Problems with Dual Brightness (Current) LED Tail Lights? Eagle Schematics!

Ok, I have tried this circuit, and it doesn't quite work as expected... Basically I have 8x 350mA capable LEDs 2 sets in parallel of 4 in series... I THINK... I am not sure exactly the load... But I'm trying to get 2 sets of brightnesses for my tail lights. And they're not exactly working how I expected.

First pic is the full brightness, it should be 640mA by the calculation of the lm317 which I am using, which is R = v/A so 1.2/.64 = 1.875 = ~1.9hm:

If each LED can take 320mA I think 640mA is correct, but I'm not sure if how I wired them is really 2 parallel strings of 4 in series. This is just the way I had to mount them based on my thermal and space contraints... pics to follow of my rationale.

**broken link removed**

Am I really getting 320mA to each LED? I'm a bit confused on the parallel, but I think I am?

Second I intended to run them at 2 different brightnesses, basically 320mA and ADD another 320mA diode isolated of course... but just adding two constant current sources together to get 640mA. It doesn't really work that way, it gets barely if any brighter. This is how I had that wired:

**broken link removed**

Now I was thinking it was possible that the two chips were not functioning properly, crossing each other out or something, so I tried just 1 LM317 with two 3.9hm: resistors in parallel to get my 1.9hm: but connecting and disconnecting that 2nd 3.9hm: resistor which SHOULD have netted me 1.9 ohms just didn't do anything at all... so I'm confused. I thought you can parallel resistors to get R1+R2/2 ?

Since I didn't get "half" of the brightness, I Thought maybe it was a non-linear brightness for power, but according to the SunLED Apollo PDF, this chart shows that I really should be getting about 1/2 the brightness for half the current through these LEDs...

**broken link removed**

So what am I doing wrong? Thanks for all the help. Bill

*I finally got Eagle down pretty good, it's an amazing program compared to Express PCB ! the auto routing feature is amazing! Now I still can't get my 555 PWM projects running, but that's probably a logical error, not the program's fault... but the Rat's nest and Auto Router features are super cool.

"2 parallel strings of 4 in series" doesn't make any sense. The way you have them wired is more like "4 groups of 2 parallel diodes in series with each other". But that's just language.

Anyways this is what's going on...

1. Adding another voltage source will give you more "current capability" but the regulators aren't going to force more current down the throats of the diodes. THey will just be able to provide more current IF something draws it.

2. The voltage sources you have (whether it be one or a few in parallel) just need to be enough to supply the maximum current you expect. What you do is you have resistors in series with the diodes (each diode, or maybe one resistor to every two diodes or something). The resistor will take up the excess voltage drop being put out by the regulator that cannot be taken up by the fixed voltage drop across the diodes. By varying this resistor you vary the current running through the resistor (and therefore the LED), thereby changin brightness. What you could probably do is make a circuit with all the resistors for the high brightness level, and then at the regulator output stick a resistor that can be shorted in and out of the circuit to change the brightness between high and low.

3. In your schematic, Each "dual-group" of diodes will be progessively darker (I think). Regardless, you they should all be connected in parallel with each other. (Sorry, no simple 2-wired chain of LEDs to lead around your car).

4. The voltage drops across LEDs varies around a nominal value...so when you parallel them take some care. Some balancing resistors or something might be needed.

It's late so some of the stuff I said might be inaccurate. Read with care.

EDIT: A way to change brightness other than shorting a resistor and the regulator output in and out of the circuit is switch in and out another voltage regulator (if the voltage regulators can handle this) which will increase the voltage to the circuit, increasing the voltage drop across the current limiting resistors (diode voltage drop stays fixed) thereby increasing the current through all resistors. They probably can't be identical regulators though since that would double the votlage and double the current which would make for

A. very dim and very bright modes
B. moderate to bright modes and a burnt out mode

yeah, 4 groups of 2 Parallel... see that's what I was getting confused, I couldn't wire them in the standard parallel/series strings because of said constraints, so I am not sure what I came up with?

The LM317 is a constant voltage regulator, and by tying a resistor from the adj pin to the output pin, you get a constant current source. Sorry should have been more explicit. There shouldn't be any need for current limiting resistors I thought, as long as I have the correct current.

If I just had 1 string of 4, 2.5v LEDs in series, I would have 320mA set on the LM317 and all would be dandy, I've done this before, and it automatically adjusts the voltage for the LEDs between 2.2 and 2.4 depending on each particular LED's requirement... But since I threw these extra 4 LEDs in, I'm not sure what I came up with? And I don't have the background to go back to the math... :-/

I don't know when you replied but I was editing my first response to this thread.

You NEED current limiting resistors since a diode basically appears as a short. Stick 5V across a short and what happens? BOOM! YOu need resistors in series with the diodes to limit the current.

The voltage acros a didoe always stays the same (let's say 1.1V). That means if you stick 5V across the diode and a current limiting resistor, then 1.1V will be taken by the diode, and the remaining 3.9V shows up across the resistor. Using V=IR, you can find out how much current is flowing through the resistor (or size the resistor to control the current through it), and in the process control the current in the diode, since resistor and diode are in series.

For V=IR
V is the voltage across the resistor. It is Vsource-Vdiode.
I is the current through the series resistor and diode combo.
R is the resistance.

Can't you make all the diodes in series and bundle all the wires together so they can lead around the same path in the car? You can also do the "one big cable around the car" and tap the LEDs and small wires into the cable as you go. It might look different on the schematic...but you can always run wires alongside each other to make it look like a chain in real life (think of christmas lights).

I'm going to assume your supply is 12V. I'll also assume these LEDs each drop more than 3V. The first problem is that the total voltage across all the LEDs is over 12V. The LM317 (from a 12V supply) can only regulate up to 9.5V at it's output. I know it's regulating current, but it just can't get it's output voltage high enough up there for the LEDs to suck 300mA through. When using the 317 you must know the forward voltage drop for an LED, and use a quantity of diodes whose total drop does not exceed the power supply voltage less 2.5V.

Then there's the parallel diodes. That's no good, because two LEDs rarely have the same voltage drop for a given current. One will always drop more than the other, and the current drawn through each will be significantly different. You could have two independent chains of LEDs powered (in contrast to a single chain of pairs or paralleled LEDs) from a single regulator, to even out the difference, but there will still be a difference. It's best to have two regulators powering two chains. Careful - if you do succeed in passing 640mA through a parallel pair, one of them is likely to pop, because the 640mA is not shared equally between them - one of them will receive more than a fatal dose.

I have not tried it, but I see no problem using two 317s as you did to double the current. But as I said, the first can't do it's job because it can't get its output voltage high enough. The other can't either, for the same reason. Connecting two together does not solve the problem, which is either insufficient power supply, or too many diodes in series. Adding the second regulator cannot result in an increase in current unless this voltage problem is solved first.

I suggest you try the circuit with a single regulator and single LED, configured to provide about 100mA. Measure the voltage across the LED. This will give you an idea of how many LEDs you can have in a chain. If the supply is 12V then you can use up to 9.5V that is (12V - 2.5V) divided by the LED's forward voltage.

Then, because this is such a low current, it is safe to connect another LED in parallel with the first. You'll see the difference in current through each causes a difference in luminosity between them, which I am sure you do not want.

Back to the single LED, now connect in the second current regulator (another 100mA) and check that the brightness increases. If you have another 3.9hm: resistor, put that in series with the LED so you can measure the voltage across it. Then with Ohm's law you can calculate the current accurately and see if the current really does double. If it doubles, then your idea of connecting two regulators together is OK.

Now the problem of powering 8 diodes. Parallel connecting diodes is not the way to go, in my opinion. Using the 317 with a 12V supply is your limiting factor here. With 3V LEDs, you can connect 3 in series (for a total of 9V dropped), and that should be OK. I suggest using three regulators to power groups of 3, 2 and 3 LEDs. Two groups (3 & 2) are powered constantly, while the third group of 3 is illuminated when the brakes are applied.

YOu could parallel the diodes if you stick an oversized resistor across each diode individually and measured the voltage drop across, then pair up a special balance resistor with each diode to make the voltage drop across them equal. I can't think because it's late but I think this only works for a fixed amount of current, and not more (due to your two brightness levels)...that is unles you switched in another special balance resistor for every single LED which is not practical.

The LM317 has a typical dropout voltage of 1.75V at 640mA, but they don't say what is its max. Plus the 1.25V across the current-sensing resistor equals a voltage loss at the output of 3.0V. It could be higher.

Your vision's response to brightness is logarithmic like your hearing's response to loudness. So half the power and double the power in the LEDs will be just a little less bright and a little more bright. Try changing the current 10 times for it to look half or double as bright.

Thanks for all the responses, I will try and answer them in turn, since they are all valuable information. I had a MUCH longer reply before, and my computer keeps getting the BSOD after plugging in a USB card reader... so it'll be a bit more brief this time.

First some pics of my project... I know it's a bit "rough" but given the challenges, it's totally functional at least... now just to get the control part down.

The SunLED Apollo chips are PLCC-4 and require the cathode/anode to be heatsinked... basically they rely on a metal core board, but if you don't have that, you have to find a slug of coppy to mount them on, and then electrically isolate that from a bigger heatsink. Hence the pink sil pads... it was a challenge to solder them on. I had to seperate the copper slugs (made from copper tubing flattened and sanded) from the aluminum heatsink, otherwise I would have never been able to solder anything onto the copper slugs.

**broken link removed**

**broken link removed**

I'm not too worried about running 2 in parallel since Lumileds specifically reccomends that as there really isn't any alternative with a limited Voltage Supply... There are LED drivers which fix this, but I'm not quite to that point... wanted to do something on my own anyways.

**broken link removed**

So running strings in parallel with ladders across every few LEDs will help equalize the differences in forward voltage drop. I think my creation is something like figure C. Lumileds PDF is here:

https://www.electro-tech-online.com/custompdfs/2006/12/AB20-3.pdf

The SunLED Apollo chip is a 2.5v Vf at 350mA... So 4 in series is 10v Vf-total. And 3v for the LM317. I measured 14.5v at the battery with the motorcycle engine running, so I should be ok, no? I measured the voltage drop across the LEDs, and they are almost always 2.3-2.4v on spot...

**broken link removed**
**broken link removed**

I think my problem is that the LM317 is only rated for 1.5amps... Am I overpowering this chip? I figured I was using 700mA for the LEDs, and maybe some overhead for the internal resistance of the chip, but still under 1.5mA?

Now my test Power supply is only 11.5v @ 5amps so that may be sagging a bit. I have a 16v @ 7.5v PSU I will try out to see if maybe that was the problem...

Part of my problem is just the limited space, and electrically isolating, while also heatsinking them... Tapping the aluminum block was a pain too... So I don't really want to redo everything if I can find a solution.

Now it says I can run the LEDs at 1/10 duty cycle for .1mS... I will have to test this out with my PWM kit I got... Could try that...

I will have to test the logarithmic scale of vision as well, probably do like 50mA and 350mA or 100 and 700mA something in that range. Brake lights are supposed to be in the 7:1 brightness range, so I'm sure the math doesn't work out precisely, but 100:700 should work. But when driving them with such low current they tend to become orangish not red... so PWM is necessary I think to get that low of brightness...

**broken link removed**
**broken link removed**
**broken link removed**

I know I missed some comments, but this is the 3rd time writing up a long post, and am brain dead, and need to tinker a while to resolve some of the questions you all have brought up. Thanks again, cheers

Thanks all for the help, I figured out my problem with the LM317, I was reading the schematic wrong, and was running them with the adj pin shorted to the outpin so of course running two together didn't make them any brighter... So I finally figured it out, and all is good. I have put together a little article for some other sites I friequent, non-electrically saavy, or at least as saavy. Let me know what you think. Be kind, it was a huge undertaking, and I am sure I made some mistakes, so I'm asking for corrections if you have constructive criticism. Imagine I'm taking this back to people who have never heard of LEDs... *i started a new post since I worked so long on it*

Ocelaris, here's a suggestion for your dual-brightness setup.

Instead of two regulators, use one regulator and a transistor on the current-sense resistors ... one set of resistors is always inline with the load, providing the basic 'running light' current to the leds. a second set of resistors is connected in parallel to the first via a transistor (probably mosfet to avoid voltage drop?). when the second set of resistors are in parallel with the first, the current limit from the regulator is increased.

here's the diagram:
**broken link removed**

The advantage of that would be less wasted power? simplicity? I suppose I wouldn't have to heatsink the mosfet because it is only providing a very small amount of feedback current to the adjustment pin?

How would I calculate what R2 should be? Since the total sense resistor is the parallel calcuation:

(Rparallel = 1/((1/R1)+(1/R2)...) )

So say [let me consult my handy dandy LM317 chart]

**broken link removed**

If I wanted to end up with 1.25 ohms for 1000mA current, I would have normally the R1 = 2 ohm = 620mA,

Then to get a total 1.25 ohm in parallel I would use a 3.5 ohm (roughly) to get the total R1+R2 in parallel = 1.25 = 1000mA Correct?

And for R3 I would use whatever the Hfe of the mosfet is, to make sure the current of the gate is say 100mA when the brake light switch hits 12volts ?

I'm glad you brought this up, because I'm just starting to use transistors in a meaningful way, but I have no idea how to select a particular transistor. Do I just calculate what my Hfe needs to be, based on my current requirements for that particular transistor?

Like you picked a Mosfet because the input because I have to assume the Source to Drain has no voltage drop? Or because the gate has an almost infinite impedence so there is no current drawn? I know this is a lengthy subject, so don't feel compelled to elaborate more than necessary. But I do sincerely appreciate the help. I sort of summed this project in a new thread, that I basically am sharing with hidplanet.com members. I sort of eluded to thanking you in the beginning of the thread, because I know you had helped in a few threads of mine.

https://www.electro-tech-online.com...brightness-leds-and-thermal-management.26364/

Thanks as always! Best, Bill

And for R3 I would use whatever the Hfe of the mosfet is, to make sure the current of the gate is say 100mA when the brake light switch hits 12volts ?

Click to expand...

You're confusing your transistors, that isn't a MOSFET, it's a BJT and yes you do need to select an approprietly sized resistor.

I think a mosfet would be better suited here so the voltage drop of a bipolar doesn't affect the feedback to the regulator.

choose a modern fet with a low RdsON ... something like 50mOhms should be pretty easy to find and cheap. At one amp, you'll drop 50mV across the mosfet, hardly enough to effect the feedback to the IC. same goes for heatsinking, which shouldn't be required.

easiest thing I can think of for figuring the resistors is this:

choose what resistance for full brightness ... 1.25v / 1a = 1.25 ohms. (four 4.7 ohm resistors in parallel is close enough)... and presto matho there is the answer. the first two resistors would give you 2.35 ohms, for roughly 530 mA of current. use the transistor to switch in the second set for the full amp of current.

if you use a nfet transistor, it should be on the opposite side of the resistor (the low side)

updated schematic attached