DIY LED driver for reef lighting

[der_wille_zur_macht]

Yes, depending on component selection, you'll hear a buzz from the PWM switching. I can hear it on my drivers if I'm within a few feet of them but it generally gets lost given the other noise on the fish tank.

Pushing the Arduino to different switching frequencies, or switching components on the board (the caps for one) can change the sound if it's really bad to you.

GreenHammer, I have a few locals who are building iterations of this board and I'm helping them with component selection. Once these things get nailed down I'll publish the BOMs here, since as you've pointed out it can be confusing and there's no need to recreate the wheel if you're using a fairly standard configuration.

 

[GreenHammer]

....
GreenHammer, I have a few locals who are building iterations of this board and I'm helping them with component selection. Once these things get nailed down I'll publish the BOMs here, since as you've pointed out it can be confusing and there's no need to recreate the wheel if you're using a fairly standard configuration.

DWZM, Thanks for the update! I will wait for your results. I have cleared some questions by studying the LM3409 data sheet, but I still have some more.

GH

 

[der_wille_zur_macht]

First though, here is my version of the LM3409 design:

http://code.google.com/p/hpled/source/browse/#svn/trunk/lm3409

The schematic, board, and gerbers are there.

My version is basically a stripped down version of Terahz's design. It allows for PWM dimming only. All of the people I know who are building these are using a Typhon or other Arduino clone for dimming hence the PWM format. Also, 0603 components make me nervous (I once made the mistake of breathing near one, and it was never to be seen again) so the passives are all bumped up in size, mostly 1206.

The inductor, MosFET, and diode are the same packages as Terahz's design.

Connectivity is via 3.5mm screw terminals.

Also, here is a BOM for this board, for 48v in, 44v out (so, roughly 13 typical HP LEDs), 850mA drive current:

http://www.mouser.com/ProjectManager/ProjectDetail.aspx?AccessID=c87dd815c5

In the customer part number column I referenced generic names for the parts (Rsense, Cin, etc) in the hopes that the BOM might prove useful for other designs, rather than naming based on my design (R1, C2, etc).

I'm working on BOMs for:

48v in, 40v out, 700mA (12 XP-series LEDs at 700mA)
48v in, 40v out, 1000mA (12 XP-series at 1A)
24v in, 20v out, 1000mA (6 XP-series at 1000mA)
28v in, 27v out, 2A (8 XM-L at 2A)

If people have other BOMs they would like to see, please let me know.

 

[der_wille_zur_macht]

BOM for my deisgn (or can be adapted to others) for 48v in, 40v out, 700mA:

http://www.mouser.com:80/ProjectManager/ProjectDetail.aspx?AccessID=ab99925606

 

[der_wille_zur_macht]

48vin, 40vout, 1000mA:

http://www.mouser.com:80/ProjectManager/ProjectDetail.aspx?AccessID=89a7dfba5c

 

[der_wille_zur_macht]

24vin, 20vout, 1000mA:

http://www.mouser.com:80/ProjectManager/ProjectDetail.aspx?AccessID=50f4794bbc

 

[der_wille_zur_macht]

28vin 27vout 2A

http://www.mouser.com:80/ProjectManager/ProjectDetail.aspx?AccessID=c2b6748f2d

 

[jgwinner]

24vin, 20vout, 1000mA:

http://www.mouser.com:80/ProjectManager/ProjectDetail.aspx?AccessID=50f4794bbc

Cool

I have a TON of old 20v Dell power supplies I was going to use to juice some boards. I realize this isn't very many LED's, 5 I think or 4? I'm typing this at work and don't have my spreadsheets.

However, I was thinking of doing some parallel strings, as I've got 90w available from these power supplies - just lower voltage.

19.6V out, 4.62A. I haven't checked ripple on these.

Think that would work?

== John ==​

 

[jc286006]

hi guys i just got some lm3409/hv v0.51 boards where can i find all the files i need for them i have found the v0.6 and v0.5 but so far have not seen anything for v0.51
thanks james

 

[der_wille_zur_macht]

Cool

I have a TON of old 20v Dell power supplies I was going to use to juice some boards. I realize this isn't very many LED's, 5 I think or 4? I'm typing this at work and don't have my spreadsheets.

However, I was thinking of doing some parallel strings, as I've got 90w available from these power supplies - just lower voltage.

19.6V out, 4.62A. I haven't checked ripple on these.

Think that would work?

== John ==​

It would definitely work. You could rig two drivers at 2A each on each supply and then put 4 or 5 LEDs in the appropriate number of series strings on each driver (probably two or three strings, to yield ~630mA or 1A on each). This would be easier than trying to get ONE LM3409 to "use up" the available current from each supply - component choice gets really hard when you get above 3 or 4A. You'd be talking about anywhere from 16 to 30 LEDs on each PS depending on number of strings and number of LEDs in each string.

Or you could put 4 or 5 LM3409s on each supply and just run a single string on each.

What LEDs are you thinking of running and what drive current do you want?

 

[der_wille_zur_macht]

The LM3409 can get very close to zero Vin-Vout. I have my XM-L drivers running with about half a volt of difference. If you're designing for a small difference in voltages, I would be sure to carefully watch the inductor ripple current (keep frequency high and/or inductor size large). If environmental conditions cause Vin-Vout to actually reach zero, the driver approaches 100% duty cycle, and the LEDs will stay at the peak inductor current (nominal LED current plus half the ripple). So you don't want a huge ripple, otherwise you may see a big increase in current as the driver hits 100% duty cycle.

Also keep in mind Vf will vary from LED to LED and will also change with drive current, so it's a bit of a crapshoot to design to a small Vin-Vout with a non-adjustable DC supply.

All that said, if you had 6 LEDs actually at 3.0 Vf and a 19.5v DC supply, I'd say you have an acceptable margin to have a reliable design. What drive current are you targeting? 1000mA? I'll take a shot at a BOM.

 

[GreenHammer]

DWZM, The BOM's are great information, just what was needed! Thanks!

I wonder, once I build a driver with one of the BOM’s setups, how much space to “stress” Vin and Vout values is recommended? Let’s say the following:

If I have the “28vin 27vout 2A” setup, but when I build my string of leds the total Vout is 28v, can I safely increase Vin to 29v?

If I have the “48vin, 40vout, 1000mA” setup, but when I build my string of leds the total Vout is 45v, can I safely run it like this?


PS: Cool dog drawing!!!

 

[jgwinner]

What drive current are you targeting? 1000mA? I'll take a shot at a BOM.

Correct! (Max)

6 LED's is a lot better as that works into the 2:1 ratio better. The only real problem is I'd have another set with the whites (CW, NW, or WW) running with just 3 LED's. Would that run on the same design? I'll look up the min voltage.

My original goal was to have banks of LED's brightening across the tank to simulate the sun moving.

== John ==​

 

[der_wille_zur_macht]

DWZM, The BOM's are great information, just what was needed! Thanks!

I wonder, once I build a driver with one of the BOM's setups, how much space to "œstress" Vin and Vout values is recommended? Let's say the following:

If I have the "œ28vin 27vout 2A" setup, but when I build my string of leds the total Vout is 28v, can I safely increase Vin to 29v?

If I have the "œ48vin, 40vout, 1000mA" setup, but when I build my string of leds the total Vout is 45v, can I safely run it like this?


PS: Cool dog drawing!!!

Ha, I wondered if people were going to comment on my "art."

I wouldn't be too worried about changes in Vin and Vout as long as they are a few percent. Especially if both values change - the real parameter that matters is the difference. As the difference changes, you change both frequency and inductor ripple (which has an indirect effect on Iled). If you're really concerned, you can always re-run the spreadsheet with the Vin and Vout that you obtain experimentally and see what the actual changes are, then adjust things if you feel like it's absolutely required. But on the whole, I wouldn't worry too much, unless you were already operating near the ragged edge of one or more components (i.e. if you were driving the LEDs right at their design maximum, I would be worried!)

My original goal was to have banks of LED's brightening across the tank to simulate the sun moving.

I appreciate that thought because it was one of my first ideas when working with LEDs. Of course you're free to do whatever you want, but after trying it myself experimentally on a few tanks I abandoned the idea. To me, it just looks weird. Casual observers to whom I hadn't explained the concept would not recognize it as simulated sun movements. If you think about it, on a real reef, the sun doesn't move across the sky by illuminating narrow strips of reef at different intensities. What really happens is the intensity on the entire reef slowly changes through the day, as well as the angle of illumination and the spectrum shifting. This is a very different look in a tank than having narrow bands at different intensities through the day.

Luckily I think it's just as achievable. The intensity and color changes we have down pat, you can just dim blues and whites separately. The angle thing is a little harder but not impossible, at least to get a rough feel. What I've done is to fix a very small LED array at one end of my tank, at a very steep angle. I fade that array on first in the mornings and keep it on the longest at night. The effect this has is that the intensity remains the same across the entire tank, but the angle of the light slowly shifts "towards the horizon" near the beginning and end of the day, while it's coming from straight overhead at noon.

 

[jgwinner]

By having more controllers and fewer LED's I could also isolate certain corals if needed, for example acclimation. But I think you're right, the narrow strips won't work without some tilting. Good to hear from someone that tried.

Ha, I wondered if people were going to comment on my "art."

...

The intensity and color changes we have down pat, you can just dim blues and whites separately. The angle thing is a little harder but not impossible, at least to get a rough feel. What I've done is to fix a very small LED array at one end of my tank, at a very steep angle. I fade that array on first in the mornings and keep it on the longest at night. The effect this has is that the intensity remains the same across the entire tank, but the angle of the light slowly shifts "towards the horizon" near the beginning and end of the day, while it's coming from straight overhead at noon.

That's sort of what I was planning - I even purchased some 20x10 degree lenses too, to attempt to shoot a narrow beam down the entire tank but keep the light off the tank walls. I haven't run a CAD simulation of that yet though, I have a bad feeling I would have to move the lights way to the side to get that to work.

I've got all these LED's and lenses now, but nothing built ... gotta actually wire something up

One of my 250W halides died a while ago, I just have 1/2 of the tank illuminated well with the other one. (Due to a phosphate problem it's mostly FOWLR anyway). I realy don't want to buy another ballast, I need to get SOMETHING LED based up and running. I do have enough parts for 4x3409 boards, and I've got 2 buck pucks just to start with. I used the spreadsheet to calculate the values, but I don't have it handy here at work.

== John ==​

 

[O2Surplus]

DWZM

Thanks for the BOM's, they're a big help, and nice artwork! I've been designing my LM3409 drivers with the same idea that you pointed out. As long as the ratio of Vin to Vout remains nearly constant the driver will behave predictably. I've been setting up the majority of my drivers to output 3,000ma max, with a Vin of 27 or 54 volts, and Vout of 25.5 or 51 volts respectively. I've found through testing that the driver will continue to operate correctly even if i adjust the Vadj voltage with a 250k ohm pot to lower the current output. It's got the same flexibility as a Mean Well, but with the potential for higher currents at a much cheaper price. Thanks again for all the valuable help that you bring to the Forum.:thumbsup:

 

[rbtron]

Terah,
V0.6 partlist still showing the old P MOSFET(DPAK) not the new TO220 package.

 

[terahz]

Thanks rbtron. The mouser url had the right part but apparently I forgot the spreadsheets. Fixed now.

 

[KRavEN]

So have any of you running one of the I2C drivers run into I2C bus length issues? From what I've read the I2C bus can go a maximum of 1 meter. With a controller beneath the stand and the drivers up top with the lights this could easily go over a meter in length. Any thoughts on using an I2C bus extender chip like the P82B715?

 

[der_wille_zur_macht]

So have any of you running one of the I2C drivers run into I2C bus length issues? From what I've read the I2C bus can go a maximum of 1 meter. With a controller beneath the stand and the drivers up top with the lights this could easily go over a meter in length. Any thoughts on using an I2C bus extender chip like the P82B715?

Where did you get the 1m limit? Maximum bus length is altered by a number of factors - clock speed (lower is better), choice of wire (twisted pair or shielded is better), number of devices (fewer is better), capacitance on the lines (lower is better), and pullup strength (stronger is better).

I've had devices with several meters of wiring and not had problems. I don't expect you'd encounter an issue on a typical reef tank installation, and if you did, you might find you could resolve it via adjustments according to the above points.