DIY LED driver for reef lighting

terahz, I posted the number somewhere in here, but I found adding a heatsink on the CAT4101 significantly reduced the heat. You may want also to consider that option. I like because there will be no extra fan noise and no wasted electricity running a fan.
 
For those interested, here is a link to my build thread, which features the CAT4101 drivers and Hydra controller with 72 XP-G / XP-E LEDS. I'm almost complete.

Here is a pic of the driver box.

so3dz5.jpg


http://www.reefcentral.com/forums/showthread.php?t=1852144

I have also found that the XP-G drivers at 700ma are getting to about 140F.
I added two fans to the driver box.

5505cg.jpg
 
If you kept the voltage drop as small as possible, I would see no issue running it right at 1A. The thing you want to watch for is hitting the temperature limit, and even dropping two or three volts at 700mA I didn't hit it - so at 1A and .5v drop, you should be totally fine.

Okay cool, so I just need to swap out the 768 ohm RSET resistor for a 549 ohm in the BOM.

I am having trouble getting Eagle 5.10.0 Light to output the gerber files that Seeedstudio wants, would you be able to post the gerber files that need to be sent to order boards?

Unless someone here has one or two extra CAT4101 boards they would be willing to let go of, please don't hesitate to pm me. Right now I only need 1 or 2 boards to do my small test build before I move on to a real one.
 
I'll email you gerbers, send me your address in a PM. Seeed publishes a cam job on their website that you need to use to get the right format. It's a little obscure your first time through but worth figuring out.
 
Today has to be the strangest day for me yet. Apart from measuring more par that the bottom of my tank from the top, I've upped the current of the drivers to 600ma (from 400ma) and now the avg driver temp is 150F (from 130F) o_0.

The whites are at 18.3V, the blues at 18.6V. PSU 20V.
 
I'll email you gerbers, send me your address in a PM. Seeed publishes a cam job on their website that you need to use to get the right format. It's a little obscure your first time through but worth figuring out.

Thanks, you got a PM with my email. I did get a cam job file from the Seeedstudio forums and ran that but the resulting files don't match either, the file extensions don't come out the same as what they have listed that they need.
 
Today has to be the strangest day for me yet. Apart from measuring more par that the bottom of my tank from the top, I've upped the current of the drivers to 600ma (from 400ma) and now the avg driver temp is 150F (from 130F) o_0.

The whites are at 18.3V, the blues at 18.6V. PSU 20V.

How much volts is the driver IC dropping? ~1.7V for the whites and ~1.4V for the blues? Is your power supply adjustable?
 
This is quoted directly from my LED thread, but I figure it is pretty relevant here as well.



I have decided to go with the DIY CAT4101 driver designed by Der Wille Zur Macht in the LED driver thread for my large tank setup. This allows me to have a more cost efficient driver for the larger setup, as well as the ability to use only two 24v high amperage power supplies that have PFC protection built in. No more worries about my house burning down with too many Meanwell drivers.

I am only using 2/3 of the board capabilities, but that is because I wanted to stick to my original plan and control 12 LEDs a channel with PWM. Heres the actual board, only ten parts in my application!
IMG_1225.jpg


This setup is will run five boards (sixty LEDs) and is powered by this bad boy:
IMG_1227.jpg


Arduino Mega clone. Currently this is providing a 5v source to the CAT4101's required 5v in as well as the 5v PWM. Once I figure out how to program it it will provide 5v to the 5v in pins of all ten boards as well as controlling ten channels of PWM.
IMG_1226.jpg


And I have light! This is two strings of six XR-E Royal Blues. When I turn down the power supply so that it is only dropping ~0.8v more then the LED strings themselves the CAT4101's barely get warm to the touch!
IMG_1223.jpg


Tested all ten of my boards with the old LED setup.
IMG_1228.jpg


I turned the extra board Seeed Studio sent me into a spare with all three channels populated.
IMG_1231.jpg


Arriving sometime tomorrow should be my five 8.4"x10" heatsinks from www.heatsinkusa.com . This time I am actually going to screw the LEDs down, but only because my father's company owns a CNC machine and he offered to have the heatsinks drilled and tapped for me. If anyone is interested in the CNC files or perhaps getting your own heatsinks tapped go ahead and send me a PM.

I also have all the LEDs I need to light the big tank (120 LEDs, 60 XR-E CW, 60 XR-E RB), but I have decided to incorporate 10 CREE XR-E Neutral Whites into the setup. These will be run two per each fixture, without optics, for more even spread and to try and add some more red light into the spectrum without driving the overall color temperature down too far.
 
How much volts is the driver IC dropping? ~1.7V for the whites and ~1.4V for the blues? Is your power supply adjustable?
Well I've increased the current to 800mA and 600mA, and now the drop on the blues is almost none and on the whites is about .5V so there is no excuse now for these to be hot, yet whites are still at 140-150F and blues at 110F.

My PSUs are adjustable between 20V-30V, and they are at 20V.

I might try doing 3x4 LEDs instead of 2x6 per board using 14V psu...


Taqpol, you gave me a good idea to space out the drivers. With all 3 soldered there is not enough space to use standard heatsinks, but with 2 side ones it will work.
 
I had the same problem with heat. Look back and you can see where I placed a heat sink on the CAT4101s. I asked if anyone else was having problems, but apparently not that many have or else have not gotten to running them. I know I am still working in mine :)
 
I didn't have a significant heat problem when I built the prototypes and ran them for a few days, but I haven't run the drivers since then. If I DO discover a heat problem in the final implementation, I'll probably just epoxy a piece of U-channel aluminum across the "tops" of the ICs, and/or make sure there's some air movement around the drivers.
 
I am surprised after "a few days" you did not have a problem. Do you remember the current and over voltage? I'll check the answer when I get back from vacation :)
 
700mA, and iirc .6v dropped across the driver. This was the setup I took the photographs of earlier in the thread. The driver was sitting on the table, a foot or so away from the LEDs, which had a fan on them. I don't think there was any direct airflow on the drivers but maybe that fan helped a bit.
 
I found the same thing running my CAT4101 drivers. 700mA current and dropping 0.8v across the driver. After running for over seven hours they were warm but I could hold my hand on them indefinitely.
 
I didn't have a significant heat problem when I built the prototypes and ran them for a few days,

The interesting thing is the drivers for the whites are much much hotter than the blues. The blues are not hot at all, they are barely over room temperature.

I moved boards around just to make sure it is not the particular boards but didn't make a difference. Even when the blues were dropping 2V they were not hot.

I'll put a couple of big fans on top of the box so that I have piece of mind.
 
Would running a cree 2 feet from the driver be an issue?
I have the 9 led array, the tenth led is needed so that I can run it all off of a 16 volt 2 amp psu.
The last LED would be used for a mini fuge,
I'll live with the fuge light not being on as much as it should be.
Thanks for the help :)
Josh
 
Would running a cree 2 feet from the driver be an issue?
I have the 9 led array, the tenth led is needed so that I can run it all off of a 16 volt 2 amp psu.

My drivers are probably 5-6ft from the LEDs and it doesn't seem to be a problem, however 10 Cree LEDs on a 16V psu will require a boost driver, most likely, because at higher currents they will be dropping more than 16V (5 XR-Es could go as high as 18.5V). Just something to keep in mind.
 
Hmm I thought it was more like 3.3 volts(plus a half volt ) per Cree at 800 MA?
Well then if that's the case then I will go with a 18 volt PSU.
 
Back
Top