DIY LED driver for reef lighting

I've gathered some of the material I've been promising for weeks (months?) in the various LED threads and figured if I make a thread about it, I'll be motivated to get the rest of the material together if there's enough interest.

If you're new to HP LEDs for reef lighting, here's some reading:

The Mother Thread:

http://reefcentral.com/forums/showthread.php?t=1678127

Some others:

http://reefcentral.com/forums/showthread.php?t=1662682

http://reefcentral.com/forums/showthread.php?t=1751598

http://reefcentral.com/forums/showthread.php?t=1672708

I'm starting this thread to detail the DIY driver I've mentioned a few times in those various threads. I decided to DIY drivers for a few reasons:

1) It seemed like an interesting project for someone like me with moderate DIY electronics skills to learn some more and expand my skillset
2) It gave me control over the end product I was getting, vs. living with the limitations of a COTS product.
3) It'll end up being a few bucks per driver cheaper than any of the commercial products that compare, though that's NOT accounting for time involved.

If you want a fun DIY project involving LEDs, read on. I'll be doing this thread in sections as I get time to document this project. Also, for all you "real" EE's out there, if you'd like to rip apart my design, please do so. It'll help me learn. :D
 
Before I get to the actual design, I want to explain how I arrived at it, so if anyone else is interested in DIY'ing drivers they'll understand why I made the decisions I did.

Design goals:

I wanted the following from a driver. These were the goals that led to this design.

1) Highest efficiency reasonable. At least as high as commercial products like the buckpuck drivers, or meanwells.
2) Controllability. Ideally, a 5v PWM signal in the ~400hz range (since that's what an Arduino can generate) would work.
3) 6-12 LEDs per driver. Any less, and you've got a billion drivers for a reasonable sized fixture. Any more, and your resolution for controllability drops off.
4) All through-hole soldering. I have no reflow oven and (at least at the time I started this design) don't/didn't trust my SMT soldering skills enough to go that route.
5) A product that was designed for constant current LED driving. I don't trust myself well enough to work voodoo magic on a fan-driving chip to get it to work, for instance.

I looked at chips from TI, OnSemi, National Semiconductor, ST, Maxim, etc. (hooray for free samples if you have a business-like sounding name/address!) and the above criteria cut the field down to a VERY small number of choices. I chose the NCP3066 from OnSemi in the end, though there are arguably a few other chips that would probably work just as well.

The NCP3066 is a switching regulator (high efficiency possible compare to linear regs) that's flexible and can run in a boost, buck, or SEPIC topology. It can use an external switch for high currents, but I'm keeping my drive current reasonable (700mA is the MAX I've done on this design, though it should be good for more than that if you watch thermal management.) I'm using it in a boost topology, since according to OnSemi, that's a little more efficient. It also lets you use more LEDs per driver on a common and cheap 24v power supply compared to buck topology.

Here's the product page for the NCP3066:

http://www.onsemi.com/PowerSolutions/product.do?id=NCP3066

There's also the NCP3065 - a pin-compatible version that's identical (and interchangable - I've swapped the two on built drivers) EXCEPT it doesn't have the EN pin used for PWM dimming:

http://www.onsemi.com/PowerSolutions/product.do?id=NCP3065

Next up, schematics and BOMs. . . .
 
Subscribed! Looking forward to this as I'm in the early stages of making my own DIY driver like I mentioned in the mother thread and this was the chip I decided to go with as well.
 
So sadly I don't have a schematic of the EXACT configuration I'm using, but it's very close to some of the schematics in the product docs, so hopefully that's good enough. :D If there's enough interest I can work on a real schematic of the actual design. Here's a schematic lifted from the app notes of the NCP3065:

ncp3066.gif


This is pretty much my circuit, with a few changes:

1) I'm using a SINGLE resistor in place of the 6 parallel resistors shown for R1-R7 in the upper left corner. As long as the final resistance is correct, you can use any combo of series/parallel resistors you want, of course.
2) I'm NOT using the dimming junk connected to the COMP pin in the lower left corner. Since I used the NCP3066 version of the chip (not the 3065 shown here) I can dim on the EN pin, so I don't need to dim on the COMP pin. What this means is you can leave off everything to the left of the NU (not used) resistor near the bottom left corner of the schematic.
3) Since I am using the 3066 version with the EN pin, the "NC" (no connection) pin shown on the upper left corner of the chip is really an EN (enable) pin. I have a 1k resistor connected to that, and I supply my PWM signal to that 1k resistor.

Now, on to parts lists. Here is the list given for that schematic in the docs:

parts.gif


Of course, these are mostly incorrect. :D They're all surface mount, and the values are wrong: this is for a 12v input 350mA driver. I want a 24v, 500mA driver. If you want something different, it usually just means one or two component changes, so stay tuned for discussion on that (if there's interest).

Up next, components I actually used. . .
 
Parts I used:

parts_2.gif


I included digikey numbers for the parts I got from them. Besides this list, I used some things I had from other sources:

1) I couldn't find a good resistor for R1 from digikey at a reasonable price. R1 is the current limiting resistor (for the entire circuit, it does NOT independently set the drive current! It basically sets the max spike the circuit can hit when trying to generate the drive current, which is controlled by Rsense.) The circuit wants .15ohm and a half-watt rating. I got a HUGE bag of .22 ohm 1w resistors from eBay (100 pcs I think) for about $6 shipped, which is what digikey wanted for only a handful of comparable parts.

2) I also struggled to find an exact match for Rsense, so I got the .47ohm resistors listed in the above BOM and used one of them in parallel with one of the .22ohm resistors I mentioned above for an Rsense value of .69ohm. This gives a drive current of around 350mA (very tame by today's standards.) You can JUST use the .47ohm resistor from digikey if you want 500mA drive current, or you can use a .34ohm resistor if you want 700mA. The 350mA and 500mA drive currents work totally fine with the rest of the components listed, BUT if you want 700mA or above you'll need to adjust other values to compensate, since the peak current on the chip will be exceeded unless you do.

3) 1k ohm resistors (for R8 and Rpwm) are a dime a dozen. If you don't have them laying around, shame on you. :D

4) I used an 8-pin DIP socket to build this, rather than soldering the drivers in. I did this to protect the drivers, and so I could switch between 3066 and 3065 chips as I tested (disconnect Rpwm if you use 3065). If you trust your soldering skills, you can just solder the chip in.

5) For layout, I don't have any diagrams. I used this protoboard from ratshack:

http://www.radioshack.com/product/index.jsp?productId=2103799

It's $2 for two of them, and they're designed exactly for little DIP-IC based circuits like this. Probably not ideal (I can hear people screaming about trace widths and ground planes now) but it works which is good enough for me. If you want to save a few bucks and you're buying your DC power supplies from mpja.com, they have this part at 75 cents for two, which brings cost down.

That should be enough to get people started. Lemme know if there are specific questions. In the next installment (maybe this afternoon, I need to get some real work done) I'll post some action photos and info about actually using these drivers.
 
Just curious, did you look at the STCS1 driver chip?

I does not meet one of your requirements - it is only available in a surface mount chip.
However you shouldnt worry about that too much.

The very first time I tried to solder surface mount, I did just fine.
It can be done with a standard soldering iron, you dont need an oven or reflow station.
You can also get "surfboards" that allow a surface mount part to be placed into a DIP socket.

Here is a pic of the schematic:

STCS1_schematic.jpg


Notice how there are far fewer components on this one - ONLY 5.
Also - NO Inductors required.

Stu
 
Stu, I'm thinking about doing some prototypes with that chip right now. At least one or two other forum members are already using that chip. I avoided it in the first round because of the lack of a through-hole package, plus it's (likely) not as efficient since it's a linear reg. However, matched very carefully to a PS so the voltage was close to what the LEDs need I'm sure it would be close.

Those chips have a ground pad for thermal management under the chip, though. I suppose you could use conductive epoxy to glue that pad down, then solder the actual connections?
 
"Those chips have a ground pad for thermal management under the chip, though. I suppose you could use conductive epoxy to glue that pad down, then solder the actual connections?"

Ahh.... I didn't notice that. You're right.

The epoxy might work fine since we would be running close to half the rating.

Otherwise we would need an oven & solder paste.

Stu
 
great....ANOTHER thread for me to follow for my impending build sometime next year (fingers crossed LOL)
 
i'm gonna let all you guys work the bugs out while in basic & ait & then when i get settled somewhere, poach all the info & start on mine :lol:

thats only 13 days away..." james"....get busy!!!!!:lol:
i said "sometime" next year....not new years :p
 
good job !

I am interested about this thread too. If you need me to do the PCB no problem ... I have everything at my job to do it. I am electronic engineer so I can help you if you want.

I will be off line during the holiday (going to France ... I am French) but will come back the 28th december
 
Thanks for the offer. Whatever drivers I end up using on my big tank will most likely be on a "real" PCB but IMHO there's value to building these things on the super-cheap protoboards if you don't want the hassle of making your own PCBs or the expense of going to a board house. Plus I think it presents an easier entry-point for people who aren't complete experts at DIY electronics (like me!)
 
Great project. I've already started planning a bigger tank and I'd definitely do the drivers as well this time. For my 10g it wasn't worth the effort but for a bigger tank I'm sure this will work quite nicely.
 
Awesome! I'll be ordering my parts from digikey tonight!!! I need to build about 15 of these for my planned hood!
Thanks for posting the digikey #'s .... it will sure make it easier than trying to figure out what's what!

I would be interested in seeing a final schematic and possibly the board layout you used...a couple pics would do that nicely!
 
In point 2 you mentioned different values for rsense for the drive current. So does this mean that if you went with the .47 ohm for 500mA then the maximum you can drive the LED's is 500mA? Just want to make sure I understand exactly what this means. For my build I would want to be able to go to 700mA in order to ensure I had the ability to get light to the bottom of my tank but still not max out the LED's. What other values would have to be adjusted to drive at 700mA?
 
Back
Top