DIY LED driver for reef lighting

[der_wille_zur_macht]

Yeah, though I would potentially segregate that into controller functionality, vs. driver functionality. If a driver can take ANY signal to alter output, you can probably build a controller or a signal converter to do whatever you want with it.

Translating that into the real world, you could put a "front end" on the LDD that took any combination of signal formats you wished and dimmed the driver accordingly.

 

[CrazyEyes]

I posted in the wrong thread. And couldn't figure out how to delete.

 

[zachts]

Yeah, though I would potentially segregate that into controller functionality, vs. driver functionality. If a driver can take ANY signal to alter output, you can probably build a controller or a signal converter to do whatever you want with it.

Translating that into the real world, you could put a "front end" on the LDD that took any combination of signal formats you wished and dimmed the driver accordingly.

that's true, would be cleaner, I still live in a knob oriented world and haven't leaped into controller/arduino design or programming yet so I lean toward simplicity and somthing that a simple typhon could controll but would have more fine tuning ability for the over all color. the ideal "white" I'm playing with requires at least 5 separate LED spectrums to create the white look that I like, using knobs to dial it in seems faster to me as I want it to just behave as a single color......

I'm mainly thinking of and planning a two channel setup, main lights dim up/ down. moonlights dim up/down. with all the color tweaking taken care of at the driver level with good old simple knobs

which on the typhon or any similar already figured out DIY project controller (doesn't require me having to learn anything new to build) then leaves you with controll for sump LEDs or PWM fans/other devices, or a second bank of lights for Left to right dimming as well.....

there are some LEDs such as whites though that look differnt at different driver currents and to get consistent color over the dimming range PWM has to be used. controlling both current and duty cycle allows for a little more flexibility in the end result and the color consistency at differet in tensities. (some day some people much smarter than I will figure out a controller and driver combo to do this all from the controller)

 

[Terminatermule]

Thanks for the reply.

Thanks for the reply.

Thanks for the reply. I had been looking at the 4101's as I wanted a driver similar to that offered by stevesleds but I wanted a little more control over what it could do. Basically a slight rejig of his driver boards to offer a few more adjustments.

I had not really looked at the LDD units, I will sus them out now.

What would you want for the boards you have? May be worth building one to compare just for fun anyway.

PS:

Just found a thread on thereefuge and realised that the boards they have the pics up of are yours, could I please hassle you for a copy of the eagle file for them to tinker with? Also what did you think of using the sockets as opposed to just soldering thru hole, do you think the cost was worth it?

Edit again:

Ok so I just read that whole thread, are you "magicJ" SurplusO2? If so (and I do suspect so, lol) can I get in on the purchase of the ldd's and boards? I am going to be driving ~100 LEDs including a few 350ma moon lights.
Cheers Glenn

 

[der_wille_zur_macht]

(some day some people much smarter than I will figure out a controller and driver combo to do this all from the controller)

Hey give yourself some credit. :lol: I did actually design a version of the CAT4101 board that had a digital pot (I2C controlled) on the Rsense pin, as well as the "standard" dimming control via PWM on the EN pin. Never "released" the design though as that was more or less right at the time that I was totally giving up on the CAT4101. Besides the fact that it's not cheaper than the LDD, it's also IME pretty darn fragile. I am 100% LM3409 at the moment. If I wanted something under 1A I would probably buy LDDs.

Regarding your overall approach - I get it, but I wouldn't do it purely to preserve the color for the white LEDs. To me, that doesn't add much value, especially since the whole reason why people are usually dimming whites is to "blend" and get a certain color. If the whites have some spectral shift at different nominal currents, all that means is your response curve (white LED current vs. overall system color) won't be totally linear, which is probably something that most people won't even notice, and at worst can be fixed with a small tweak of the knob.

Instead, I would want to control current separately from PWM dimming to preserve efficiency. If you have a driver running at 2A, but have dimmed it 50% via PWM, you'll be effectively running with the efficiency of a 2A drive current. By contrast, if you just redesigned the driver to run at 1A, you'd have slightly better efficiency.

In the end IMHO the best approach is to design the driver such that the drive current is pretty close to what you "normally" want, then use dimming for MINOR tweaks and "effects" like sunrise/sunset.

 

[O2Surplus]

Thanks for the reply. I had been looking at the 4101's as I wanted a driver similar to that offered by stevesleds but I wanted a little more control over what it could do. Basically a slight rejig of his driver boards to offer a few more adjustments.

I had not really looked at the LDD units, I will sus them out now.

What would you want for the boards you have? May be worth building one to compare just for fun anyway.

PS:

Just found a thread on thereefuge and realised that the boards they have the pics up of are yours, could I please hassle you for a copy of the eagle file for them to tinker with? Also what did you think of using the sockets as opposed to just soldering thru hole, do you think the cost was worth it?

Edit again:

Ok so I just read that whole thread, are you "magicJ" SurplusO2? If so (and I do suspect so, lol) can I get in on the purchase of the ldd's and boards? I am going to be driving ~100 LEDs including a few 350ma moon lights.
Cheers Glenn

LOL - I'm not "Magic J". Here's a link to the newest version of the 5up LDD-H PcB -Reef Central Online Community - View Single Post - Meanwell LDD driver: for those who want to dim to 0 using Arduino

 

[KRavEN]

Hey give yourself some credit. :lol: I did actually design a version of the CAT4101 board that had a digital pot (I2C controlled) on the Rsense pin, as well as the "standard" dimming control via PWM on the EN pin. Never "released" the design though as that was more or less right at the time that I was totally giving up on the CAT4101.

Any chance you still have the eagle files for that? I'm interested to see how you did the I2C digital pot.

 

[der_wille_zur_macht]

I will look but probably not. That development work was done on a laptop that I no longer have, and when I turn over PCs I tend to only keep the stuff that had been stored in a repo (which is pretty much just the projects I've published here or in other threads). I tend to end up with dozens of abandoned projects every time I have a burst of creativity, archiving them all would be quite the task.

If you spend some time looking at mouser or another electronics supplier you'll find plenty of digital pots though so you should be able to find something that works.

 

[zachts]

Regarding your overall approach - I get it, but I wouldn't do it purely to preserve the color for the white LEDs. To me, that doesn't add much value, especially since the whole reason why people are usually dimming whites is to "blend" and get a certain color. If the whites have some spectral shift at different nominal currents, all that means is your response curve (white LED current vs. overall system color) won't be totally linear, which is probably something that most people won't even notice, and at worst can be fixed with a small tweak of the knob.

Instead, I would want to control current separately from PWM dimming to preserve efficiency. If you have a driver running at 2A, but have dimmed it 50% via PWM, you'll be effectively running with the efficiency of a 2A drive current. By contrast, if you just redesigned the driver to run at 1A, you'd have slightly better efficiency.

In the end IMHO the best approach is to design the driver such that the drive current is pretty close to what you "normally" want, then use dimming for MINOR tweaks and "effects" like sunrise/sunset.

We're thinking along very similar lines, the efficiency gain is a big benifit, also I like the simplicity of having one driver that can be adjusted to suit the current needs, makes project implementation eaisier on a DIY level, a one part fit's all sort of appoach.

my main point with the "whites" was that I intend to use red, green, and blue in suplementing my white (probalby a remote phosphpor type rig when all is said and done) and i want their relative intensity and the total "white" color to dim as one rather than shift color as it would if only using pwm. (since if you have one color dimed slightly to achieve the all on color then it will have fewer dimming steps available to it and shift intensity more quickly than the other colors) Mainly I want to aviod what the Philips L prize bulb did and have a light that turns red at low dimming values due to the diferent response of each color LED. I tend to over complicate things sometimes, which is why I havent jumped into controller design yet, I'll end up building a monster that can controll my whole house from another country and spend two years developing it :lolspin:

 

[der_wille_zur_macht]

Ok, gotcha. If you can deal with each color on it's own driver and sending a separate signal to each color's driver you can fix the rest in software - essentially build a different response curve for each color with respect to "total" dimming. So if your overall dimming level is 50%, the controller knows that means the reds are 45%, blues are 52%, and whites are 50%, or whatever it works out to be. You'd basically be writing color correction software, which would be a pretty neat challenge. But probably a unique challenge, as most people want a significant change in spectrum during "total" dimming (usually more blue at lower light levels though).

 

[Terminatermule]

Sorry O2 !!

Sorry O2 !!

Thanks Guys.

I hate that, I had found this thread somewhere in the middle, read a bit, decided I wanted to read it through, so I posted the initial questions, then started reading it tip to toe, but in doing so I missed the late discussion of the the LDDs.

Ok O2, The other thread sort of read like you were possibly the same person, I'm sure I read in their a claim to having designed the boards, then the pic had your name on the silkscreening.

I will evaluate my layout today and I may have a few other questions for you guys. Thanks for the input, has been great.

Cheers Glenn

 

[zachts]

Ok, gotcha. If you can deal with each color on it's own driver and sending a separate signal to each color's driver you can fix the rest in software - essentially build a different response curve for each color with respect to "total" dimming. So if your overall dimming level is 50%, the controller knows that means the reds are 45%, blues are 52%, and whites are 50%, or whatever it works out to be. You'd basically be writing color correction software, which would be a pretty neat challenge. But probably a unique challenge, as most people want a significant change in spectrum during "total" dimming (usually more blue at lower light levels though).

that does sound like an intersting challenge, just one I'm not ready to tackle just yet for now I'm sticking good old anolauge simplicity for some of the work but at some point the software based approach would be the way to go. (future projects, yay!)

 

[MagicJ]

Edit again:

Ok so I just read that whole thread, are you "magicJ" SurplusO2? If so (and I do suspect so, lol) can I get in on the purchase of the ldd's and boards? I am going to be driving ~100 LEDs including a few 350ma moon lights.
Cheers Glenn

:lolspin: No he is not me, although I wish I had the ability to knock up some great PCB's like O2surplus can

 

[indutor_ki]

DEVS

Here's a dual CAT driver that will work. As it is, both channels are independent of each other. If you want to Parallel the channels, just connect the negative side of your LED string to both negative terminals on the board. I included pots with both chips for current adjustment. All you'd have to do is balance the load between the two CAT chips by adjusting each one to drive half your target current. For example- Say you want to drive a string of six XPG's to 1400 ma. Connect the string to one channel only, power-up, and adjust the current to 700ma. Now connect to the other channel only, power up, and adjust it for 700ma output. finally connect both channels and check the current, It should now be around your target of 1400ma.This design is nice because you're not forced to parallel drive if you don't need to, plus it has a 5V reg included for powering up the driver for situations where there is not a controller providing the 5V turn on signal or dimming signal. Feel free to mod it any way you like.

hello, first let me thank you for your work and for sharing it.
i want to now one thing about the dual cat4101 driver you design. i don't know much about electronic but in that board you are using a 12v regulator, but the led and the vin of the cat, i believe is 5v. why is not a 5v output for your regulator.
sorry if i missunerstand

 

[O2Surplus]

hello, first let me thank you for your work and for sharing it.
i want to now one thing about the dual cat4101 driver you design. i don't know much about electronic but in that board you are using a 12v regulator, but the led and the vin of the cat, i believe is 5v. why is not a 5v output for your regulator.
sorry if i missunerstand

I probably made a mistake when I drew up that PcB and used a 12V reg instead of the 5V version. Both regulators look the same. Just substitute a UA7805 in place of the UA7812, and it will work fine.

 

[indutor_ki]

Ok, thank you for your attencion.
I think i will try the smartcat, i will try to convert it to gerber and send it seeedstudio, to make some pcb's.

 

[marspeed]

if anyone needs any cat4101 i have 25 to many i also have some 3 up boards left over

 

[indutor_ki]

Ok, good to know. I'm still waiting for the pcb to arrive.

 

[ShadowNoob]

opinions?

opinions?

I was wondering what your opinions were on TPS92510DGQ from Texas instrument. I plan on running 12 cree xp-g r5 5w per driver they run at 3.25v 1500ma. Thats 39v total, That was the only chip i found that would fit the requirements of 1.5a with 39+v. Im not saying I plan on running it at that just looking at possibilities. I'm leaning more towards running them at 1000mah. But if i decide to just run them at 1500mah i would like to test one out.

 

[ShadowNoob]

i just seen the new LM3409 at mouser and its a 2A 75v max so thats another one and its a bit cheaper.