voltage regulator help for - Homemade Desk Light/Portable Emergency Lamp

[EAPellow]

I have a project that is effectively done, but I have a power supply question to take it the extra step.

I moved recently and have plans on making a work bench... somehow plans of making an LED desk-mounted work light came first. I basically threw together most of the decent LED's I had on hand that I thought might be bright enough, and just wired things together from there without much planning. At the very least, I mounted it into an old incandescent desk light, so it does look ok for now.

I have 3 banks of 3 different LED's. the components are as follows (and all LED's are wired in parallel, each with a dedicated 1/4W series resistor):

(8) 5050 RGB LED's, 60mA max, running at >40, (I fused the neg/pos of each of the 3 chips because I wasn't going to be mixing colors at all, and I still haven't figured out SMD soldering), wired with two 22 ohm resistors in series
(4) 9mm Strawhat LED's, 120 mA max, running at 90mA (If I had used only these, I would have been happy, but I only had 4), wired with a 100 and 22 ohm resistor in parallel
(8) 4.8mm Strawhat LED's, 20mA max, running at 16mA, wired with a 100 ohm resistor
Old cell phone power supply: regulated 5V, 1A max. With all LED's lit up, I'm using just under 500 mA.

All LED's have a 120 degree beam angle. Some of the SMD's blew one of their three chips during soldering (which is why I cut their current down by just over a third), and one of them just never lit up again after I wired in the resistors (which is why I like parallel circuits with dedicated resistors). I arranged the resistors on a separate board and sandwiched the two together, passing half of the LED leads through to the resistors. The three banks of LED's have separate positive wires, as I plan on having a switch for each one. I also plan on putting a big potentiometer on it for good measure.


Now here's where I'm stuck. I originally was going to wire this with a 12V power supply, which turned out to be dead. After I had finished wiring everything together, I realized that having it underpowered somewhat and running on 5V, I could easily get away with connecting it to 4 AA's and making it portable. I then thought that if I made a little box for it to be attached to that would act as a counterweight, I might as well wire it up to work on a range of different battery voltages (preferably up to 12 or even 24VDC) and have everything sitting neatly away in the box. We get power outages fairly often, and where my girlfriend works it's pretty much routine. They could always use an extra emergency light.

First I thought I'd just make a bank of plugs. Each would be wired to a different resistor to limit a specified voltage over 5V, and you'd just plug some battery terminals into them. Unfortunately, once I start getting up around 7.5V, I need 1/2 Watt resistors, and at 12V I start needing 1W resistors. Since the LED's have 1/4W resistors for the 5V wall supply sandwiched between a a board and some silicone right behind them, I didn't want them to start overheating. They might be ok since they're only limiting about 20% of the power at 12V input, but I don't really want to risk it.

So now I'm wondering about just making a 5V regulated circuit. Can anyone help point me in the direction of which I should choose and why? I have one LM317T on hand, and pretty much everything else to make a basic regulator (I've been looking at datasheets). My only problem with it is that it would limit the mid-range option of battery supplies, since the 317 would need at least 6V input to put out 5V. I wouldn't be able to drain lower voltage battery banks enough to really call this an emergency light.

The other circuit I was looking at is just a zener diode and resitor hooked up before the load. I think it still needs the input voltage to be a volt higher than the output voltage, but I think would consume less energy than the 317.

Any thoughts on how I might configure a supply? I'm still not above having to make a few different supplies for different input voltages (the box IS a counterweight, after all). Some of the key voltages I would input are 6, 7.5, 9, 11.1, 12, 14.8, and possibly 24. after the PSU their will be a dimmer pot and then 3 on/off switches for each bank of LED's.

Any help would be appreciated. Next project I have lined up, I'm just throwing DynaOhms in and forgetting about resistors and voltage regulation altogether.

*Edit: Here are some photos http://s204.photobucket.com/albums/bb274/eapellow/LED Desk Light/

 

[AnAppleSnail]

With sane LED mixing, an lm317 can deliver constant current. It will dissipate (I x (Vin-Vout)). Look up "Yh's lazy man constant current." It will take up to 32v in and about 30 out. I think you could get equal lighting performance with one or two.$3 Cree power LEDs. You would want your battery to delivery about 6v (4-5 NiMH. I suggest using a proper charger for them.

LEDs don't thrive in voltage-regulated environments. Especially mixed forward voltages, as you found when your RGBs broke. I bet you six lumen that the red diode blew. Red has a low forward voltage.

Efficiency: As far as I know, all regulators in series consume (Vin-Vout) x current power. Many LED.drivers are available with a wide input range in the 80% efficient range. These are constant current drivers that cost a few bucks when working DC/DC below an amp or two.

 

[EAPellow]

I had read somewhere that the LM317, while versatile, was a bit of a power hog compared to its non-adjustable brethren.

I'm fairly certain that the SMD's had problems because I was impatient about soldering them. I don't have a reflow workstation, and, rather than looking up clever ways to solder them without reflow, I just went ahead and scorched them. I took a heavy gauge solid copper wire, bent them in a "U" and soldered one on each side of the LED, bridging all respective anodes and cathodes. The maximum heat for these is a little lower than the others that I used, and it can withstand it for a briefer period. I, however, soldered these for a longer period than the other LED's. So some of these were screwed to start with, as I always test multi-chip LED's after soldering at super low currents first to see if all the chips are lit up. I've got a DIY reflow station in the works. These are nice little LED's, same kind used in commercial emergency lighting.

All that said, I'm also fairly certain that it WAS a few red chips that blew on the 2 or 3 that had one burn out. I wonder if red is more sensitive to temperatures too, or maybe they were the middle contacts of the three and got more heat?

Thanks for bringing up "Yh's lazy man constant current". I had forgotten about that little circuit. It's a little cheaper than the dynaohms I was going to get, and I bet would be pretty useful as a quick and easy bench power supply if you used a pot on it. I'm going to run some numbers and see if I could even use that for this, I'd have to make 3, one for each LED bank, and use the existing resistors as R1 plus a pot to fine tune everything to pre-measured values when using different input voltages...

 

[EAPellow]

It looks like using the LM317 as a constant current driver for varied battery inputs wouldn't work perfectly with the resistors I already have wired to the LED's. Even if I had a dedicated wire to each resistor/LED series, than the LM317 would be driving them just above 50% of their rated currents. That wouldn't be too bad...

But I have each bank of the 3 different LED's wired with one anode to the resistors. How will the LM317's output behave with one wired to 8 resistors who are wired to 8 LED's? Would trying to put one LM317 per bank put all the resistors in parallel, in which case the value of the (8) 100 ohm resistors separately wired to the (8) 4.8mm LED's become 0.00125 ohms? Or would they act as lone 100 ohm resistors in the constant current circuit, supplying about 12.5 mA per LED?

Sorry if it's a dumb question, I'm still getting into teaching myself circuits like this. I would venture a bench test of the LM317, but I only have one on hand. Maybe I have another from an old solar charging circuit somewhere though...

 

[AnAppleSnail]

Right now you're in a voltage regulation regime. This works okay if everything in parallel has the same forward voltage. The red LEDs have a forward voltage (at a.rated current) of about 2v. The blue, green, and white are at about 3.4v. If you voltage-drive an LED about 0.1v over its rated voltage, you may double the current passing through it. This is because diodes have a nonlinear voltage-current response. In other words, connecting a red led in parallel with a blue or green will always kill it at reasonable current...because its lower forward voltage lets it take ALL the current until it burns out.

Draw the circuit out. If each LED is in series with a 100 ohm resistor at 20mA, the resistor dissipates 2v. V=ir is always true. The LEDs could be wired to one common 100 ohm, or each have their own.

 

[EAPellow]

Yeah, I should've planned it out better before putting this thing together. At the very least, I could've gone without the SMD's if I had some other 9mm's on hand. It still works well though, and I threw 4 nearly new AA's on it and it very nicely lit up a whole room of animals that I was looking after during Sandy. Lasted about a whole day until it started getting pretty dim, with part of that usage having just the 4.8mm bank lit up.

I may scrap this as the frame ended up getting some damage, but I like its utility thus far (not that I got to actually mount it to a bench and use it as a work light). I think maybe I'll make the cone detachable. It should allow me to bundle some more wires there for voltage/current regulators rather than having anymore run down the armature.

Say, did you end up getting snow in VA? I heard W. VA got a foot or so.