DIY LEDs - The write-up

[rcgates]

Technically, in order for one or more LEDs to stop working, you need MULTIPLE shorts to the heatsink (or any other otherwise-isolated conductor not intended to be part of the circuit). It's wise to find and eliminate all of them - don't stop just because your fixture starts working, otherwise you may end up in a situation where your heatsink is live, which could be dangerous.

goes without saying, but worth reminding all ;-)

 

[TheFishMan65]

Grim, earlier IIRC kcress said either don't or it wasn't needed on the coating. Can you shed any more light other than your opinion. I know someone said they placed the LED in salt water at a show and once again IIRC it lasted for a day. Thanks

 

[der_wille_zur_macht]

I'm with kcress regarding the conformal coating. Regardless of wether or not it "waterproofs" the LEDs or cuts down on intensity, I just don't see the point. If a fixture is low enough that splashes and spritzes are a problem (under 18" or so, IMHO) then, personally, I would still want a splash shield - a bit of water hitting an LED is going to cause damage even with a coating, via thermal stress and leaving behind salt spray (which I would MUCH rather clean off a flat shield you can quickly wipe down vs. a bunch of delicate LEDs on a heatsink). If the fixture is high enough that splashes and spritzes aren't a problem, then you don't need the coating either! The only exposed metals on these builds are aluminum, tin, lead, silver, and gold, which are all fairly corrosion resistant. And condensation shouldn't be an issue, as the heatsink will (typicallly ) be hotter than the surrounding air anyways.

That's my take, at any rate. People are free to disagree!

 

[The Grim Reefer]

Grim, earlier IIRC kcress said either don't or it wasn't needed on the coating. Can you shed any more light other than your opinion. I know someone said they placed the LED in salt water at a show and once again IIRC it lasted for a day. Thanks

I did it because mine are about 6" above the water in a canopy and I was concerned about corrosion, even with the splash shield. When I sprayed mine I placed a straw over the LED domes. Didn't want anything coating them. We'll see how it works. My tank doesn't really spit a lot of water so splashing is the only real concern with any great amount of water hitting the LED's. If it all goes to hell in a couple years it will give me an excuse to do it all again.

 

[kcress]

Thanks for the extra information!

You are now getting some great suggestions above for troubleshooting.

So I now have the following.
16 gauge wires from my driver to the terminals, then the fuses, then resistors and then 24 gauge from the terminals to the LED's. They are now only around 5' from the terminals to LED's.

Excellent!

When I had the drivers wires disconnected from the terminals I could get a solid reading of 42v ( dead on ). When I tried to hook the multimeter to the same spot with the wires connected to the terminals I did not get a reading? Should I try an test it on the resistor and see what I get? Am I testing it on the wrong spot?

What this looks like from here is that you have a dead short in your array. So the driver is protecting itself by not driving.

Disconnect the two strings in question which is easy to do. (A nice reward for terminal blocks)

AGAIN make sure you have the HLG's current pot turned down because now you have 2 fewer strings.

Power it up and see if removing the two questionable strings helps.

If this doesn't do it. Power down and disconnect another string. Power up.

Doesn't fix it? POWER DOWN. Reconnect that string and disconnect the next one. Power up.

Step and repeat for each remaining string. DO NOT FAIL TO POWER DOWN each time. Hooking up to a powered driver could blast a lot of LEDs.

I am now keeping the adjusting wires disconnected. I found that I need a much more adjustable pot to get it to work correctly. The data sheet it says keeping the ADJ wires disconnected it is 100% and shorting them is ~50%. This still had no effect on getting them to fire.

I'd short them for testing. 50% is more than adequate for testing and troubleshooting.

After trying multiple things I went back to test the strings of LED's with a Meanwell LPC-700. I two strings with one LED that would not fire and the rest were fine. The odd thing is the series of LED's will still fire even with the one not working. I thought if one LED in the series did not work they all would not work.
First string : D = Dead G = Good
D-G-G-G-G-G-G-G-G-G-G
Second String :
G-D-G-G-G-G-G-G-G-G-G
Could the two LED's that wont fire change anything? Of course I'll be swapping them out.

I'd side step this issue by disconnecting these strings temporarily.

Un-powered you can use an ohm meter as previously suggested to measure from each star pad to the heatsinking. You should have infinite resistance in every case on those dark LEDs.

If it came down to it is there a problem running a high number of LPC drivers like there is with the ELN? I need to finish this VERY soon and if I have to I may have to run a high number of LPC's ONLY if it is safe.

This solution doesn't even compute for me.. You have other problems not the driver. The driver is protecting itself from those problems.

At this point I believe you have the two driver output leads shorted somehow. This is very likely occurring at the terminal blocks. You may have jumpered them wrong or what often happens is you have a flying lead touching something it shouldn't. Follow each driver wire to the terminal block and then follow each jumper down the blocks. Look for anything questionable.

If you see nothing start the process of elimination. With the power off, disconnect ALL the fuses so the driver just powers the + terminal block. Power up and measure from it to the - terminal block. You should see your 42V just like you see it with no connection at all.

If you don't see the 42V disconnect all the strings from the - block. Measuring across the two driver blocks you should now see 42V. If you don't, you need to understand why.

Come back and tell us if you don't see 42V at the terminal blocks at this point if all the previous tips don't get it done.

cant thank you guys enough.
-Dave

:thumbsup:

 

[DFason]

Ok great. I am going to be a busy man tomorrow.

I just got off of work so I will be waking up nice and early to try everything. First I will disconnect the two strings and try it that way. After that I will look for shorts. Fun Fun!

I was hoping to start on my paludarium this weekend too!

Again I cant thank you guys for all your input. I've learned so much from all of you guys.

-Dave

 

[cjones1344]

So I'm in the very early stages of planning my LED setup, going with 2 fixtures with 24 LEDs each, driven by 4 ELN 60-48-D's. I know they aren't the newest model, but I think I'm going to go with all XR-E's, 12 CW and 12 RB on each fixture. This is mainly because a 50/50 mix will be easier, since each driver will be driving all the same LED's and I won't have to mess with all the complicated measuring stuff. Is this a bad idea? Should I just suck it up and go with something newer like XP-G's?

My real question pertains to wiring the LED's, though. I saw this diagram for wiring two strings of 6 in parallel on each driver, but I don't understand the wiring on the terminal blocks.

Is this wiring only there because they are 10 terminal blocks? Can I use smaller blocks or is that wiring actually doing something?

Excuse the ignorance, I really have very little idea what I'm doing so I'm trying to get the wiring down before I consider buying anything.

 

[kcress]

Hi Jones.

Terminal blocks are only electrically connected to the single other screw across the block. Here the left screw is internally electrically connected to the neighboring screw on the right.

So if you want your power supply to feed more than just that ONE MATING screw you have to do something else. In this case that something else is jumpering from the next screw up - over and over and over.

Of course you only need to jumper to the screws you want. Of course you can have as few as you want depending on your current driver's abilities.

Use whatever size terminal blocks you want. That is only a guideline.

Keep in mind that some terminal blocks can be pretty small and the fuses and resistors AND your fingers need some room to work with. So if you need only five parallel strings you might want to blow the extra measly dollar to buy 10 long terminal blocks so you can skip every other screw to give you some room.

 

[cjones1344]

Thanks, that cleared it up for me. I don't think using smaller terminal blocks should be a problem for me, though, since I will only have two strings on each block. From what I understand, I can get away with driving 24 XR-E's on one ELN 60-48D, but since there will be 12 RB and 12 CW on each fixture, in order to be able to dim them separately I will have to use 2 drivers per fixture. Lol every new page I read on diy LED fixtures makes the task seem more daunting.

 

[cjones1344]

I noticed this on the reefledlights website:

This is how reefledlights says to run XR-E's on ELN 60-48 drivers. I understand that running two series splits the amps in half, giving each one 650mA, but how does the voltage work out? Is your total voltage the total for each series or for both added together? (I assume its the voltage for each since the driver is only 48V) What I'm getting at is I'd like to run 16 XR-E's in two series of 8 on each driver, but I don't understand how wiring series in parallel affects voltage.

 

[kcress]

That is fundamentally what you need to run.

As drawn, that could work well or be a disaster depending on the luck of the draw.

So yes, do it right and add the fuses and the 1 ohm resistors and plan to check the currents in the two parallel strings once you have it running. Also plan to have to disconnect a few LEDs and move them to the opposite strings and vis a verse to properly balance the strings.

You might want to re-think this and just mix the two colors as depicted above. That will save a bunch of complexity and keep the LEDs homogenized.

If you know what color temperature you want. 8k 10k 15k 20k 500k we can advise on the mix and you could NOT have all the drivers.

 

[widmer]

Parallel strings is such a trendy question, maybe one of you guys wants to put together a comprehensive post about it (ie pros/cons, how to etc) that we can just point people to?

 

[cjones1344]

I'm looking for a 15k color, I was only planning to do one color per driver so I could have more control for dawn/dusk. So I can do 8 LEDs in 2 series on each driver? That will let me get a few extra LEDs in each fixture, since I was worried about 24 being enough. Since y'all are saying I should mix colors anyway, what do you recomend for 36 led fixtures? I'm going to have two of the fixtures over a 75 gallon and was only planning on going all xr-e because it would be less complicated. If anyone can recommend a combo that uses newer LEDs like xp-gs that will be a 14-15k look I'd appreciate it.

 

[rcgates]

Parallel strings is such a trendy question, maybe one of you guys wants to put together a comprehensive post about it (ie pros/cons, how to etc) that we can just point people to?

Parallel strings done right: http://www.ledsmagazine.com/features/6/2/2

The resistor only method helps, but isn't the most optimal.

look at the final method described, "Over-current protection", a few extra parts, but cheaper than blowing LEDs

 

[cjones1344]

Thanks that article was really interesting, I definitely plan on doin my build the correct way. Now I just have figure out what these extra parts would be called on mcmaster or mouser

 

[kcress]

Parallel strings done right: http://www.ledsmagazine.com/features/6/2/2

The resistor only method helps, but isn't the most optimal.
look at the final method described, "Over-current protection", a few extra parts, but cheaper than blowing LEDs

Depends on your definition of optimal. If finding a bunch of extra parts that in no way present themselves as being easily usable is optimal - then that is definitely optimal.
.

How do you propose this for a non-technical user? It's not trivial to select, mount, hook-up, test, and troubleshoot a current mirror with over-current protection! How does one even mechanically protect it?

How does one even measure the string currents in that design?

I know of no commercial designs using that scheme.

 

[wld1783]

That is fundamentally what you need to run.

As drawn, that could work well or be a disaster depending on the luck of the draw.

So yes, do it right and add the fuses and the 1 ohm resistors and plan to check the currents in the two parallel strings once you have it running. Also plan to have to disconnect a few LEDs and move them to the opposite strings and vis a verse to properly balance the strings.

You might want to re-think this and just mix the two colors as depicted above. That will save a bunch of complexity and keep the LEDs homogenized.

If you know what color temperature you want. 8k 10k 15k 20k 500k we can advise on the mix and you could NOT have all the drivers.

+1x

If you do mix leds in a parallel string they must be balanced. Mixing Blue and White is ok but only if they each are from the same lot and evenly balanced.

Its best to keep each parallel string with the same LED from the same lot. Especially if you drive them hard.

On a side note I was experimenting with a 1 Ohm resistor based on the recommendation of kcress and measured the output of my ELN 60-48D at 1.8A from the source. This is bad if your just hooking it up to XP-Gs and expecting 1.3A max...Always check your driver output...

As far as the 1 Ohm resistor it looks like a trip to Home Depot for another meter to do more measurements. A single meter just wont do it as you need to measure the current through both strings at the same time. Will post the results with and without resistors in the next couple days. So far no harm in adding a 1 ohm resistor or fuses all works well as expected.

I'm 100% positive a disclaimer will be posted tomorrow recognizing that this is an artist rendition simplifying parallel string wiring and does not endorse the use of different LEDs while using parallel strings...

Bill

 

[cjones1344]

So I can run 16 LEDs on one ELN 60-48 as long as I put them in two parallel series of 8? I will still prob be going with 50/50 all xr-es but I really just need to know if I can run that extra 4 per driver.

 

[kcress]

As far as the 1 Ohm resistor it looks like a trip to Home Depot for another meter to do more measurements. A single meter just wont do it as you need to measure the current through both strings at the same time.

Bill

Bill. Why? You don't need two meters. In fact it's sort of a bad idea. If you have resistors in each string just measure the voltage across the resistors.

If they're 1 ohm the voltage you measure is the current.

That's the whole point of adding the resistors, to allow instant current checks of the various strings.

cjones1344; If you're looking for 15k and you're not using XGs for the whites I think you want about a B:W ration of 60:40.

Hopefully someone will confirm that.

If you understand paralleling and do it as described in a dozen places here you can run two strings of 12 or 13 LEDs for a total of 24 or 26 on a single ELN60-48.

If you want to use xpg's and xre's you need about 35% white 65% blue.

 

[rguyler]

The boss (wife) finally gave me the green light to set our 90G (48"x18" surface area) display up for reef, which I interpret to mean "go ahead and put LED lighting on it".

Is the general consensus right now to go with the Cree XP-G for the white and XP-E RBs for the blue? I'm questioning the benefit of going with the higher-priced units instead of staying with the XR-Es albeit the XP-Gs provide a few more lumens.