to the + of the next LED etc.

The - of the last LED in the string goes to the V- of the driver.

The two DIM signals go to your ALC dimming channel.

Set the dimming output of the ALC to max (+10V) then set the meanwell max current to what you want the LEDs to run at. DO NOT turn on the driver until you have turned the internal dim pot counter-clockwise all the way or you will blow the string when you turn it on the first time.

THEN adjust upwards. To measure the current properly you need a digital Multi Meter placed inline on the LED string to read the current.

Stu

Not everyone found that information, but this shows it all pretty clearly:
Meanwell and how to use it - for Idiots like me

Internal Current Adjustment
Some meanwells (like the D) have an internal adjustment to limit current. Der_wille_zur_amacht explains the different ways of limiting these very nicely:

Originally Posted by der_wille_zur_macht

To help clear it up. The driver you have has TWO ways to adjust the power that the LEDs are running at (actually three, but the third - voltage limiting - doesn't come into play when it's run as a constant current LED driver.)

First, there is the internal trimpot. Think of this as the max. limiter for current provided to the LEDs. The trimpot sets the max current (amps) your LEDs will run at. You adjust this internal pot simply by turning it with a screwdriver and reading the CURRENT (amps, not volts) on the output string, i.e. in series with the LEDs.

Second, there is the external dimming signal, connected to the external wires. This external circuit allows you to dim DOWN from the max current set by the internal pot. You control this function by providing a voltage signal to these wires, between zero and 10 volts. The LEDs aren't actually running at this voltage, it's merely a signal. The driver translates the signal into a percentage of the current set by the internal trimpot.

So, when your external signal source (ALC, whatever) is providing a 10v signal, your LEDs aren't running at 10v, they're running at 100% of the current limit set by the internal pot.

The internal pot is a set and forget kind of thing - you configure it when you install the system, and then you likely won't touch it again. Depending on your plans and which LEDs you're using, typical values might be between 700 and 1000mA.

The procedure Stu suggested above is 100% exactly what I would suggest. Turn the trimpot all the way down (counter clockwise). Wire everything up, with your external control signal set at 100% (i.e. 10v). Put a multimeter set to measure amps in series with the LEDs. (make sure it's well connected, you don't want loose connections!) Plug the whole thing in, and measure the current. Leaving your external signal set to 10v (100%), adjust the internal trimpot upwards until you get the current you want, which will be between 700 - 1000mA for most people. Then, unplug the driver, remove the multimeter from the circuit, close the circuit, and you're ready to go.

BuckPuck
These are nice because they are small, but a separate power supply is needed.

LENSES
I found a really good post by der_wille_zur_macht so I will just leave this with what he said.
[QUOTE=der_wille_zur_macht;17061951]

Originally Posted by TheFishMan65

I think I remember reading that there was a recommendation of 1 LED for every 10 to 20 square inches. The above spacing would give 1 for every six inches. taqpol used 1.75 by 2. Lots of light in the middle (especially with lenses), but I little dark for what I think I want on the edges. So I will try and adjust my plans based what I have learned for taqpol.QUOTE]

In the past, I was suggesting people end up with between 10 - 20 square inches of tank surface area per LED. That's where the vast majority of "successful" medium-sized LED rigs have ended up, depending on drive current, livestock requirements, height, optics, LED model and bin, etc.

So, for instance, if you have a 120g tank, it has 48*24 = 1152 square inches of surface, so you might want 60 - 110 LEDs.

Those rough guidelines are probably about right. It's hard to give really specific suggestions because of all the variables, and not every type of LED has every width of optic available (i.e. there really aren't any 60 degree optics for XP-G that I'm aware of). I might cut your ranges a little as follows, for "average" depth tanks (say, 18").

80 degree: anything up to a foot
60 degree: 18"
40 degree: 24"
20 degree: 36"
anything smaller: on the ceiling!

As noted above though, concentrating the light with optics has other inherent effects besides the obvious one of increasing intensity and reducing spread. The spacing between your LEDs becomes a little more important as your optics get narrower, both from a coverage perspective and a "color separation" perspective. When you have wide optics or no optics, it doesn't matter that you are alternating blue and white LEDs every few inches (typically) because by the time the light hits the tank, it's "mixed" together well. With really tight optics, you can get an effect where shadows look blue on one side, and yellow on the other side - this is because the different colors of light are so laminar that they don't "mix" evenly.

And, of course, you can get the typically undesirable "shafts of light" effect if you use tight optics and big spacing between your LEDs. So, typically, as your optics get narrower, you are forced into a closer LED spacing to compensate for these effects, which COMPOUNDS the typical issues of using tight optics (less light spread and more intensity directly under the LEDs.) Effectively, if you want or need to use tight optics on a larger tank, you are FORCING yourself into a very high output light fixture, because you'll have to increase your LED count (decrease spacing) to get good coverage. The alternative if you really want a lower-output fixture is to use lower drive currents along with the closer spacing and tighter optics.

One final thought on optics that isn't brought up often. As stated above, tighter optics result in more laminar light. As you change from wider to tighter optics, your light behaves less like an MH and more like a laser. As this happens, we start to get some effects that may be considered desirable, but it's very possible for them to become overwhelming with very tight optics. Namely, shimmer and "contrast" in the sense of shadows becoming very, very well defined.

Shimmer is pretty much a "good" thing to most reefers, and you can control/influence it easily by changing the amount of turbulence on the surface of your tank. So it's a non-issue.

Level of contrast in the shadows might be considered a good thing, up to a point. As your optics get tighter, there is less "sideways" light in the tank, which means that corals/rock located in a shadow will get less and less light. Up to a point, IMHO, this can be very desirable - the sharply defined, dark shadows can look really cool. BUT, for people with lots of corals, it can be undesirable if taken TOO far, because lower branches on big colonies will be receiving very little light, which might cause growth problems or die-back for the coral. This can be somewhat avoided by letting your corals grow up in the LED-lit tank, and/or creative effects with LEDs.

Which brings me to my last thought for this post (I promise!). The above "dark shadow" issue could be easily thwarted if a few LEDs were angled with respect to the tank, instead of all of them pointing straight down. To date, 99% of LED builds have been BORING! The typical approach is to get a big heatsink and spread X number of LEDs evenly across it, then hang it above the tank. This gives a final result very similar to what you'd get with T5 and MH. A perfectly even and vertical distribution of light from edge to edge across the tank. LEDs let us BREAK FREE from that mold if we so desire! You can't bend a T5 lamp to get "angled" light at one end of the tank and vertical light at the other end. BUT that is easy to do with LEDs! Us reef folk seem happy to have actinic lights coming on before white lights as "sunrise" simulation. Imagine a tank where the "sunrise" actually involved a narrow shaft of light piercing the tank in a nearly horizontal fashion. The possibilities are endless! We can position individual LEDs, or small banks of LEDs, at different heights, angles, and positions to get some really cool effects. Time to think outside the box!

XP-G can be hard to get lenses for.

XP-G Optics: I ordered the 60 degree optics from etgtech.com for $1.25/pc plus shipping.
[IMG] http://www.reefcentral.com/forums/pi...ictureid=21604 [/IMG] I don't believe they have them listed on the site so you will likely have to email [email protected]. !

Lenses for the XP-G are getting easier to find:

Originally Posted by terahz

TheFishMan65, just to add on the xpg lenses for your summary, I ordered last week from Anna, and they have 60, 40 and 25 degrees (as others have mentioned earlier) that fit over XP leds. Same price. They are made by these guys:
http://www.ledlink-optics.com/ll01cr_dfxxl.html

I started quote everyone else, because they are much more wordy and descriptive than I am, oh and did I mention it is easier. Most of this quoted information was added in as the question came up after the original post so rather than paraphrasing and trying to remember they get full credit.

A EASIER WAY TO MEASURE CURRENT

Originally Posted by kcress

Monitor resistors are large carnivorous animals that ea... Wait no! Wrong monitors!



Your question has been asked and answered about a dozen times so I'll just repeat one of those here if you don't mind.



This is why.
It's required so you can measure the inter-string currents in situ.

You can put them anywhere in the string. By definition with series, all current in a string must pass thru all elements of the string. The resistor will be an element of the string. Stick where ever you want. However.. If you stick them all on one end of all your strings, (one driver lead), when you want to compare the currents in all the strings to each other you only need to move ONE lead of your meter to measure the various strings. If you put the resistor in any other locations you would have to move both meter leads for every measurement.

In parallel string driving YOU MUST HAVE BALANCED STRINGS. Otherwise you can have two or three times more current running in adjacent strings. This means you could plan on having 700mA running thru all your strings before you balance them and you could easily have 2A running thru one and 0.05A running in the neighbors! The 2A string is gonna die in seconds.

You cannot do this by looking at it. The human eye can't decipher in detail ANYTHING about a light source that is as bright as these LEDs. (You should never even look straight at them anyway.)

With a (one)(1)(uno) ohm resistor installed in each and every string you can instantly measure the exact current flowing thru that string with just a voltmeter. No wires to open. No careful power down, then cool down while you cut/solder open the string, followed by installing an ammeter. Then powering up. Now reading the current that is no longer the same because the whole system has cooled down. Then taking this not quite right reading, writing it down, powering down, re-soldering the open string. Powering up. Next... Only 10 more?

No, No, NO. With the aforementioned resistor permanently installed in the chain, just whip out your voltmeter, set the selector to volts, and measure across the resistor. The number you read IS THE ACTUAL CURRENT flowing thru the string. No soldering. No temperature issues.

If you measure your strings like I just described and see 680mA, 720mA, 702mA, 698mA. You are good to go! Do not be an anal-idiot and try to get them all identical. They will all drift with age anyway. However, with the handy dandy resistors permanently in place you can check any time you want.

More likely you will probably see 590mA, 639mA, 727mA, 880mA, etc.

You then need to take your meter and while everything is running measure all the individual LED Vfs of the 590mA string. Write them down.
Do the same for the 880mA string. Write them down too.

Take the highest Vf one out of the 590mA string and swap it for the one with the lowest Vf in the 880mA string.

Remeasure ALL the string currents with the handy resistor again. You would now see something like: 690mA, 639mA, 727mA, 780mA, etc.

Keep working this same game plan until your strings are all within about 30mA. This will result in a long trouble free life for your strings. Periodically check all the string currents with the handy permanent resistors. Check maybe a month later. Make sure they are still in that 30mA range. If they are, recheck in six months. After that check every year.

Your neighborhood street signal lights can't be checked like this so in about 5 or six years the string currents shift so far that only certain strings carry most the current and fry. Not us though - because of the little miracle resistor.


The terminal blocks are to keep things clean, stable, cooling in the air, and available. They also allow easy fuse replacement. If you run dangerous sting voltages you would need to protect from accidental touches.

You use both meter probes across the resistors.

No formula. Use 1 ohm, 2 or 3W resistors. Use 1A fuses and use whatever terminal blocks you want to.

Originally Posted by kcress

Online:

Use these terminal blocks if you want 2-position:
http://search.digikey.com/scripts/Dk...name=A98503-ND

or use these if you want 6 positions:
http://search.digikey.com/scripts/Dk...name=A98507-ND

Use these resistors:
http://search.digikey.com/scripts/Dk...me=UB5C-1.0-ND

Use these fuses:
http://search.digikey.com/scripts/Dk...&name=F2313-ND

Originally Posted by kcress

A 2W is fine a 3W a little better. You can't run 1W thru a one watt or it will fail in months. It will also burn you the instant you touch it. A two watt would last indefinitely and would probably be painful to touch but wouldn't burn you. A 3watt would not probably be painful and would again last indefinitely.

If you're running 0.8W (800mA) thru a one watt rather than 1A it will probably last as long as the LEDs. It will burn you if you touch it for any length of time. In all cases we since we're mounting the resistor up in the air suspended between terminal blocks it will be as cool as it could possibly be in an application.

If you are really hung up about the power aspect you can actually use a 0.1 ohm resistor in the 1/4W rating and it won't get warm. But then you have to always multiply the meter readings by 10 which is often another point of confusion for beginners.

Here are some recommended parts:

Originally Posted by kcress

Online:

Use these terminal blocks if you want 2-position:
http://search.digikey.com/scripts/Dk...name=A98503-ND

or use these if you want 6 positions:
http://search.digikey.com/scripts/Dk...name=A98507-ND

Use these resistors:
http://search.digikey.com/scripts/Dk...me=UB5C-1.0-ND

Use these fuses:
http://search.digikey.com/scripts/Dk...&name=F2313-ND

[TheFishMan65]

Paralleling LED Strings – Make Sure You Understand the Risks
Maybe I should write this really small so it gets ignored. I started thinking about this kind of late so did not try and remember what I read on parallel, but here is some basic information:

Originally Posted by TheFishMan65

Since I started the summary I have been thinking more about the parallel configuration.

As a general rule we have been saying avoid parallel configurations. After thinking I wonder if the Meanwell's are meant to driven in parallel. The advice was to ignore the internal voltage setting potentiometer. But if we set the voltage to the measured voltage when running at the desired current then when a string is lost the Meanwell us unable to increase the voltage to get the additional current out. I don't think any additional safety precautions are needed.

I think this was mentioned before, but just did not sink in for me. So where is the problem if both internal adjustments are PROPERLY set with running parallel strings?

And kcress pointed out what I forgot:

Originally Posted by kcress

The problem is shorts in a string.

With a short the required voltage of the string drops and the current increases, a vicious circle.

All the streetlights and signal lights run parallel with no protections and the strings all burn out eventually. I was just noting yesterday that about 1 signal light in 3 has failing LEDs showing. My town embraced LED signal lights the minute they came out. So what, 6 or 7 years? And now failures are starting to occur. If you are fine with retooling your fixture in that period it will probably work fine to parallel with no added protections.

So I think with a properly adjusted Meawell and some fuses you should be ok.

So for those interested here are notes CJO (who was interested in this as he read the threads) took:

Series or Parallel

Fairly simple actually. With the same power supply, in this case for 4 LEDs with a forward voltage of 3.3v, of 15v.

The series circuit, uses less current draw from the power supply, and a single resistor.

The parallel circuit, uses a higher current draw from the power supply, and larger multiple resistors.

Another way to look at it, you need a larger voltage, lower amperage, power supply for series circuits. A smaller voltage, higher amperage power supply for parallel circuits. Assuming the same number of LEDs.

With power supplies, having less than the total forward voltage of all the LEDs, it is necessary to use a series/parallel circuit. In this particular case, the power supply is easy to come by, and the closest standard size resistor 120 ohms, is the exact size needed. The other circuits use the next closest higher resistor. (results in dimmer LED output, because less current will flow.

So from a design point of view, you design for the type circuit that will give the best results, in this case, although the series circuit is close enough, the series/parallel circuit with a 9 volt supply, will perform the best. (LED output wise, power is a secondary concern-- except for dissipation, depending on the size of the array)

See circuits below. If there are math errors, it is the calculators fault (it is an old TI-36X)

The greatest difference is seen in the total power (wattage) for the circuit.


kcress on the possible issue with parallel strings (if proper precautions aren’t taken):
You should only have one driver for each string. Period.

Any other scheme risks all the LEDs as soon as one fails shorted.

Two stings in parallel will toast them all. Why?

If you are running two stings in parallel and each string is, for example, 700mA, your driver would need to put out 1400mA. Now if one LED shorts the driver will continue to drive 1400mA into the two stings. But the string with the shorted LED will have a different voltage requirement than the good remaining string. This causes what is termed as "current hogging". The good sting will either go dim or OFF completely while the bad string may have 1200mA running thru it. The remaining LEDS will fail in seconds.

Once the entire string with a short in it has blown or one of the LEDs fails OPEN the driver will then focus on driving the 1400mA thru the remaining good string. Every LED in that string will also fail within seconds in a domino effect.

One driver per string!

Drivers cost money.. How do you deal with this?

Two ways: The first is to string far more LEDs in a string. Using a 36V or 48V driver or at least 24V. 12V borders on the ridiculous.

48V/2.2V = 21 LEDs
36V/2.2V = 16 LEDs
24V/2.2V = 10 LEDs

Alternatively you could run strings in parallel but you would need to put a fuse in series with each string. As soon as an LED fails shorted that string would hog current, exceeding the fuse rating, and the fuse would blow. Promptly the full current would try to run thru the adjacent parallel strings and those fuses would also promptly blow. It would take some careful fuse selection however.

How to test forward voltage of individual LED’s:
Wire them to a driver (you can do lots at once if you want). Run them at your target current. Read voltage across each one with a multimeter, by probing right at that LED's solder pads.

Really, the only trick to it is not blinding yourself. It helps if you have optics or an optic holder that leaves the pads exposed, because then you can easily look at the LED from the side without getting blinded.

(der_wille_zur_macht)
Parallel strings are not ideal in these applications, for a few reasons. First of all, if there are any tiny differences in characteristics in your various LEDs, you'll have inconsistent performance. If you have one string that ends up requiring 10.3v to drive at 700mA, and another string that requires 10.5v to be driven to 700mA, then the driver will end up over driving one string and under driving the other. Since even a small variation in voltage can lead to huge variations in light ouput, this might mean poor performance from some LEDs. In practice, I've seen 3 - 4% variation in drive voltage to achieve a target drive current from LED to LED (even in the same bin) which is enough to make me worry about performance in parallel applications.


A: (der_will_zur_macht) Daniel, fusing the parallel strings would prevent failure, but it leads to some other (potential) issues:

1) If the LEDs in one string have a different total forward voltage at a given current than the LEDs in the other string, they won't balance out well. This would be especially true if you mixed different colors/types of LEDs on the same driver. I'd want to carefully "bin" the LEDs I was using (set up a test station where you could drive a single LED for a few seconds to record it's voltage at a given current) to avoid this.

2) As you get more LEDs on a driver, you start to lose control resolution. Maybe this isn't an issue on a very large tank, but on a smaller tank, if you had drivers doing 12, or 24 LEDs each (for example) you quickly lose resolution to the point that it would be hard to implement the sort of control people are starting to show interest in. For an extreme example, I have a nano rig with 16 LEDs run at very low current. This is two of my DIY drivers, 8 LEDs each. A driver capable of doing all 16 wouldn't even let me dim blue and white separately.

From the sounds of your posts, neither of these would be huge stumbling blocks for you, but I wanted to point them out in case others were following along.

Binning LED’s, adding Fuses (Kress)
You could theoretically run 4 strings of 48/3.5 = 13 LEDs.

Or 52 total.

You would be limited to 1.3A / 4 = 325mA per string.

To do it right though you'd need to do some additional work.

It would consist of some detailed meter work.

You would set up a string on a Mean Well and set the string current to 325mA using an ammeter.

Turn it on and wait until the string is warmed up. As you wait, use a Sharpie to number every one of them. Once warm measure the voltage across each one and write it down in a numerical table.

Do this for all 52.

Now take this table and mix and match the values to end up with the same total voltage in each string. You could do this many different ways. Use, say, the highest 5 with the lowest 6 if that works. Or just match across one low one in each string then the next higher one in the next string, etc, etc.

Once you have them grouped build your 4 strings.

You need to build the strings normally BUT you need to add fuses in each string.

Something like a 375mA fuse. Digikey F1504-ND in a holder F1467-ND.

Now when a LED opens or one shorts the fuse will open protecting the rest of the string.

Note that if any fuse opens they will all open, so keep spares.

If you can't pull this off as described, don't run parallel strings.

(der_will_zur_macht)
Skeptic, it's an easy problem to solve. Set up a "test station" with a constant current driver that can power a few LEDs at a time at some reasonable current, while allowing you to probe each individual LED with a multimeter. Turn the test array on, test the voltage drop across each LED, and write them all down. Then, arrange your LEDs into groups such that the total voltage drop for all groups is as close as possible. That's what I meant about "binning" your own LEDs. It should take an hour or two max, and it's cheap insurance if you're running parallel strings.

.2v CAN be quite significant (like 100mA!!!), but I'm not sure you'd see variation that high unless you randomly stacked things up in the worst possible way.

Originally Posted by kcress

GEORGEDOPE; If you want to use that driver with parallel strings, go ahead. Add a 5 Ohm 10Watt resistor in each string. This resistor will help balance the string currents due to the inevitable mismatched string voltages.


Next turn down the voltage limit on the driver until you just detect some dimming. Then turn it back up a wee bit.

Now when one of the three 700mA strings opens the remaining two will theoretically have 1050mA sent thru them. But to have this happen the voltage must rise significantly, except, you just limited that voltage rise with the adjustment. The limitation will keep the current below 1050mA by a significant amount.

The only thing you aren't fully protected from is a shorted LED in one of the strings. The math shows that with those resistors you would probably see one string increase about 50mA and the other two drop by 25mA. Someday I will check that on a build. For piece of mind you can add a 3/4A fast blow fuse to each string.

Partial Summary
First, I don't think there is any difference between the cool white and royal blue (XR-E). Both are 3W, have a forward voltage of ~3.6v, and take a max current of 1000mA (note that the XP-G can run up to 1500ma). However, I've read that the royal blue (and white to a lesser extent) are best run at a slightly lower current. Say 700mA. This extends the life of both types of LED with only a slight decrease in light output. This means that both colors could be on the same driver. But, most people don't so they can dimm them by color and turn them on at different times to simulate sunrise/sunset. I'll try and get confirmation on this.

The Meanwell, specifically that in the group buy ELN-60-48P, is a line voltage constant current source. It provides a constant 1.3A (-25%,+3%) to the LED load on it. It can handle up to 48V max on the load. And the P means it is externally dimmable with an analog signal. The line voltage part means it just plugs into the wall (90 - 240v), no additional power supply is needed. The max constant current is changeable by an internal potentiometer -25% or +3%. This means it can output from 1A to 1.4A or so.

LEDs have a current requirement and a forward voltage. For the Cree XR-E that is ~1A and 3.6v (on average). In series, voltage adds and current is the same. Thus, the Meanwell can power up to 13 XR-E in series at 1A (with the max current at -25%).

It can also run two parallel strings of 13 LEDs at 700mA per string with the max current set to +3%. Current divides in parallel so a total of 1.4A is being provided but each string gets 700mA.

A word of caution with parallel strings of LEDs. This is a recipe for disaster. If you aren't comfortable building this next item, don't run LED strings in parallel. What happens is if one LED in one string dies, then that string shorts and all of the current is sent through the other string. This either kills part or all of the LEDs in the other string or it severely limits the life of the LEDs in the second string. The higher current will work but it will also heat up the LEDs a lot. That is how they die.

But, you can do parallel strings as long as you build in a current mirror. Evil66 found and posted this in the Meanwell thread on nano-reef. Basically it forces the same current in both strands. If one strand shorts then it shorts the other strand too. The Meanwell thread:circumventing the filters is generally frowned upon

(CJO)
This is how I set up my parallel strings. I soldered together two random strings of 12 LEDs each. I then connected one string at a time to the driver used a multimeter to set the driver to the amperage at which I was going to run each string (700mA). I used some magazines to cover the LEDs so I wouldn’t get blinded by the light and let them warm up. Next, I used the multimeter to read the forward voltage of each individual LED and recording them on a table.

By adding up the forward voltages in each string, I was able to determine an LED from each string that I could swap to balance out the forward voltages for both strings. I then connect a 1A quick-burn fuse to each string and connected them together so that they would run in parallel. I again used the multimeter to increase the amperage on the strings so that each string was running at 700mA. I’ve had it running for a couple of months now with no issues.

[TheFishMan65]

Oops I lost a bracket (seems and appropriate time) and who know what else, but let's give Wille credit where I should have.
LENSES
I found a really good post by der_wille_zur_macht so I will just leave this with what he said.

Originally Posted by der_wille_zur_macht

In the past, I was suggesting people end up with between 10 - 20 square inches of tank surface area per LED. That's where the vast majority of "successful" medium-sized LED rigs have ended up, depending on drive current, livestock requirements, height, optics, LED model and bin, etc.

So, for instance, if you have a 120g tank, it has 48*24 = 1152 square inches of surface, so you might want 60 - 110 LEDs.

[bamf25]

The eln-60-48's that most people use are rated for 60 watts. About how many watts per string are these drivers really using? If you say the leds are using a foward votage of about 3.2 for the XP series and you string 12 that is 38.2 volts of the 48 avaible used, how does this translate to watts used.

[TheFishMan65]

You have to look at the curve the the LED you are using. IIRC the XP-G at 3.2 volts is about 700ma. So about 27 watts.

[terahz]

TheFishMan65, great effort! Just want to add that small cpu heatsinks work very well for cooling (no fans required).

On a side note, this really begs for a wiki! If anyone of the mods/admins is around, can you consider /wiki so that summary posts like the above can be created into articles and referred to when needed? It will make things so much easier to find.

[bamf25]

Originally Posted by TheFishMan65

You have to look at the curve the the LED you are using. IIRC the XP-G at 3.2 volts is about 700ma. So about 27 watts.

Seems like un under usage of the eln-60' in some respects so conservative at 700mA we are using 30 watts per string of 12. Running at 1000 mA like many do for the cool whites which bring the forward voltage in the to like 3.3 to 3.4 we are then using about 35 watts? Is ther a good way to figure this, or where are you estimating this from?

[reefmusic]

Originally Posted by bamf25

Seems like un under usage of the eln-60' in some respects so conservative at 700mA we are using 30 watts per string of 12. Running at 1000 mA like many do for the cool whites which bring the forward voltage in the to like 3.3 to 3.4 we are then using about 35 watts? Is ther a good way to figure this, or where are you estimating this from?

Power is just voltage times current. Since current is 1 (amp), power just happens to equal to total voltage across the string. Just measure it. If you want more accuracy, put a 1 ohm resistor in series with the string, and measure the voltage across it - that will be your current (1volt=1amp), then multiply voltage by current. I suspect the total will be somewhere in the 40 or so watts, as LEDs do drop a bit more voltage at higher currents.