DIY LED driver for reef lighting

I do not have a precise target
I would like to have a driver that has no limits of Ampere and Volt or at least coming to the max about 10Amp 50volt
 
Well, the LM3409 is only rated to 5A drive current so you might have to look elsewhere to get a 10A driver (or, just use several lower-current drivers).

sorry, you have reason, I have not considered the limitation of the chip lm3409HV to 5A.
10Amp to my conception, it was only to think of ways around this restriction for a prototype of the future drivers changing components.
This driver is excellent, and 5Amp are sufficient and go very well, not easy to find drivers on the market like this!
terahz more compliment!
my request was just to look in a future prototype drivers to increase this limitation.

Sorry for my writing but I do not speak English very well :D
 
No problem Baghyyy.

I don't know about anyone else, but I have really struggled to find components for a "theoretical" maxed-out LM3409. Seems hard to find a MOSFET that has low Qc, low Rds, and high enough voltage limit, at least in the package we're currently using. Would probably be easy in a bigger through-hole package. Same for the diode and the inductor, at least in higher inductance values. I'm not really losing sleep over this though, as I'm happy that I've got parts to work with my parameters (roughly 3A and 28v maximums).
 
Ok so I am looking at going the different PSU route while i play with the old ones. Instead of using 2 24v 6.5A PSUs can I use one 24v 14A or something similar? I am only running 8 cat 4101s at a max of 900mA?
 
That would work if all CATs are dropping the same voltage. Another way to say that is if all LED strings have the same voltage. If they don't then some CATs will get hotter than others. If they get too hot they will go into thermal shutdown.
 
I will this week my multi meter has decided to stop measuring voltage so i am limited there but i did pick up a 100 pc pack of assorted resistors to play with!
 
Kcress, DWZM, TeraHz

Please take a look at this driver board and let me know what you think. I up-sized the board to 70mm x 75mm and mirrored the high current traces on the bottom plane for cooler operation. I'd like to be able to run up to 3 amps with this design- if you guys think this design is up to par.

LM3409Mk3.png
 
my triple cat4101 is overheating causing my string of neutral white leds to go off and on. all the blues and royals blues are fine. i put a small fan above blowing to the drivers and no issues. i dont have any meter readings now but when i get time, what should i look for?
im using a 150 watt 24v (adj min to 21.7v) mean well ps for two sets of drivers, 2 power supplies for the setup.
total of 72 leds, 36 leds in each end. four triple cat4100 drivers.

driver #1 for 6 RBs, 6 NW, and 6 NW
driver #2 for 6 RBs, 6 RBs, 4 RBs + 2 Blues
driver #3 for 6 RBs, 6 NW, and 6 NW
driver #4 for 6 RBs, 6 RBs, 4 RBs + 2 Blues

And only the Nuetral Whites string of leds are shutting off and then on after 3-5 seconds. After power is on, a few minutes later one set starts shutting down, then a minute later the other set...soon all four sets, its Xmas already!


40 CREE XP-E Royal Blue 3W LED
24 CREE XP-G Neutral White 3W LED
8 CREE XP-E Blue 3W LED

d7830516.jpg
 
Kcress, DWZM, TeraHz

Please take a look at this driver board and let me know what you think. I up-sized the board to 70mm x 75mm and mirrored the high current traces on the bottom plane for cooler operation. I'd like to be able to run up to 3 amps with this design- if you guys think this design is up to par.

LM3409Mk3.png

From a thermal perspective, if you choose components carefully, that design will most certainly work.

It's interesting to point out though that some of the board layout comments in the datasheet aren't really well implemented in that design. Though honestly I don't know how you'd determine if that was important or not other than firing it up and seeing what happened.

my triple cat4101 is overheating causing my string of neutral white leds to go off and on.

How are you supplying DC to the input side of these drivers?

Thermal cycling means the driver IC is dissipating too much power. It's dissipating too much power because it is dropping too much voltage. The voltage drop is determined by the difference between input voltage (i.e. voltage supplied by your DC power supply) and output voltage (i.e. voltage dropped by the LEDs).

If you can get your hands on a multimeter, measure input and output voltage for each driver. The ones that are shutting down will likely have the biggest difference. The best way to fix this is to use a DC power supply that is as close as possible to the desired output voltage plus the minimum drop (which is a half a volt). So if your string of LEDs is dropping 21.8 volts, you'd want a DC power supply running at 22.3 volts.

This is one of the biggest drawbacks of the CAT4101 chip. It needs careful selection of input voltage to not have thermal problems. Most people who are using this design with various different strings of LEDs of different composition (i.e. strings of RB and strings of CW) have a DC supply dedicated to each "type" of string, which lets you tune the supply close to that color's typical voltage drop. For instance, your strings of RB might all be dropping 23v, but your strings of CW are only dropping 19v, and you try to run both off a 24v supply, the RBs will be fine but the CWs will fry.
 
Or add a heat sink to the board. I just ran a test on my 5 chip board with mixed colors off the same, but tuned, psu for a full light cycle. Max temp my probe ever recorded was 30.6875 C.
 
Kcress, DWZM, TeraHz

Please take a look at this driver board and let me know what you think. I up-sized the board to 70mm x 75mm and mirrored the high current traces on the bottom plane for cooler operation. I'd like to be able to run up to 3 amps with this design- if you guys think this design is up to par.

LM3409Mk3.png


YeAH! That is probably pushing the maximum rational area. Probably even a little bigger than necessary. You should not have a problem
 
so would this power supply be ok? It is adjustable down to 21v where the 6.5A 150 watt only goes down to 22.8. Is anyone using this particular power supply?

http://www.ebay.com/itm/Original-Me...ultDomain_0&hash=item19c99957ae#ht_6383wt_977

That should work fine, though you really need to break out the multimeter and test each string to be sure whatever power supply is supplying it is in an acceptable range. Besides the danger of thermal problems if you supply too much voltage, you can easily under-drive these drivers by not providing enough voltage - so you really have to tune on a case by case basis, not just assume you need a lower voltage than what you already have.

O2, another comment on your design - though clearly not an important or critical one. :D If you want to get the pours on the bottom of the board to line up exactly with the pours on the top, just delete all the pours you currently have on the bottom, then use the copy tool to copy the pours you have on top - then change the layer for the new instance of each pour to the bottom layer. This will result in pours that are exactly mirrored instead of a few thousandths off - again this is definitely not a critical thing but my OCD streak likes to see things line up perfectly. :D
 
Kcress, DWZM

Thanks for taking a look and giving me the :thumbsup: on my design. I appreciate your input immensely. DWZM, I'll revise the polygons with your "copy" approach. That will be a real time saver, it never even occurred to me that it could be done that way.

O2
 
How are you supplying DC to the input side of these drivers?

Thermal cycling means the driver IC is dissipating too much power. It's dissipating too much power because it is dropping too much voltage. The voltage drop is determined by the difference between input voltage (i.e. voltage supplied by your DC power supply) and output voltage (i.e. voltage dropped by the LEDs).

If you can get your hands on a multimeter, measure input and output voltage for each driver. The ones that are shutting down will likely have the biggest difference. The best way to fix this is to use a DC power supply that is as close as possible to the desired output voltage plus the minimum drop (which is a half a volt). So if your string of LEDs is dropping 21.8 volts, you'd want a DC power supply running at 22.3 volts.


This is one of the biggest drawbacks of the CAT4101 chip. It needs careful selection of input voltage to not have thermal problems. Most people who are using this design with various different strings of LEDs of different composition (i.e. strings of RB and strings of CW) have a DC supply dedicated to each "type" of string, which lets you tune the supply close to that color's typical voltage drop. For instance, your strings of RB might all be dropping 23v, but your strings of CW are only dropping 19v, and you try to run both off a 24v supply, the RBs will be fine but the CWs will fry.

im using a mean well RS-150-24, two of them, one for two sets of boards. and also 5 v power supply for the 5v and pwm. i have no pwm arduino yet.

the only adjustemnt made was the 24v, i adj to the min at 21.7v. one meanwell ps for the NWs and some RBs. the other for all the other RBs and Blues.

so i will check the in and out and make the +5v adjustment to the 24v PS. If the whites are too low in V, then i may end up adding more components. thanks
 
Back
Top