# Urgent help needed

Discussion in 'General Electronics Discussion' started by JeffS, Oct 7, 2010.

1. ### JeffS

9
0
Oct 7, 2010
Hello,

I need help determing the PROPER/BEST way to light up a bunch of led's. I've done a BUNCH of reading on this forum, but haven't been able to determine my best option yet. These are being installed in a vehicle with a massive power supply (5 batteries, HO alternator, plug in charging capacity, etc). Voltage in the vehicle will range from 11.5 - 14.5 volts. It may rarely dip below the 11.5 from time to time, but I'm not too concerned if the LED's dim or go out at that time. Once the LED's are installed, they will take hours and hours of work to replace should they burn out. As such, I want to get the circuit right the 1st time!

What I need is to figure out what to control, the voltage, the current or both and how to do that.

I have a bunch of NTE956 (lm317) on hand and all the resistors I could ever need. I also have some generic NPN and PNP transistors. I can get other stuff as needed but would like to make use of what I have already.

Specs for WHITE LED's:
5mm FLAT TOP Wide Angle 20,000 mcd LEDs
Reverse Current (uA) : <=30
Absolute Maximum Ratings (Ta=25°C)
Max Power Dissipation : 80mw
Max Continuous Forward Current : 24mA
Max Peak Forward Current : 75mA
Reverse Voltage : 5~6V

Specs for Green LED's:
If=25mA
Min voltage 3.0
Typcial voltage 3.2
MCD 14,000
reverse current w/ Vr=5V = 10 uA

( the led's came from different sources and each source only listed the above info)

I would like, if possible, to wire strings of 3 leds (of the same color) in series and then parallel them together or run each string to a "device"

What I need to know if how to do this the best way possible?

Can I run 36'd led's off one device (12 strings with 3 in series per string?) Do I use one nte956 to control the current and another to control the voltage. Do I need resistors? Will each LED color require different resistors? I am not good with figuring this stuff out, so a drawing would be most helpful!

Oh, and of course, I need to try to get this done asap.....

THANKS!!!!!!!!

2. ### Militoy

180
0
Aug 24, 2010
I’m not quite following what you are trying to achieve. If all you need to do is light the LEDs without burning them out, then all you need are the LEDs and resistors to limit current. The transistors might be helpful if you need to switch them on and off selectively – but are not necessary to set up intensity, etc. Your first step will be to determine a test current that will give you a brightness you will be happy with. This will typically be some current well under the max rated current of the LED. Maybe start with 5mA for each color – select a resistor that will limit to that current at 12V (maybe 2K to 2.4K), then gradually increase current by lowering the resistor value – until you have a safe current selected that will give you an intensity you are happy with. The current should be well under about half the max rated LED current – to keep well within the safe operating area. Then – plan your final lighting array based on limiting current to each LED in the series or parallel string to the selected current. Set the whole thing up for a nominal 12V input. As long as you are operating well under the SOA for the LEDs – the input voltage variation will have little effect. As long as you’re not planning fancy gymnastics with your lamps - your project is based on Ohm’s Law – plain and simple.

3. ### (*steve*)¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥdModerator

25,490
2,832
Jan 21, 2010
What seems simplest is to create a couple of 9V regulators using LM317's (these will stay in regulation down to about 11V). Each one can supply 1A, but I'd not draw more than half of that so you can limit the amount of heat each one generates (they will need a small heatsink).

attached to your 9 volt rails should be strings of LEDs where the forward voltages add up to 7 volts or thereabouts (certainly not much higher). Each should include a resistor to limit the current to 20mA. It may be best to make each string a single colour, but that is up to you.

each string (of say 2 or 3 LEDs will require 20mA, so you can have approx 25 strings (50 to 75 LEDs) per regulator.

Oh, only 36 LEDs. Easy!

4. ### (*steve*)¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥdModerator

25,490
2,832
Jan 21, 2010
Militor's suggestion of one resistor per LED is perfectly fine. Because the resistor drops a large amount of the voltage with respect to its variation, the LEDs will not change brightness too much.

Placing LEDs in strings will increase the effect of voltage variation which is why I suggested the use of a regulator (that you said you have)

The easiest method really depends on what you perceive as easiest.

Whatever you do, mock it up first (perhaps with only a couple of LEDs) and try it out with the range of voltages you expect to make sure the LEDs are appropriately bright, and that they don't vary in brightness more than you think is acceptable.

5. ### Militoy

180
0
Aug 24, 2010
Definitely nothing wrong with using 3-terminal regulators to stabilize a string of LEDs - I just tend to naturally look at the "cheap - dirty" solutions, before the elegant ones. Resistors cost close to nothing.

"Mock it up first" should be a rule taught in all the electronics and engineering classrooms throughout the world.

BTW - I kind of like the handle that you called me - "Militor". Sounds like a robot superhero or something. Just giving you a hard time on the typo - I picked up the name "Militoy" a number of years ago - as a reference to my hobby of collecting and restoring military vehicles. My favorite ride is a 1970 M561 "Gama Goat" that I've owned for around 10 years now.

6. ### JeffS

9
0
Oct 7, 2010
Ok, so I have lots of options.

Then, can you help with the details of wiring up this type of setup as it would be ideal for our layout:

1) 12 parallel strings of 3 LED's in series for a total of 36 LED's per device nte956 (lm317)

2) Run the paralleled groups from a nte956 (lm317) set up for voltage regulation of 9 volts (or should it be lower or higher voltage, like 7 volts or 10.5 volts ???)
using the specs here: http://www.reuk.co.uk/Using-The-LM317T-To-Regulate-Voltage.htm
I get R1 = 240ohms
& R2 = 1490 ohms

That will give me a controlled voltage.

3) Each group of 3 in series will need a resistor to limit current, which should be in the 15mA range x 3 = 45mA per group. I can measure that easy enough to determine the exact resistor to get it exact with a constant output voltage of the above device. Is 15mA x 3 = 45mA where I should be? higher or lower?

and a last question, I assume I'll need to heatsink the lm317's, can I use the mounting hole on the lm317 and screw it directly to a clean piece of metal on the vehicle with some heatsink compound? Or is the tab wired internally to the unit? Or is this just a bad idea all around and it needs a separate heatsink?

And if you have any other comments that may help, please pass em along!

This site rocks by the way. Your help is most appreciated!

7. ### (*steve*)¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥdModerator

25,490
2,832
Jan 21, 2010
note that Vf for white leds is around 3.2 volts, so you can only put 2 of them in a string inless you have more than about 10V available from your regulator (which you wont).

I'm very surprised that the green LEDs are rated at 3.2V Vf. I would have expected them to be lower. Still, if that is correct, you can only have 2 of them in a string as well. So that means 18 strings of 2 LEDs

9V should be fine until the battery voltage falls below about 11 to 11.5V. The LM317 probably needs about 2V at the sort of current you'll be using.

Group of 2 now and if you want 15mA, then it's just 15mA -- the devices are in series. The resistor needs to drop (9 - (3.2 + 3.2)) volts at 15mA. Thats 2.6V @ 15mA, so the resistance is pretty close to 180 ohms.

No, the tab is connected to Vout, definitely NOT what you want to connect to ground.

You need to insulate the device (electrically, not thermally) from the heatsink. There are mounting kits that you can buy to do this.

You could mount it on the car body, but make sure that the part you mount it to doesn't get too hot.

You should really take precautions against voltage spikes doing damage to your regulator.

This might be as simple as a low value resistor (say a couple of ohms) in series with the input power, and a 24V zener -- preferably 5W -- across the power rails following the resistor. The voltage rating of the zener is not critical. As long as it exceeds say 18V and is less than the maximum voltage for the 317 by 5 volts or so.

We like to think so, thanks.

8. ### JeffS

9
0
Oct 7, 2010
Ok, makes sense. Rarely should we get below 11.0 volts in the vehicle.

Ok, I follow that also. I'll measure to be sure. Is 15mA good for the specs on the LED's I have? Should I go higher or lower? Keeping in mind I need this to work for a long time.

It's a good thing I asked. It looks like the nte956 has a mounting kit included ( a flat piece of plastic to mount behind the device and a small plastic spacer/washer to mount in front to keep the screw from contacting the tab. Do I need any sort of compound when attaching to a heatsink?

Well, I would like to build this once and have it work reliably for years, and yes, it's possible the regulator on the alternator goes whacko or something else spikes and causes a problem. The nte956 (lm317) has specs of Vo 1.2V to 37V and Vi of 3V to 40V.

Would a part like this: "NTE5137A, Diode, Zener; 24V(Nom.)@50 mAVoltage, Zener; 5W" be fine?
For the low value resistor, what kinf of power rating does it need? I have 1/4 and 1/2 watt resistors on hand and a few 5 ohm 50 watt big guys on hand also.
and can this one circuit protect all the lm317's or do I need one per lm317?

I follow all the stuff for the led's and lm317's, but could you do a quick drawing for the protection part of it and how it runs into the lm317 (or multiple lm317's)?

Thanks again!

Last edited: Oct 7, 2010
9. ### (*steve*)¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥdModerator

25,490
2,832
Jan 21, 2010
Is 15mA ok? That's why you mock some up and try it out

Given the relatively low dissipation, you can probably get away without using any heatsink compound. If the insulating washer id a blue-grey soft and flexible piece of what appears to be rubber, then you don't need it -- the soft rubber does the trick itself.

10. ### JeffS

9
0
Oct 7, 2010
Maybe a better way to state the question, considering the specs on the LED's, what mA draw would be safe for long life and good brightness? Can I go up to the 25mA they spec and have long life? I know they will be bright at 25mA, but how about the life expectancy?

On the zener, should I wire it up like the diagram at the bottom of this page:
http://www.kpsec.freeuk.com/components/diode.htm
and if so, how do I calculate the resistor size? You had said low resistance, so could I use a 5 ohm, 50 watt resistor to feed all the lm317's (There will be like 8 lm317's in total, 4 which will run 36 led's each and the other 4 which will only run a couple led's each). And would the zener be large enough to handle all of this?

Or should I use one zener and one resistor (say 20 ohm, 1/4 watt) per lm317?

11. ### (*steve*)¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥdModerator

25,490
2,832
Jan 21, 2010
The junction temperature defines the life of the device more than the actual current. You'll need to read the specs to find out what they say. In any case, the LEDs don't suddenly fail, they just get dimmer with age. Their lifetime is typically defined as when their brightness falls to some percentage of the "new" brightness.

Mock it up. Is 15mA bright enough? Do you need 25mA? What is the lifetime? if you reduced it from (say) 50,000 hours to (say) 10,000 hours, would it make a practical difference?

Yes, the circuit you have is appropriate. Calculate the resistor to drop approx 1V at the normal current the LEDs will draw (you may like to measure it first (without the zener and resistor) and then calculate a value.

You should have one of these for each regulator (note that the resistors will vary in size if there are substantially different numbers of diodes attached).

Note also that if you have a very small number of diodes (total current less than 100mA) then you won't need a heatsink for the 317.

The wattage for the resistor should be calculated based on the current (use I^2R), then pick a resistor of approx twice that rating.

Because the resistor will drop your voltage, either the regulator will fall out of regulation earlier, or you should drop the regulated voltage a little (maybe to 8V).

You should also ensure that you have a fuse protecting all of this. The zener diodes will probably fail short circuit if/when they eventually fail, and a fuse will prevent smoke and (possibly) flames. Calculate the total current consumed by all devices and use a fuse of twice this value.

A zener diode may not be the most appropriate part to use here. I considered recommending a transorb (aka TVS diode). The other problem is negative going transients which threaten the regulator by reversing the voltage briefly. A diode in series with the power input will prevent these issues, but will also drop an additional 0.7V to 1V. Placing a regular diode in parallel with the zener (with the same polarity) is a compromise that may be effective enough. I would recommend something with a fairly significant surge rating just to ensure a long life. A 2A rectifier diode is probably the sort of think you'd be looking at.

12. ### JeffS

9
0
Oct 7, 2010
So I mocked up different scenarios, starting with a few LED's then up to about 40 led's per lm317. The major testing was done on a bench with a 5A 13.8V supply.

I have the lm317 set for about 8V of output and was stable with an input voltage from 10V to 14.6 volts so that part seems stable.

I have banks of 2 LED's wired in series with a resistor and then each bank w/ resistor wired in parallel to the lm317. What I found was that with only a few LED's I could use a 150 ohm resistor and get about 18 mA of draw (measured w/ a fluke meter on 300mA setting) which had an acceptable brightness and should give long life. As I increased the total number of LED's, I found total draw increasing at a lower than expected rate (I expected the current draw to double every time I doubled the number of LED's). For example I measured:
2 led's (1 bank of LED's), 18 mA
6 led's, (3 series banks in parallel) 51mA or so (17 mA per bank)
20 led's (10 series banks in parallel) 150mA or so (15 mA per bank)
40 led's (20 series banks in parallel) 200 mA or so (10 mA per bank)
(note: I don't have my paperwork with me, so the numbers above are approximates, but you get the idea)

I had to go down to 100 ohm resistors to get the total current back to where expected and get the appropriate brightness. Is this from the resistance of the wire and connections? I should note, the total feet of wire for the 40 led's was about 30 feet of 18 awg wire for (+) and the same for (-). All connections were soldered (properly).

Is there something I missed here? Or would this be as expected?
The NTE956 (lm317) is rated for 1.5A so I was well under it's capacity and also under the power supplies capacity.

I want to get this part figured out then I'll get back to protecting the lm317's........

Once again, thanks to all those that have responded!

13. ### (*steve*)¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥdModerator

25,490
2,832
Jan 21, 2010
You may need to take into account the resistance of 60 feet of cable, although I am somewhat surprised at the magnitude of the effect.

Measure the voltage across the resistor and 2 LEDs (which I presume are close together). If any of these show significantly less than 8V then cable resistance may be your problem.

By my calculation 60 feet of 18 awg wire has a resistance of less than 0.5 ohm.