DIY LED driver for reef lighting

I've got my 5x cat4101 driver board running and it's not producing any heat. Currently running 5 luxeon cool whites (3.4v) @ 700ma. Is that a product of running lower vin (5v)? My overnight test resulted in the die running at 81.5F with an ambient air temp of 79.5F.
 
biggs, I think we're getting some confusion since the circuit essentially requires two voltages to operate. On the Vin pin, it requires 5 volts. This is the IC's operating supply that it uses to power it's internal circuits. This supply should have very very little current flowing on it. The other supply is the +LED voltage, which is applied to the positive end of your LED string. The LEDs drop "most" of that voltage and the remainder is applied to the LED pin. The IC must dissipate this entire voltage at the LED's current, which is where the potential for heat and waste comes from.

So, the difference between our circuits and yours might be in the voltage "left over" after the LEDs have dropped your supply voltage, i.e. the voltage sinked at the LED pin.

Can you tell us your input voltage for the LED string? i.e. your supply voltage? Also, the total Vf for your LED string (guessing it's 3.4 * 5 = 17v)? The difference between those two voltages is what your chip needs to drop.

Can you verify with a known-good multimeter that your LEDs are operating at 700mA?

To operate, the IC needs to be dropping at least .5v on the LED pin. At 700mA, that's .35w, so not very much at all, but I'd still expect it to create more than 2 degrees of temperature rise on the IC.
 
biggs, I think we're getting some confusion since the circuit essentially requires two voltages to operate. On the Vin pin, it requires 5 volts. This is the IC's operating supply that it uses to power it's internal circuits. This supply should have very very little current flowing on it. The other supply is the +LED voltage, which is applied to the positive end of your LED string. The LEDs drop "most" of that voltage and the remainder is applied to the LED pin. The IC must dissipate this entire voltage at the LED's current, which is where the potential for heat and waste comes from.

So, the difference between our circuits and yours might be in the voltage "left over" after the LEDs have dropped your supply voltage, i.e. the voltage sinked at the LED pin.

Can you tell us your input voltage for the LED string? i.e. your supply voltage? Also, the total Vf for your LED string (guessing it's 3.4 * 5 = 17v)? The difference between those two voltages is what your chip needs to drop.

Can you verify with a known-good multimeter that your LEDs are operating at 700mA?

To operate, the IC needs to be dropping at least .5v on the LED pin. At 700mA, that's .35w, so not very much at all, but I'd still expect it to create more than 2 degrees of temperature rise on the IC.

I drop the voltage with a 1/2w resistor for the led's (5v -> 3.4v). vin for the chip is 5v. I'm running 1 led per driver. Yes, I checked the current and it's running at 660mA per driver.
 
biggs,

My I am reading wrong. Sounds like it is not regulating. You say it is configured for 700, but then you measure 660. What is the resistance 1/2 resistor? If it is larger than about 2.0 ohm there is your problem. 3.4 (LED) + 2.0 * .66 (resistor) + 0.5 (for CAT4101) = 5.22 volts.
 
my voltage supply i chose was 5v and the led's require 4.3v or less. So to power the led's my i need a voltage drop. My R1 resistor (680ohm i believe, could be wrong) was a 10% + my pot that I added was a 5% so there exists the possibility that my current regulation is correct (.66/.7 = 95% of what is expected). I actually selected the limiting resistor and pot combo that would allow the LED to consume 800 mA maximum, but with the 100 ohm pot trimmed to full resistance, it should run at the 700 mA.

led's
Voltage In -> 2.7 ohm -> led -> cat4101 current reg -> internal chip ground

cat4101 voltage input
Voltage in -> cat4101 chip vin

cat4101 current limiting resistor chain
cat4101 -> 680ohm ->100ohm varpot -> gnd
 
Is that how you plan to run it in "production?" How many LEDs are you going to drive?

It strikes me as inefficient from a $$$ perspective to only drive one LED per CAT4101 chip, when it'll happily drive 6 or 7 - you'd just need to cough up $15 for a 24v supply.

Also much less efficient in the long run, because you're "wasting" the power that the resistor and driver are dropping for each LED. If you were to run 6 LEDs per driver, you'd have the same waste per driver, but it would be "spread out" across 6 or 7 LEDs, so overall efficiency would be much higher.
 
I still think something is wrong, but given to tolerances 660 ma is possible.

Here is my issue:
5 (volt supply) - 2.7 * .660 (resistor drop) - .5 (Cat4101) = 2.718 (volts for the LED)

I think your resistor is to high. Short it out and measure the voltage and current of the LED. Then we can figure out if it is correct or what is should be.
 
I designed my circuit to be a 1-n driver. Right now, it's the way I have it configured with the power supply I purchased. I eventually can see this board running a 1a or higher (cat4101's can be run in parallel for more amperage). One adjustment of my power supply to run at 5.5v and two LED's 500 mA would be much more efficient.

For right now, I'm playing around with the driver. The < 1 watt wasted on the resistor doesn't bother me that much.
 
I still think something is wrong, but given to tolerances 660 ma is possible.

Here is my issue:
5 (volt supply) - 2.7 * .660 (resistor drop) - .5 (Cat4101) = 2.718 (volts for the LED)

I think your resistor is to high. Short it out and measure the voltage and current of the LED. Then we can figure out if it is correct or what is should be.

I bet I forgot about the .5v drop. I'll have to check this when I have some time to putz.
 
It shouldn't matter. There's no specific wattage called for in the datasheet and the power through the resistor SHOULD be really small. I used 1/4w resistors (Mouser part 660-RK73H2BTTD7680F) and they're just fine.

Edit - one thing you MAY want to consider is accuracy. The part I used has a 1% tolerance. It was only 5 cents each so it's not like I was paying "extra" for that precision.
 
I think DWZM said he found it to be closer to 0.6.

I did, but IMHO that's within acceptable tolerances of .5v when you consider the different factors that could change the value for a given circuit (temperature, loss in wiring or terminals, etc.) Especially differences in measuring technique. If you're measuring drops across each individual LED right at the LED, and then you just subtract that from the voltage you measure from your power supply right at the power supply, your "drop" will include many (very small) losses besides just measuring across the IC, for instance.
 
DWZM, Took the stuff from your BOM Only. Thanks a ton for those useful info. I also ordered the 866 Ohms (600ma) and 680Ohms (800mA) for testing purposes...

Fishman,Future Electronics, doest ship to India. And Mouser.com seemed cheaper with a flat 40$ shipping for the entire setup of 25 Cats and 100 in each of the resistors and caps!!. So went with that.. and thanks a ton for your guidance.

Will be coming back once i get the components in 3 days!!
 
Are all you guys running the boost config? I'm trying to use the NCP3066's that I have but was thinking about running them in buck. What's the reasoning you used for going boost?
 
Back
Top