LED Fixture Build --Kcress--

[RTMA]

Very nice build - wish I had access to a cnc. Although, sorry to say, I think if you add optics, those bends may be rendered useless. Tagging along - can't wait to see the finished product.

[JW65]

That is awesome! Cant wait to see it in action! I am thinking about a DIY led build and this just gave me some good ideas for when I finally start it.

[ktg113]

I forget what Cnc stands for but basically it is a nice toy I mean tool that takes a auto cad drawing and then makes all the cuts that are in the auto cad drawing


Sent from my iPhone using Tapatalk

[jpccusa]

Originally Posted by skimate

I’m a little concerned that bending those tabs may have slightly distorted the area where the LED’s are fixed, possibly creating a tiny void under them?

Originally Posted by smithj108

Agreed!!!!!!! It looks like there's a slight bow - but it could be just camera optics in the picture.

I believe there is thermal compound under each star... that would fill the void.

[sfsuphysics]

Originally Posted by jpccusa

I believe there is thermal compound under each star... that would fill the void.

Yeah but thermal compound is meant for very tiny gaps not huge ones, of course I haven't used thermal compound in a long time maybe it actually transfers heat better than it used to

One thing to help with the bending, maybe on version 2.0 Since you have the CNC why not grind down a channel along the bend area that's 1/32 or so so the bend doesn't take as much effort? Nifty little setting you have, on my man-list for things I'd like.

Also I know by the looks of the wiring, the jumper popped out at me, are you running two strings in parallel? I'm assuming the fuse and the resistors are for that? I'd like to see the specs on those fuses/resistors when you have the chance.

[westreef]

Kcress,
I think your resistors are for the current monitoring purposes. Is it 1 ohm resistors? From the look of it, it is about a watt or less. In my LED built, I use 1 ohm 5 watts wire-wound resistors and I can still feel they are quite warm to the touch with 1 Amp going through it. Just curious.
BTW, very nice built!

[Spaced Cowboy]

Originally Posted by westreef

Kcress, sorry to ask: What is CNC?

CNC = "Computer Numeric Control". It's basically a way of controlling a device from a computer. Typically you use a CAD (Computer Aided Design) program to draw something on the computer, and then go through some process (it varies depending on what you want to do) which results in a file you can send to a CNC device (typically a router, on an X,Y,Z table, that describes where in space to move the router, and how fast the spindle ought to be rotating etc.). It's all numbers, hence the acronym.

Beware, these things are addictive. I now have 3 of them...

- A PCB "engraving" machine that starts off with a sheet of copper-clad FR4, and "rubs out" the copper, leaving behind a circuit to which you can attach components. It's accurate down to about 6/1000 of an inch - which is to say it can create "wires" only 6/1000 of an inch wide connecting pins on ICs. I use it a lot

- A traditional CNC machine made by K2 - see http://k2cnc.com. This is actually still awaiting software to use it fully. I intend to use it when I get around to making housings for my reef controller.

- A Thing-o-matic from http://makerbot.com/ which allows me to "print" in plastic, making a 3D object. This is an additive CNC machine (it deposit plastic, one layer after another, to form the model) whereas the other two are subtractive CNC machines (they route away material from a solid block, leaving behind the desired model).

If someone says "CNC machine" without any qualification, they're almost always talking about a router on an X,Y,Z table, so it's a subtractive device.

Hope that clears things up

Simon

[kcress]

What exactly are you using for your PCB machine? I'm always interested in that subject.

[westreef]

Thanks Spaced Cowboy. Very clear.

[Rybren]

kcress,

I'm curious as to why you didn't go with your DIY Driver? Did you ever end up building a rig with it?

[Spaced Cowboy]

Originally Posted by kcress

What exactly are you using for your PCB machine? I'm always interested in that subject.

It's an EverPrecision EP2002H with a controllable (60,000 rpm max) spindle. You can see the specs at http://everprecision.com/e-epseries-3.html - that's for the 2006 model, but its essentially the same device. Mine claims 4-mil accuracy as well, but on a test grid, only about 50% of the traces actually conducted at 4 mil. At 6 mil, they all did. I don't need better-than-6-mil accuracy, so I'm ok with that.

One of the reasons it repeatedly works at that resolution is that it does a surface-scan of the copper before it mills away the surface. I attach a GND-lead to the copper top, and the toolhead makes an electrical connection as soon as it touches the copper. That gives it a very sensitive measurement of height variations in the FR4, and you can see it compensating in Z as the toolhead moves across the surface when milling. Very cool.

I think the machine itself is excellent, but the service frankly sucks. The current dealer won't upgrade my software because I bought it from the previous dealer... EverPrecision themselves don't want to get involved, so basically I'm stuck with the old software. Fortunately, that software is pretty good, and I'm unlikely to need an upgrade. If the worst comes to the worst, I guess I can write something - it's all just g-code, and I am a software engineer

Simon.

[kcress]

Originally Posted by Rybren

kcress,

I'm curious as to why you didn't go with your DIY Driver? Did you ever end up building a rig with it?

Dear sir I think you have me confused with DWZM. I've built no LED driver I've shown in RC.


Spaced Cowboy; Thanks for the info. I'll PM you so we don't run this baby off-the rails any further.

[Rybren]

Originally Posted by kcress

Dear sir I think you have me confused with DWZM. I've built no LED driver I've shown in RC.

My bad.

I was thinking of the LED Lighting on a budget thread that you started here http://www.reefcentral.com/forums/sh...diy+led+driver

[kcress]

Picking up were we left off.
I want to go back a bit and show the fire-up process of a parallel string build.

Once you have everything wired up you need to check your wiring. You should really NEVER turn on a build and have parts of strings not correctly go on. That's really bad form. So the first thing you do is use a DMM with a diode test. Set it to DIODE and run down every single LED. Touch the solder blobs not the other pads. We don't care about the other pads, we only care about the ones we actually used. (Sorry for the hazy pictures the camera focused on my fingers!!)

Here I probe a blue LED in the proper direction and you can see it lights up. If it doesn't you have a problem. You could also do this with the two AA batteries method.



Next I move one probe to the chassis leaving the other in place. If I botched something that shorted the other lead to the chassis the LED will light up again - incorrectly.



Then I go back across the LED to light it up again and swap the opposite probe to the chassis to check for a short on the other side of the LED. Do this on every single LED on your build before powering up. It takes only a few minutes and will prevent any ugly, expensive, surprises.



That done we can move on to adjusting the driver. I turn the current limit allllll the way down. Sometimes they ship them all the way up. If they do this and your strings are badly balanced you could toast one. If you turn it all the way down you likely won't toast anything even with horribly mismatched strings.



Also take the time to turn the voltage limit all the way down too. This is a safety thing that can also save you in a mess up. Large currents tend to require higher voltages. Turning the voltage limit down limits the current that can occur through a secondary aspect.

[kcress]

The moment of truth!

Flame - on!

Immediately - if not sooner - measure the current thru the strings. You need to know if you are over driving a string so you can power down if you have to.

So, quickly, quickly, I measure one string.
I have 0.1 ohm resistors so I have to do some math. Notice the meter is set to show a maximum of 300mV. This means I'm seeing 43.8mV. Multiplying this by 10 because of my resistor values, tells me I have 438mA running thru the left string. This is great, not too much.



Over to the right string...
425mA

Wow. Happy days! They are very close together in current. Close enough that I don't have to do anything about it.. At this point I turn the voltage limit back up all the way. (more on that later)




This means I can just turn up the current to where I want to run at.
And so I do. I crank it up to 754mA on this side.



And that gives 740mA on the other side. No problems here.

Everything looks good. So now I turn down the voltage limit again while monitoring the current in one of the strings. I turn down the limit until I see the current start to drop. I then turn the limit back up until the drop disappears again. Then I turn it up a tiny bit more to cover temperature and age drifts. However in this particular build I could turn the voltage limit all the way down and never had the current drop! This means these string's Vf totals are so low that the driver voltage needed to reach 750mA is even below the voltage limit range. Perhaps I could've run another LED in each string..



The astute reader will note that this 1.3A driver appears to be delivering 748mA + 754mA which equals 1.5A.
How can this be?!
There are two reasons. One is that our drivers are really power drivers at their cores. Since this is only 12 LEDs they are not demanding anywhere near 48V to run at this current level. This means the driver can provide a little more current and still be below the 60W rating.
Another reason is probably the ripple. My meter reads the RMS current. That means it's looking at the peaks and averaging them. There are peaks with this driver.

What are they? Let's look. Setting my meter to AC Volts will show the AC current running in my strings.

Here we see 12mV (note this meter doesn't have a 300mV AC setting).

Back to my multiplying by 10. This means each string has about 120mA of ripple current running thru them. So my average current is about 750mA and this means I really have a cycling value of about 690mA to 810mA.

So, several hundred or thousand times a second my strings have the current varying thru them between those two values. This is well below the current limit for these LEDs and so does not concern me in the least. Ultimately this will make them run probably 1 or 2 degrees warmer than a totally smooth 750mA pure DC current would. If this build was running right at the maximum current limit the heating would be greater and might need further study. In this 750mA case - not at all.

At this point



On to checking out the light! Here's the color against a white ceiling. Note the slight blue.




Here's a mug shot of it.



Here's a shot with the camera stopped way down running a 1/4000 sec shutter. This really shows the LED colors. Neutral whites, cool whites, and royal blues.



Here it is tipped up lighting my towel.



Here it is for chuckles propped up on two books lighting my big build.

It fell off the books into the bottom of the tank and never blinked. Try that will a running MH and see what happens... LOL

[tahiriqbal]

Nice work sir..... quality DIY but still need a sharp heatsink to keep everything cool unless its a test run, love the whole idea behind custom reflectors

[widmer]

Color @ last pic looks great to me. So what's your plan? Going to make a couple more? Same color mix?

[skimate]

Absolutely fantastic!
I’m pleased to see that driver worked out.

Thanks for taking the time to show the set up process in so much detail. How to test the ripple was especially useful. I’m now not so nervous about testing the ripple monster drivers I have.

It’s so hard to tell from a photo, what do you think of the colour mix you have there?

[BeanAnimal]

Would you stop messin around and put some fish in that thing already...

[kcress]

Originally Posted by tahiriqbal

but still need a sharp heatsink to keep everything cool unless its a test run

Hi tahiriqbal. Not sure what what you're getting at here. This is it. This is the heatsink. This is what's been runnig for a week. I'm running a temp test right now and will let you know.

Originally Posted by widmer

Color @ last pic looks great to me. So what's your plan? Going to make a couple more? Same color mix?

Not sure. Of course I'd need about 5 of these for my tank and optics are also a definite requirement. But then I also have to avoid the solar beam paths. I will likely use this style for for whatever I end up doing. Perhaps a giant one with two holes for the sun? This one was sort of spec'd to be an ATS light.. But man.. I do love the color for the livingroom as does everyone who sees it.

Originally Posted by skimate

Absolutely fantastic!
I’m pleased to see that driver worked out.

Thanks for taking the time to show the set up process in so much detail. How to test the ripple was especially useful. I’m now not so nervous about testing the ripple monster drivers I have.

It’s so hard to tell from a photo, what do you think of the colour mix you have there?

I really love the color. Realllly love it. But since I have no coral I can't really speak for its coral 'popping' ability.

After seeing the ripple result here I don't think I'd worry about it further, if you're going to run around the 7-900mA region. I was really fearing it would be causing 500mA current ripple or worse. Having sub 200mA ripple is a don't-care. As a person I am exceedingly flicker annoyed. I see flicker where no one else does. There are stores I can't even go into because of their lighting. These drivers run at such a high frequency that the actual light modulation this ripple causes is completely undetectable. It's likely way too fast for any aquarium inhabitants to respond to. Go ahead and let your ripple monsters out of the closet.

Originally Posted by BeanAnimal

Would you stop messin around and put some fish in that thing already...

??? And end up with fish poop all over my nice rock? Never!! I'm switching to Terrarium Central...

[BeanAnimal]

I would just get drunk and sleep in it to reassure the wife that it will get done when it gets done

Did you steel my bench supply and meters? Do you want my variac too?

[sfsuphysics]

Looks good, just curious if its fish only why would you need more than one? Looks like it has decent cover as it is.

Also good to know about the diode tester, had no idea it would actually light the led, I simply used a wall wart hooked to a decade box (with a ton of resistance) to make sure each fired pre-wiring.

[kcress]

Originally Posted by BeanAnimal

I would just get drunk and sleep in it to reassure the wife that it will get done when it gets done

Did you steel my bench supply and meters? Do you want my variac too?

Why yes I would like your Variac too. Mine's a huge one for mounting in a stage lighting panel - that's not mounted in a panel. So I have to prop it up and be v e r y careful of where I touch it.

Originally Posted by sfsuphysics

Looks good, just curious if its fish only why would you need more than one? Looks like it has decent cover as it is.

I agree. I think it would be fine, as is, (maybe without the books holding it up ), for a FO tank.
I'm going for a coral tank this time around.

Originally Posted by sfsuphysics

Also good to know about the diode tester, had no idea it would actually light the led, I simply used a wall wart hooked to a decade box (with a ton of resistance) to make sure each fired pre-wiring.

Yeah, they all light up on Diode Test. It's nice. You have to be careful using a wallwart as these LEDs have a low reverse voltage. Resistors or not you could possibly exceed the reverse voltage and damage the LEDs. Remember that the circuit open is where all the voltage is. You could have high resistor values and you will still have the total WW voltage appear across a reversed LED.

A meter keeps the diode test voltage low enough that it won't do this. If you do use a WW you should probably try to keep it around 5V.

[smithj108]

Originally Posted by kcress

..... I see flicker where no one else does. There are stores I can't even go into because of their lighting. These drivers run at such a high frequency that the actual light modulation this ripple causes is completely undetectable. It's likely way too fast for any aquarium inhabitants to respond to. Go ahead and let your ripple monsters out of the closet.

Yes. Without a scope I would have to guess at somewhere around ~500Hz or more. Contrast that with the 60Hz you commonly see in store lights.

[nauticac4]

Out of curiosity why bend the fins down instead of up. I suppose it doesn't make much difference in heat managment, but I also dont see it impacting the light as much either. Any chance of PAR readings and some running surface temps?

Regardless the rig looks awesome can't wait to see it enclosed.