Small Constant Current Sources

[covert1]

Hi everyone - gotta start off by saying, great place! With lots of discourse, I'm sure to learn even more. Hence my reason for joining the fray.

Short background - I'm an amateur to intermediate LED tinkerer, but lately I have been succumbing to my interest in it such that it's starting to become a hobby. What got me started in all of this, was a quirk on the car I drive. It has an issue with the license plate lamps dying prematurely. Without going into the discussion of possible failures, I must admit I had an idea to do something, but got lazy. Then, one evening, I was pulled over by the local PD, for none other than my license lamps being out. That was mid-late March. Then I said, "damn it, I'm going to do this" and certainly had the motivation to do so, as I got a fix-it ticket with a 14 day window for visual inspection and signature.

A few $100 later (plus research, trial and error), and, being part of a car discussion forum, I mentioned my work to the masses and now others have chosen to sport these license lamps. Fast forward to now: I had another idea - why not retrofit my dome light with a nice, "pure white" array instead of the common as dirt festoon bulb? So then I got started making dome lamp modules. In the last three weeks I've obtained a ton of resistors and even a bench-top power supply. And, thanks for the resistor parallel/series array calculator, because what I learned with it, and what I observed with my setup, I think I have much more insight as to why some of my creations meet an earlier demise than others. That list begins with the fact that my car seems to allow 14.5V while running and 13.8 on the battery alone.

Now, to the point: My two designs, utilize 5mm LEDs. The License Lamp has a string of 4, 5mm white LEDs rated 3.4-3.8V, 20mA (sorry, V is from memory) and the Dome Lamp uses 24, 5mm LEDs, in a configuration of 6 parallel strings of 4 of the same LEDs. This one consumes, then, 120mA. The working area on the board inside the license lamp is maybe 2/3" by 1.2", whereas the dome lamp board is about, say, 1.2" by 2". Both are reasonably confined (my judgement) when assembled (board + LEDs + resistor(s)). This small space is important, because... after observation with my LEDs on the power supply, I really need to invest in something TINY, which can limit current.

To the question: I happened to be searching for some Superflux LEDs, but stumbled upon a German website:
http://shop.dotlight.de/

Does anyone here, perhaps, have any experiences with them? The reason I ask, is because I saw that they sold various preassembled (and SMALL!!!) current limiting PCB's. I almost got giddy, because it seems exactly what I need. However, I tried looking for these things (briefly, since I am at work) and didn't really turn up anything of use in the states. Of the brief results I saw, the output current was WAY too high.

They offer a wide array: http://shop.dotlight.de/shop/index.php/cPath/100

They provide options that are lower than the common 350/700/1000 that I kept seeing. And, did I say they're tiny? I can easily stack these on top of my project.

But, that's why I am here - I wish to get some opinions from those of you who with much more expertise. Like others out there, my interest is in being able to purchase the preassembled thing, instead of putting together the components myself. The latter would result in something bad, just like the DifEQ class I took years ago. 

Can these things be had, ready made, in the states? If so, does anyone know where? Also, if these look good, is it possible to do a group buy?

Thank you in advance!

[Rob]

Welcome to the boards.

There's a lot to deal with in that first post...

That list begins with the fact that my car seems to allow 14.5V while running and 13.8 on the battery alone.

Yep, you have attained enlightenment on this point.  Constant-current design is a great advantage in this environment.

I really need to invest in something TINY, which can limit current.

I'm with you.

http://shop.dotlight.de/

Does anyone here, perhaps, have any experiences with them?

Not I. I'll check it out though.

Like others out there, my interest is in being able to purchase the preassembled thing, instead of putting together the components myself. The latter would result in something bad, just like the DifEQ class I took years ago.

Have more faith in yourself here, covert1. You've proven yourself capable of building a working LED array--this isn't really a lot harder.

So wrapping up, you'd like a small constant-current regulator targeted at a lower current than 350 mA. Correct me if I have that wrong.

As far as US offerings, I think that LEDSupply has a nice selection for you to look over:
http://www.ledsupply.com/led-drivers.html

The flashlight crazies (and I mean that respectfully) are the leaders in current-regulated little circuits, but they tend to be concerned with stepping voltage up rather than down. Having said that, I've never seen any smaller than those mikrokon ones on the German site:
http://shop.dotlight.de/shop/index.php/cPath/100_185

The groundrules for your appliaction are different than a flashlight in significant ways, that's the reason I think you're not seeing something tailor made for you. So if you can stomach it, I still think DIY is the one true path for you.

For the price of one LM317LZ (datasheet link for your convenience) and one resistor, you have a constant-current reg that will tuck into a small corner of your board. You don't need the TO-220 package as long as you a) deal with lower currents (<100mA in this case, but you can stack these in parallel), and b) maximize the series dimension of your arrays so that the voltage drop across the reg is minimized. Plus, cheap. If you're willing to tackle surface-mount, there's smaller options too.

[covert1]

Thanks for the response, and offering to look into the products on that site.

I checked the Led-Supply site, but those offerings, again, looked inappropriate for what I needed. Maybe when/if I design something much larger, but not quite yet. Theirs were 350/500/750/1000mA.

I have a minor concern about the posted datasheet. You had mentioned that I can accomplish what I needed with the LM317, with one resistor. I found the diagrams on page 9 of the sheet, and did the basic calculations. Now, presuming that I have to use the larger package TO-220 (which I can probably get away with, on the dome lamp module) can I ignore the restriction underneath the "basic current limiter" figure? It says that the Resistance, R1, needs to be between 12 and 240 Ohms. If I want 120mA for my 4(series)x6(parallel) array of LEDs, then I need a 10 Ohm resistor.

Is that limitation there because the sheet is for the "L" and not the "T" model? Or, can I ignore that? Oh, another thing I don't understand - does this thing leech any voltage of its own? If I start at 14.5V, put this regulator together so the current stays at 120mA, and if the 317 doesn't factor in there, I'd have 14.5V / 4 LEDs = 3.6V per LED per "column" (within spec)?

And, now that you got me thinking of it - would it be possible to use the 12V regulator (7812?) in the same fashion? For the license lamps, which only have 4 LEDs @ 20mA? If I can do that trick here, then maybe I have much less to worry about.

My main motivation for all the questioning, is because I need to streamline how I put these things together. The fewer things that I have to assemble, the fewer the mistakes... and greater the output of work

Anyhow, I definitely plan to play with these things when I get home - Radio S*ack is on the way, so I can pick up one of each to test, and later, if/when I figure it all out, I can order more online. Thanks again!

[Rob]

The current limitation is a package thing. So you can use a LM317T to play with on the bench and use a LM317LZ in the finished product.
LM317T = TO220 package, 1.5A
LM317LZ = TO92 package, 100mA

Yes, it does "leech" about 1.5V to do its regulation work. That's perceptive of you to catch, and I meant to point it out in the first reply. There are regulators available with a lower "dropout" voltage, and they would work fine as current regulators if that was a concern. I think 3.6V forward voltage is high, but get out your meter and convince yourself of that.

If you need more than 100 mA, just use two or more regulators. Alternatively--underdrive the LEDs.

The 12 ohm lower limit is where you hit the 100 mA output current limitation. You can maybe squeeze another few mA, but for 50 cents I'd just add regulators.

[covert1]

Oh - neat. If the thing saps up about 1.5V, then 14.5-1.5 = 13V. That would give me 3.25 across each LED, which I think is all right! Can't wait to get home and rig this stuff up, now. If it turns out right, then I'll be ordering a bunch of 50 cent safety valves.

[justDIY]

Oh - neat. If the thing saps up about 1.5V, then 14.5-1.5 = 13V. That would give me 3.25 across each LED

that's not exactly what Rob meant.

The LM317 is a regulator ... it requires 1.5v to do its job;  so if you wanted 5 volts output, you need at least 6.5 volts input but could go as high as 35(?) volts input

but the method Rob perscribes for the lm317 is as a constant current regulator, so voltage does not apply.  You set the current you want output, and the LM317 dutifully obeys by providing that current... voltage will be variable depending on the requirements of the load.

so if you have 4 leds that require 3vF each and you wire them in series to keep the current constant; you will need a minimum source voltage of (3 * 4) + 1.5 = 13.5 v to deliver the current you've selected

if you only have 2 leds that require the same 3vF each, wired in series, your supply needs to provide the LM317 with at least (3 * 2) + 1.5v = 7.5 v ... but that is a minimum.  If you supply your LM317 with 14v, it will "eat" the excess voltage and turn it into heat

so you want to build LED strings that require forward voltages of approx 1.5v less than your intended supply voltage, this way the LM317 will not have to dissipate or 'eat' too much excess voltage

[Rob]

More thoughts: Have you looked at these 1W LEDs at all? For something like a dome light they are ideal, honestly.

[covert1]

Thanks for the clarification, justdiy. In fact, I'm glad you said something, because now I need to characterize my LEDs before I string them together, if I want this idea to work. I found that many were hitting their nominal current at slightly lower voltages (like, 4 supposed 3.3V LEDs in series only consumed 11.8V but the current was near 20mA) So in the case of the car, I need to be sure that the voltage drop of the LEDs is 12.3V or less, so that the thing will light up when the car isn't running the alternator.

Research on 1W LEDs is a work in progress. Maybe worthy of its own thread In terms of a small, personal business/hobby, it's more cost effective for me to stick with the traditional LEDs for now. If 1W LEDs were like $20/100qty, then, hells yes I'd be all over it.  I might have to post pics to explain some of the other issues.

I bought a small sampling from Mark over at Quickar and was impressed with what I saw, but I didn't get to brain storm, since other hobbies got in the way (bowling, two nights a week). His Baby UFO's might be usable, but I need to get pics this time so I don't forget how the beams look.

[Rob]

I see an LED bowling ball in your future....

[sanjed]

I see an LED bowling ball in your future....

Excellent idea.  You could cast your array in resin, along with the batteries and a kinetic charger, and then somehow have it all set in a bowling ball.

If you were feeling really clever you could put a microcontroller in there, and some buttons for choosing sequences in the finger holes.

[justDIY]

kinetic charger! .... bowl enough games to store energy for moonlight bowling after the xmas party

or maybe a induction charging interface

and don't forget an infrared com port for downloading firmware updates

[covert1]

Actually, someone out there beat us all to that. Prior to the winter leagues starting, I was at one particular bowling center, thinking about their Thursday league. While waiting for the meeting to start, I noticed someone on the far right side of the house - they had a ball with flickering red LEDs on it. Just a handful were embedded, but easy enough to be seen in contrast to the black plastic color of the bowling ball. It was like those keds shoes that some little kids get, where the LED lights up with every step they take.

With what I'm going for, I don't think I need to go quite that far. I certainly don't with to mess with the equipment I have (pair of 300's) but in order to offset the slight envy of the kid with the LED ball, I had to get a cow print spare ball, instead  So far, it's quite the attention getter.

Now that I'm refreshed from a long nap, it's time to play around with LEDs.

[cpemma]

Oh - neat. If the thing saps up about 1.5V, then 14.5-1.5 = 13V. That would give me 3.25 across each LED

...so you want to build LED strings that require forward voltages of approx 1.5v less than your intended supply voltage, this way the LM317 will not have to dissipate or 'eat' too much excess voltage

With the CSS application, you lose voltage in two places; the approx 1.5V drop-out across the internal transistors, etc, still applies, and by definition you lose a further 1.25V across the current-setting resistor. I think Rob's '317T CSS for Luxeons' article suggests allowing a 3V head over the LED requirement.

With a more complex CSS circuit (discrete transistors) you can reduce the loss, or you can just fit a low-drop-out 12V regulator and take it from constant voltage with a current-limit resistor. TBH, I'd go for parallel strings of 2 LEDs in series with a resistor on each string, pick a resistor to give a safe current at 14.5V and the LEDs won't be much dimmer at 12V.

eg, 2 whites @3.6V V_f with a 300R resistor; 14.5V gives 24mA, at 12V current is 16mA.

[covert1]

Yeah, I learned more about the circuit over the last three day weekend and stumbled on that very issue. The car's system did not allow me to overcome that 3V threshold. Resistors will do for now.