DIY LED driver for reef lighting

[terahz]

ingoh, thanks for pointing that out. The "value" for the opamp was different (I don't know why it didn't update the schematic when I changed the board). It is just a string difference, absolutely nothing else. I've fixed the schematic and reuploaded the files.

While I'm fairly certain the pcb will work just fine (I have only added a few small components) until I test it I have no way of knowing for sure. My order of PCBs was shipped yesterday and it seems to take about 6 weekdays to come to me, so probably mid next week I'll be able to confirm that everything is working fine.

 

[ingoh]

Hi terahz,

the new files are fine now. I'll wait until you confirm that the layout is OK. Would be nice if you drop me a line.

regards,
Ingo

 

[matiLanza]

hey everyone, im starting my LED build now for my 72 bowfront and i was thinking of doing a 72bridgelux kit but i want to make my own drivers. This forum is 79 pages long!!! my eyes will bleed if i read comment by comment lol can anyone help me out

basically i need a driver that is PWM dimmable since i started my arduino build for the controller which i was planning on using meanwell P's so i have a 10v power source and 10k pots already and the basic sketch..

anyone?

 

[TheFishMan65]

There have been several driver discussed. Here is a summary for the CAT4101.

 

[der_wille_zur_macht]

The CAT4101 is indeed a good choice. Advantages are that it is tried and true, used by many, cheap, easy to assemble, requires few parts, and can interface directly with an arduino for pwm dimming. The only real disadvantage is that it is linear, so the difference in voltage between your power supply and your LEDs gets turned straight into heat. It really helps to have a ps with a trimpot to adjust voltage so you can "tune" it to your LED strings.

The LM3409 design is probably the other front runner - this is the one Terahz has been talking about lately. It is potentially more efficient over a wider range of conditions, and way more flexible in terms of drive current and voltage (i.e. number of LEDs per driver). The disadvantages are that, in this thread at least, it is a newer design and not as well tested, plus it requires more external components and soldering a fairly tricky IC package.

 

[matiLanza]

Thanks for the respons guys, ill look into the cat4101

 

[der_wille_zur_macht]

Some good news, and a PSA.

Terahz's 3409 design works fine at 2.6A running XM-Ls on 28v. At least, one of the two boards I have does. The other works like it is firing on one less cylinder. It is always at about half the current it should be. I suspect it is a problem with this particular board which has seen a rough life. I am going to go comment in my tank build thread on my impressions of these LEDs.

Speaking of rough lives, here is today's public service reminder. Test everything! I tested all the LED strings on my new 360 build for shorts with a multimeter, and test ran each string with a known-good driver. They were all fine. What I neglected to do was test each driver independantly - I wired them all up and flipped the switch. There was a POP! And the driver box filled with smoke. (Then the fans in the box blew it all in to my face!). Turns out one of the PCBs had a short on the board itself. Thanks, seeedstudio. 8 4101's in the garbage. Luckily the other components and the LEDs are fine. If I had bothered to test the boards more thoroughly, i.e. check for shorts before assembling them, I wouldn't have had this problem. This caused me to review my testing procedures, which I hadn't really put any formal thought in to, and I thought I would share them since I haven't seen a comprehensive discussion in this thread yet. It is one thing to build and test an LED rig with bofff the shelf components, but when you are DIY'ing every component it requires more rigor.

0) Set up a test station, including a multimeter, known-good driver, DC supply, LED string, and dimming signal. The idea is to have a complete "environment" separate from the rig you are building such that you can swap components one at a time and you know that everything in this environment is good except the component in question.

1) Solder the LEDs, test for shorts to the heatsink with a multimeter. Fire up each string on it's own with a known-good driver and known-good control (dimming) signal.

2) Test driver boards for shorts with a multimeter. Assemble boards. Test each driver independantly with a known-good LED string and control signal by rigging it up and checking for appropriate voltages and?or currents at each individual input or output point on the board.

3) Assemble the "infrastructure" for the production rig, i.e. wiring, terminals, dc power supply, and so on. Tag or label all wiring. Do not connect the drivers yet. First, test with a multimeter to ensure wiring is correct, nothing is shorted or labeled wrong.

4) Insert drivers and fire it up! Check again for appropriate voltages and/or currents on each I/O point for each driver.

Thoughts anyone? Does anyone else want to share your steps for testing a DIY driver or LED rig?

 

[O2Surplus]

I think you're spot on with the driver testing DWZM. I have had my share of bad PCB's from seeedstudio also. I used to pass up the extra charge for 100% testing, not any more, and I even test the circuit traces by hand prior to soldering, just to be sure . I've had enough cats and Atmel avr's go up in smoke to justify the additional cost of testing. It's cheap insurance.

 

[NanoNana]

OMFG...

That was five days well spent (I'm a slow reader and a busy person), but thanks for all of the info folks.

It looks like I'm gonna try the design by Terahz. Of course, I am slow to order parts and to solder them so it will be awhile.

But seriously, THANKS TO ALL YOU GUYS!

 

[crisaguilar]

hi guys

i be a passive follower of this thread and some others.

i make mi LED setup for my tank whit the follow specs.

-tank top area 39 X 15.5 in.
-4 20W leds for china (very good leds) 2 royal blue and 2 whites.
-12volts
-2000mA (2amps)

-2 driver boards whit 4 cats each one(one for blues and one for whites) configuration 2 parallel for each led. (8 cat in two boards)
-Arduino Mega with one sketch of the thread arduino sketches. Modified by me.


All work very good in all test, but the Sunday (2 days working very good) i have a problem, one of the RB not turn off complete maybe low the dimming to 40 or 30 percent and stop dimming.

i start to make some test and found one cat is the problem, i test the voltage in the pwm pin (pin1) and all the 4 signal are in 0 volts, and all the cat have 0 volts in the output except one, this one have 7 volts.

check all the tracks in the board and all are ok, well i proceed to change the faulty one, and all work again.

yesterday i see how the sunset work but my surprise be when one white led don´t continue dimming, again stop to dimming at 40 or 30 percent, y check and one of the cats have 8 volts again :headwallblue: this cat are in the other board.

but i don´t want replace because is too late in the night, i be very afraid because i can´t continue change one cat each day. Some one have this problem whit the cat, i check all connections and solders and all are ok.

i check the temperature, is hot but i can touch the cat for a long time whit out burn my fingers, what is the problem whit my configuration.

i live in Guatemala and only buy 10 cats i don´t have more to continue changing, i buy today other 5 or 7 but is a problem because i need wait a 3 or 4 weeks for the shipping.

some one can help me or have some tips to resolve my problem

tanx

sorry for my bad English but is not my natural language.

 

[TheFishMan65]

I was going to say check the heat, but you did that. I am lost, but the following may help.
What is the voltage across the LEDs?
Do you have datasheets on the LEDs?
What is the current for each CAT?

 

[shark boy]

I have a question.

I have a 24v power supply running 2 strings of 7 cree xp-e royal blue leds at 1000ma on a 40x1 inch heat sink The leds are 2.5 inches apart.
I decided to do this because the voltage differance with 6 leds would leave almost 3 extra volts that would get turned to heat and 7 would power the leds at 3.43 volts. When fully on, my leds (not my drivers) are creating a lot of heat. I thought that because of the extra led I could run under better conditions and would have "cool" leds. The heat is about the same as a 54 watt T5.
I powered them up last night for the first time and have not installed them into my hood. I don't have any fans on my heatsink yet, I wanted to wait and see if I needed them.

Is this kinda heat normal?
Thanks

 

[der_wille_zur_macht]

Shark boy,

You need to seperate power dissipation at the driver from power dissipation at the LEDs. Voltage drop at the driver only impacts heat produced by the driver. As long as you are within operational parameters, the power at the LED will not change.

 

[crisaguilar]

I was going to say check the heat, but you did that. I am lost, but the following may help.
What is the voltage across the LEDs?
Do you have datasheets on the LEDs?
What is the current for each CAT?

the specs of the leds

Royal Blue

Wavelength: 445nm Royal Blue
LED chips: EPILEDS EP-B4545V-A3 45x45 mil
DC Forward Voltage (VF): 10-11V DC
DC Forward Current (IF): 1000-2100mA max
Viewing Angle: 120 Degree
Intensity Luminous (Iv): 180-450LM
Life span 30000 hours guarantee

White

Color temperature: 20530K
LED chips: EPILEDS EP-B4545V-A3 45x45 mil
DC Forward Voltage (VF): 10-11V DC
DC Forward Current (IF): 1000-2100mA max
Viewing Angle: 120 Degree
Intensity Luminous (Iv): 800-1400LM
Life span 30000 hours guarantee

maybe you want see the other items in actinic leds, this are asome leds :eek2: i don´t know if i can write the ebay item number.....

i take the actual values of the voltage and current in the night, but i remember the voltage was 10.5v. and the current was 1.4A i configure the cats to 1 amp with 560 ohm rset but never reach the 2 amp.

i don´t know how much current are driving each cat because i can´t separate the connections in the board. but each pair of cats are driving 1.4 amps.

my power supply is a PC power supply of a old desktop i have in my garage.

tanx for the help maybe some one have the same problem with his drivers all the suggestions are welcome.

 

[shark boy]

Shark boy,

You need to seperate power dissipation at the driver from power dissipation at the LEDs. Voltage drop at the driver only impacts heat produced by the driver. As long as you are within operational parameters, the power at the LED will not change.

If I have the right power going to the driver ie 24v dc driving 7 leds at 3.43 volts and 1000ma max current output from the driver to the leds how do I figure power dissipation, and how can I fix it?.
I thought it should be plug and play.

It sounds like something is amiss.

 

[der_wille_zur_macht]

There may not be a problem. The LEDs SHOULD be fairly hot.

Dissipation at the CAT4101 driver is easy to estimate fairly accurately. Measure your input and output voltages at the driver and fins the difference. Multiply by your drive current. The idea is to make this number as low as possible, but only for the driver's sake - "tuning" the driver like this does not change any aspects of how the LEDs are operating. Generally, people shoot for a voltage drop (difference) across the driver of half a volt, since that is the minimum this particular driver needs to operate. Hence, people often try to use a number of LEDs that puts then close to (but not under) half a volt velow the Vin in terms of Vf for the LEDs. LEDs often vary from their specified Vf so it is common to use a supply slightly bigger than the half-volt formula specifies, especially if the power supply has a trimpot to adjust output voltage as many do

As far as power dissipation at the LEDs, that is just voltage times current again. Given your specs, that means 3.43W per LED. Altering this is done by choosiing a different drive currentt, but of course that changes light output AND heat output. As long as your heatsink or other thermal management device(s) are adequate, 3.43w is fine for those LEDs.

 

[der_wille_zur_macht]

Ok i just did the math. If you are trying to get 1A current and your LEDs drop 3.43v at 1A (according to their datasheet) you need 24.5v input, not 24. At 24v your LEDs are under driven right now.

Where did you get the 3.43 number from? Something makes me think you are doing your math backwards.

 

[TheFishMan65]

crisaguilar,

I am not ignoring you just don't have any thoughts at the moment. If I think of something I will let you know.

There is not a lot of experience running these in parallel. If you loose another I would remove the bad one and let the singe run at 1 amp. See if you have any problems running singularly. It also seems odd that you are only getting 1.4 amps when you specified 2 amps. Maybe you are voltage limited and this damages the CATs. Do you have resistors so you can set them to 1.3-1.4 amps?

 

[shark boy]

Ok i just did the math. If you are trying to get 1A current and your LEDs drop 3.43v at 1A (according to their datasheet) you need 24.5v input, not 24. At 24v your LEDs are under driven right now.

Where did you get the 3.43 number from? Something makes me think you are doing your math backwards.

I got the number from deviding 7 into 24. I thought 6 leds at 3.5 volts would leave too much voltage to create heat. My power supply does not have a trim pot. I may have to change supplies.
Thanks for the help.

 

[TheFishMan65]

shark_boy,

Do you have a CAT in there then it will eat up at least .5 volts.
23.7 / 7 =3.36. Not much difference, but I think this is what DWZM figured.