DIY LED driver for reef lighting

[der_wille_zur_macht]

Posting at the same time again.

My only thought on Rs - if you DO pick a value to give a higher Iout with the intention of dimming, make sure your peak current with the driver 100% on is not outside the limits for any of the external components (I'm thinking primarily of the inductor and LEDs here). If you build for 1A Iout and run at 50% duty cycle, your components are still seeing whatever the peak value for 1A operation is (it's gonna be more than one amp) even though your "effective" Iout is only 500mA.

 

[hkfoot]

Thanks der_wille and FishMan. I learn a lot from reading your posts. I definitely will try to build one and report it here. As my skill is very limited so it will take time. If you want to try this chip, I can send some to you once I got the chip.

 

[terahz]

Indeed a nice chip. According to findchips.com this is nowhere to be found in the US. Maybe hkfoot can send us a bag at that price to play around with ?

hkfoot, the application circuit in the datasheet should get you started as the others have stated.It shouldn't be too hard.

 

[TheFishMan65]

DWZM, didn't you learn form the sketch thread? Sorry, I am getting caught up on my reading and you make so few error that when you do they stand out
.1 / .12 = .833
.1 / .14 = .714
.1 / .142857 = 0.7000007000007000007000007000007

 

[der_wille_zur_macht]

Drat. That was a typo. I meant .14. I swear!

 

[terahz]

Drat. That was a typo. I meant .14. I swear!

:lol2::lolspin::lol2:

 

[der_wille_zur_macht]

You guys have to forgive me. I'm used to projects with a handful of QA analysts checking everything I spew. This whole concept of actually being personally responsible for the accuracy of your technical work products is new. :lol:

 

[hkfoot]

I found a driver kit like this so I am going to order some and see if I can build one. Is it difficult to solder this board? I have no SMT soldering experience :worried:

 

[TheFishMan65]

That looks like a great option - depending on the price. All i got was a picture so a little hard to tell. Don't suppose you have a link to a better descriptions (list of materials) - in english

 

[der_wille_zur_macht]

Let's see - I can spot two diodes (two? maybe one is for reverse polarity protection on the input?). Two caps, look like tantalums. Means they must have one on the output. And a 47uH inductor, and the IC in question of course. Two sense resistors - I wonder if the kit includes instructions on using one/both to get different drive currents?

It doesn't look terribly hard to solder, but if this is your first SMT soldering experience, I'd probably buy two of the kits in case you mess something up first. I'd solder the components smallest to largest unless there would be some clear interference. For the caps, diodes, and inductor, tin one of the two pads on the PCB, then hold the component in place with tweezers or a pick and touch the iron to the tinned pad to melt the solder. It'll bond the part to the board. Then go solder the other pad. For the IC, I'd do the same thing, but start by tinning the ground pad on the PCB and getting that well soldered, then solder the other pins down.

It looks like a kit for a flashlight or something?

 

[hkfoot]

The driver kit only cost 1.5 USD, I am going to order 12 (actually I only need to build 6). The resistors provided in the kit is 0.27 ohm each, the board can solder upto 3 resistor in parallel, so for 1 resistor it is ~370ma, for 2 resistor the current will be 740ma.

I failed to find a supplier in Hong Kong, so ordering it online from China is my only option. Still it is dead cheap even including the shipping charge.

 

[der_wille_zur_macht]

Any specs on the other components in the kit? i.e. max voltage/current ratings? What voltage is the kit designed to work on? Given the inductor size I'm guessing it's about right for 24v and 6 LEDs but unless you know the ratings on all those components it would make me nervous. Would also be nice to know what type of diodes and the capacitances on those (tant?) caps.

Also, is the DIM pin brought out? I don't see it in that photo. Would be a shame if it wasn't brought out or if it was tied to Vin or something like that.

$1.50 USD is an insane price. Low enough to make me nervous. :lol:

 

[hkfoot]

All it said is the board can drive 6 x 3W LED, 8-24V, 350-1000ma. I will ask the supplier for more info. when I place the order. I think it would be better to build your own board so you can use better components.

 

[5ft24]

Curious what the cost is for a PS, and the board/parts for the regulator...
I was looking at the meanwells and can get the 700ma meanwells that will drive 13 LED's each for $16.95usd I know they don't do PWM dimming, but the output could be run thru a power MOSFET and driven with a PWM Signal for the dimming...

 

[der_wille_zur_macht]

Curious what the cost is for a PS, and the board/parts for the regulator...

Check through the thread, we've done a handful of cost calculations. I find it useful to speak in terms of cost per driven LED. The CAT4101 design can put you around $1 - $1.50 per driven LED (off the top of my head).

the output could be run thru a power MOSFET and driven with a PWM Signal for the dimming...

I'd really want to see what that did to the meanwell's ability to regulate current before assuming it was OK. Opening the circuit (i.e. when the MOSFET was off) on the output of a constant current driver like the Meanwells results in the driver going "wide open" in terms of output voltage. When the MOSFET switched back on and connected the LEDs, they would instantaneously be subject to the full output voltage. I doubt the Meanwell could react fast enough to prevent disaster or damage.

 

[5ft24]

Check through the thread, we've done a handful of cost calculations. I find it useful to speak in terms of cost per driven LED. The CAT4101 design can put you around $1 - $1.50 per driven LED (off the top of my head).



I'd really want to see what that did to the meanwell's ability to regulate current before assuming it was OK. Opening the circuit (i.e. when the MOSFET was off) on the output of a constant current driver like the Meanwells results in the driver going "wide open" in terms of output voltage. When the MOSFET switched back on and connected the LEDs, they would instantaneously be subject to the full output voltage. I doubt the Meanwell could react fast enough to prevent disaster or damage.

Good point, never thought of that...

 

[blasterman789]

When the MOSFET switched back on and connected the LEDs, they would instantaneously be subject to the full output voltage. I doubt the Meanwell could react fast enough to prevent disaster or damage.

Side-bar, but don't they ALL react this way? I've had the same problem with Xitaniums, Lightech, etc. kicking out a voltage pulse under the same conditions.

 

[der_wille_zur_macht]

It would be safest to assume so, though I would not be shocked to find an off the shelf driver that included some sort of open output circuit shutdown feature. If you tried to use a MOSFET to switch the output of a driver with such a feature, you'd basically be slamming off that shutdown each time the MOSFET opened the circuit. This may be safer (or maybe not, depending on how fast the open circuit shutdown reacted), but would still make me nervous purely from a stability standpoint.

I know that some LED driver ICs include provisions for implementing such a feature, but I have yet to see a commercially produced driver that implements it. Or, at least, that advertises it - I haven't made a regular practice of opening the outputs of live LED drivers to find out. Meanwhile, in the case of the Meanwell ELN drivers, we can say for certain that this feature does NOT exist, as there have been reports of people frying LEDs by connecting the output-side wiring while the driver was live.

 

[akindbro4u]

just to get this back out of archives, and see if anything has changed. BUMP

 

[theatrus]

Has anyone played around with the LM3404 for LED driving? Its a buck-regulator and has built-in PWM control, plus has a VinMax of 40V (longer chains of LEDs). While the CAT4501 sounds nice, I'd much prefer a switching controller for efficiency.