DIY LED driver for reef lighting

and reduce the Voltage to get the Current @ 650mA and hopefully LESS heat!!!

To be clear, you DON'T want to control current on the output side by reducing the voltage! You want to control it by picking the right sense resistor.

To hash back through what I explained above - if you start out with an input voltage too low, the chip won't actually be regulating anything, it'll just pass that voltage through to the LED string. As you increase input voltage, suddenly you hit the minimum point for the chip to actually regulate, and it takes over. When you're below that point, the driver circuit is doing nothing but burning off a tiny bit of power. You want to be in the operational range for the chip (i.e. at or above the point at which it starts working) but as low as possible in that range, to reduce the amount of voltage it has to burn off.
 
cowboy, after how long of operation did you get the smoke? and what was the wattage of the resistor that you had used? what if, we use higher wattage resistor, say 5Watts

and is it alright to touch the resistor when in operation? isint 48V the border of danger?
 
perikaruppan,

That power supply is capable of driving 8 stringa at 650ma without a problem(7.3/.65 = 11.23 20% safetey 8 maybe 9). So if you are like me and have extra drivers you can add LED any way you want. More on a string is better effiecency, but if you need more than 7 LEDs add a string.
 
oh ok. As the specs the formula for output current = 0.2/Rs, Rs=0.33, so it is 606mA.
Say in my current scenario, can i do like this, adjust the input voltage till i get 600mA to match the perfect current? and check out the heat? will this be alright.. or a quicker way to do.

how will i know whether the IC is regulating the input or not? or how will i know that the IC is working? at which point?
 
fishman, but in the XLS sheet it indicates that the current is not CONSTANT, it varies between 2 values, and if i choose a 600ma average current, then the max current is around 900mA... shouldnt i calculate the string based on the max current?

that is only 8 strings!!!

I was thinking of adding another string to get 12 LEDS to light up my sump!!!... or increase the count in each string by 1, to get 7 Extra LEDS to light up the sump...

What DIY driver are you using fishman?
 
Depends on which of the three driver circuits you're asking about. The original NCP chip, in the mode I used it (boost) had a very big cutoff, worse than the meanwells. In a buck configuration I'd guess it would perform better, but I never tested it.

After reading your big post on the first page, it sounds like buck or sepic is the way I'd want to go. Is it even possible to have no cutoff at all? hope i'm not chasing something that doesn't exist.
 
cowboy, after how long of operation did you get the smoke? and what was the wattage of the resistor that you had used? what if, we use higher wattage resistor, say 5Watts

and is it alright to touch the resistor when in operation? isint 48V the border of danger?

I was using a 1W resistor, and I wasn't touching the contacts of the resistor, just the top of it. Gently. And doing a 'touch'/'remove finger' very quickly :)

As for danger, I think ~80v is considered dangerous. I was only running at 24... The main thing is the current though, not the voltage, and whether the electricity flows across your heart. I've been electrocuted in the UK (which is 250V and 13A) without getting anything more than a buzz in my fingers, but it's not something I'd like to repeat :)

Having said that, http://www.ednasia.com/article-7565-hazardousvoltageprimer-Asia.html would tend to agree with you :)

Simon.
 
You need a better analog engineer than me. I also forgot this is not a true constant current device. The 900ma is max through the LED not necessarily from the power supply (capacitor can help here). I have no idea the waveform of the current draw from the power supply. I would think that there is no way all the drivers can be in sync (all drawing 900ma at the same moment). So I think you could safely add 1 string.

Anyone else's thoughts?
 
brought down the voltage 43V to get 600mA and increased the voltage 43.5V as fishman said to maintain.

But after 5 min of operation, there is heat, on the IC that is bearable. So what can be done for the heat elimination?

The IC is too tiny, have a fan blow over it regularly? or Switch over to CAT?
 
After reading your big post on the first page, it sounds like buck or sepic is the way I'd want to go. Is it even possible to have no cutoff at all? hope i'm not chasing something that doesn't exist.

I wouldn't do a SEPIC driver for fishtank use - they're mostly good for when your input voltage is actually going to be all over the place, i.e. in an automotive setting. If your input voltage is stable, known, and well-regulated (it will be all three of those things in our case) then you should just pick buck or boost based on the relationship between input and output voltage.

Is it possible to have no cutoff? Probably, but IMHO it's murky territory, since you're essentially strobing the LEDs with a very low duty cycle at that point. I am pretty confident that you'd be able to build something that would meet your needs in terms of starting your fade from a very low brightness.
 
thanks simon.

DWZM, so @ 600ma there is min heat, and when i increase the V, the current increases to 650ma havent tried further than that.. so if it stabilises at 650ma then that is the current? but at that current there is huge heat generated...
 
If you have really stable powe over there you cn probably get away with 43 volts output. The problem is that if it drops below 43 volts the LEDs don't get the right current. It may also damage the chip if run long term with too little power - can anyone verify/negate this?

After 10 minutes, 15 .. is is still bearable. And what if you fully populate the board. Do you have more than one driver per board? If only one let it run and see if it is still bearable.

you could also try 43.25 volts (or 43.33). Pretty much whatever you are omfortable wihth.
 
DWZM, which driver IC would you choose to be the safest of safest bets, with regards to safety and performance is concerned. If not too much bothered about the costs?

will it be the NCP3066? or CAT? as ZXLD there is heat.. do these, the NCP and CAT also get heated up?
 
brought down the voltage 43V to get 600mA and increased the voltage 43.5V as fishman said to maintain.

But after 5 min of operation, there is heat, on the IC that is bearable. So what can be done for the heat elimination?

The IC is too tiny, have a fan blow over it regularly? or Switch over to CAT?

You'll have SOME heat no matter what, really - it's no big deal. If you can touch the chip and it doesn't burn you, and/or it's not shutting itself off, you're probably OK.
 
DWZM, which driver IC would you choose to be the safest of safest bets, with regards to safety and performance is concerned. If not too much bothered about the costs?

will it be the NCP3066? or CAT? as ZXLD there is heat.. do these, the NCP and CAT also get heated up?

It's really hard to pick a "best" chip. I think any of the three will work fine. If I were you, I wouldn't switch at this point just because it's getting a little warm. I'd set one up, run it for a week or two, and evaluate at that point. If it works stably for a week, you're probably fine.
 
fishman, there are 7 DRIVERS in a single board of 100mm X 100mm.. maybe i should revise the board.. the drivers are spaced within 30mm radius from each other and that is pretty CLOSE..

I havent touched the IC when in operation (as simon said touch the top) as the IC is tiny so dont want to risk that.. i switch off and immediately touch the IC to determine the heat.. so may be will make it run till it turns OFF by itself of the resistor/cap sends smoke!!!
 
thanks dwzm.. maybe i will run it overnight now (now its midnight here) and see whether it turns ON or OFF by itself or does it blow..

and as cowboy mentioned, if the Resistor or IC blows, will it damage the LEDS?
 
Back
Top