View Full Version : OT: Basic Question About LED’s

Mad Scientist
08-18-2001, 03:32 AM
As Dvido mentioned voltage spikes are going to be your LED’s biggest danger.
The way to avoid this AND to control the brightness at the same time is to use a voltage regulator. (LM317 Radio Shack aprox. $2) This plus a 5K-ohm trimpot and 240 ohm resistor and you can build a voltage regulator that will eat any incoming spikes and accurately maintain the output voltage to the LED’s. Also with this no current limiting resistors are needed.

The Lm317 is a three-legged device (Looks like a power transistor) and can control 1.5 amps.

Power comes in on one pin and goes out on another. Between the output and the third pin is the 240-ohm resistor and from the third pin and ground is the trimpot.

J Tiers
08-18-2001, 07:23 AM
D'y know.... voltage spikes have yet to blow ANY LED I have seen.

There is indeed a max current on an LED, but in general, they will take quite a bit of current, more than the continuous limit, if it is in the form of a short pulse.

There are applications I have seen which happily pulse the LED with 2, or 3 , or more times the max continuous, and operate for years. They use very short pulses, and rely on the eye's "peak detector" to increase apparent brightness.

An LED functions as a voltage regulator, it is a fwd biased diode..... The volts are lost mostly in the dropping resistor, IF you don't use a max-length string, with the acoompanying very small value resistor.

I'd be very suprised if any "voltage spike" that is in the forward biased area would kill them

If you have a max "load dump" pulse, you might get closer. I forget the length of time spec'd for those, but it is longer. Still, a reasonable resistor will allow accepting even that. It is only about 40V for 12V autos.

BTW, the voltage regulator chips mostly have a 35V limit.... lower than teh load dump voltage. The regulator will need more protection than the LED...... unless you use a type with 65V limits.

[This message has been edited by J Tiers (edited 08-18-2001).]

03-01-2006, 02:57 PM
I have a panel that I am lighting with a total of 18 - 2mm x 5mm rectangular LED’s on pigtails. I don’t have any specs on the LED’s but they seem to like 2.2-2.5 volts. They will be wired into a dimmable vehicle dash lighting circuit so I have about 14 volts to supply them.

Should I wire 3 banks of 6 LED’s in series and parallel the banks, wire in parallel and reduce the voltage with resistance, or perhaps something else? The first option seems to make the most sense, but obviously I am far from expert when it comes to electronics.

I wouldn’t mind adding a little pot so that the brightness could be fine tuned to match the other lights. How much current will 18 of these little guys draw?

03-01-2006, 03:41 PM
calculate for each red LED drawing between 15 and 30 mA.

I = V / R

R = V / I => 14 volt / .030 Amp

R = 466 ohm resistor for brightest

03-01-2006, 04:08 PM

LED's are current devices and are typically all wired in parallel with one current limiting resister for each LED. Wiring in series can result in uneven lighting. A single dimmer pot can be used to control the overall light output.

03-01-2006, 05:05 PM
18 LEDs at 20 ma each will be about 5 watts of power on 14vdc. Unless you use a wirewound pot that's too much for an ordinary pot to dim. Usually for dimming large numbers of LEDs such as in displays etc pulse width modulation is used. This is trivial with a 555 timer IC and a small power fet.

03-01-2006, 05:19 PM
You should be able to string them in series groups of 4 (maybe 5 depending on LED) each and a dropping resistor (for each group). Now the resistor is only dropping about 4 - 5 volts at 20mA so 1/4W resistors should be fine.

The 4 groups will then only draw 80mA or 1.2W on the 14 volts.

I'd combine this with Evan's 555 pulse width timer and an n-channel, power mosfet thru which all of the groups return to ground.

That way, everything is nice and cool http://bbs.homeshopmachinist.net//wink.gif


[This message has been edited by nheng (edited 03-01-2006).]

03-01-2006, 09:34 PM
You're wiring these into a vehicle, so you basically have 12 to 14v to play with. Four leds in series will need about 9 volts to light, according to your data. Three in series will need about 7 or so. Three in series will give you the option to make 6 strings to wire in parallel, thus 18 leds can be powered decently. No other combination except 9 strings of 2 leds will allow you to light all of them without having to have a resistor for each one. Use 6 strings of 3, and 6 resistors. For each string of three then, add a series resistor of about 150 or 180 ohms. The current draw will vary as the alternator brings the battery voltage up, so maybe use the 180 ohm resistors and be a little safer with the current draw through the leds. With the battery voltage up to 14 volts, the 150 ohm resistor will allow 50 ma to pass through each string, which is usually a maximum. Using a 330 ohm resistor will drop the current through each led to about 20 ma, so there's your range of resistor values.
Now a potential problem- if the dimmer is powering only these leds, then they will always be bright, since the dimmer is just a variable series resistor inline with the dash lights. It's not a potentiometer. Because the dash lights have a much higher current draw than the led array will have, the resistance of the dimmer is way too low to give a proper dimming action to the array. But you stated that you would be adding this array to the existing dash lights, so they will dim ok, though they may go right out before the dash lights get very dim.
The led array will look to the dimmer like one, maybe two extra dash lights, that's all.
Most leds, if of same color and type (coming from the same batch, basically) will be pretty closely matched, and it will be unlikely to have a noticable brightness difference between any in a string. If me, and one led appeared to be weak in a string, I would eliminate that one and wire in another.
You didn't say if you were going to mix colors with these 18 leds. If so, you'll have to make equal strings, like each string would have a red, a white, a green- whatever. If you make a string of greens, for instance, and parallel that with a string of reds, you'll have to calculate a different resistor value for each string so each will draw an appropriate current. If you do a mix and make each string the same, the calculated resistor would be the same for each string. Different colored leds have a different voltage drop so you would need to have that data before you could calculate for the resistor value.

[This message has been edited by darryl (edited 03-01-2006).]

J Tiers
03-01-2006, 11:13 PM
Most non-high brightness LEDs are about 1.2 to 1.6V drop.

Some new high brightness types, like the white and blue ones, are up to around 4V, IIRC.

We happily use strings of several old-type higher-brightness LEDs on 15V in our products. No problem.

In fact, since they are current-operated devices, wiring in series is one of the better ways to get them the same brightness. You know the whole string gets the same current.

As for dimming, I wouldn't count on them dimming along with incandescent lights at same voltages. Almost certainly they will not, since their brightness is not related to applied voltage in the same way as an incandescent bulb.

But they may be close enough for you, depending on the chosen series resistor, etc, etc.

[This message has been edited by J Tiers (edited 03-01-2006).]

03-02-2006, 12:37 AM
Excellent dimming response synced to the dash lights could be had by controlling a 555 timer duty cycle with a home made opto isolator consisting of a small dash light in a film can or similar together with a cadmium sulphide cell used as the timing resistor for the 555.

03-02-2006, 03:58 AM
Thanks for all the information.

I measured the supply voltage at my preferred dimness setting and it is 11 volts. 13 volts is the maximum the dimmer appears to provide to this circuit.

I tried wiring up a bank of 5 and a bank of 6 to see how the brightness compared and the bank of 5 looks pretty good. Since this is backlighting, I can add 2 more led’s without too much trouble and keep it simple by going with 4 parallel banks of 5 led’s in series. I am unable to detect any difference in brightness between any of the led’s in the array (all green). The max voltage per led should only be 2.6 volts so it looks like I will be ok – unless I am missing something.

Sorry Darryl, I don’t quite understand; are you recommending the 6 string of three and 6 resistors so I can get the brightness spot on? Would I be better off if I series wired 10 banks of 2 paralleled led’s instead of the 4 banks of 5, or is this just a matter of preference?

As was mentioned, when the dimmer is turned down very low, the led’s go out. Fortunately, this doesn’t happen within my normal range of adjustment, so it indeed appears to be close enough.

Paul Alciatore
03-02-2006, 08:00 AM
As mentioned above, an LED is a current operated device. This differs from a standard light bulb which is voltage operated or rated. The problem is that they will attempt to regulate any voltage placed on them to a relatively constant value (voltage value). So a LED with a 2.3 volt characteristic that is powered directly with a 12 volt source will try to bring that voltage down to it's 2.3 volt level - by drawing a lot of current. Instant fuse or flashbulb effect, small amount of smoke is released and the LED no longer works.

This requires that a current limiting resistor be placed in series with the LED to limit that current to a safe value. That resistor is calculated by the supply voltage minus the LED operating voltage divided by the desired LED operating current. The LED operating current determines the brightness and should be limited to the max that the spec sheet shows for safe operation.

Unfortunately, some "LEDs" today are actually a combination of the LED and the series current limiting resistor. Such combination devices are operated more like a standard light bulb and if they are supplied with the rated voltage, an external resistor is not needed. You MAY have combination devices. You should determine which they are before deciding on a circuit. If at all possible, get the manufacturer's spec sheet.

If that is not possible, you may have to experiment with different circuits to determing the best.

As for putting them in series, results may vary. I had a similar application and I contacted the manufacturer. They said that THEIR devices were matched as close as possible but they could not guarantee anything. I had an indirect lighting situation so I put two in series and it worked out OK. They did seem fairly well matched. But they were all from the same batch and if I have to replace any of them later, results may vary.

Chances are, series operation would be OK. Six of them in series would be about 13.5 volts from your figures. If they are operating at 25 ma then the total current would be 3 X 25 = 75 ma which any automotive dimmer should be able to handle. An additional series resistor should be added to EACH series string. It should be (I am using 15 volts for the supply to be on the safe side) R = (15 - (6 X 2.25)) / 0.025 = 60 Ohms. A 62 ohm resistor is the closest standard value. Wattage is low so a 1/4 or 1/2 Watt resistor would be OK.

Paul A.

03-02-2006, 10:30 PM

03-03-2006, 12:43 AM
I just finished a "ultrabright" white LED design... It is more interesting than you might think. Specs say that your LED is probably 50,000 hours reliability... that's great - but the real world can bite...

When you are a long way out on the end of the wire - voltage may be poor - and spikes may be large. Temperature will affect the Current - Vf numbers a lot. Running pulse style will allow more "apparant" light to come from the LED - if you like.

Your battery wire also hooks to the starter. It hooks to a battery that gets cold in winter. It has nasty motors every where (spikes high and low). Automotive regulators typically should be rated at 60Vin on the line. They are go see 4.5 V at times as well. be safe at those numbers.

Current is affected by temperature and process. You can't count on "consistancy". So allow a good safe operating area. If you can, use a good "low drop regulator" to drop a 1/2 Volt or so to your LED operating voltage. That way, the regulator can be rated at 60V - and protect your LEDs from blowing.

You don't need "high precision" - just to keep the LEDs from Blowing - and you want to look at "safe operating region" numbers.

BTW - I'm building some lights for my boat. LEDs are getting EXTREMELY cool... LOTS of light. Putting them on my car's interor operating lights - as well.

good luck.. much good has already been said...


03-03-2006, 12:57 AM
I just picked up an LED head lamp (for wearing on your head). It has a one watt Luxeon LED. Holy crow! Is it ever bright! You cannot stand to look into it directly. It's as bright or brighter than my three D cell maglight, throws a beam just as far and runs on three AAAs.

It has a strobe mode that looks just like a xenon flash. White LEDs are really improving lately. In a couple of years they will very likely be practical for most lighting situations.

03-03-2006, 01:34 AM
Joel, I suggested the six strings of three leds each for a few different reasons. One is because it gets you the 18 leds in an equalized array. This way you only need to calculate one resistor value, and it's the same for each string. Another reason I suggested only three in each string is because the voltage drop is going to be much less than the supply voltage- this means that the brightness will vary less with changes in the voltage in the vehicle, and the problem of them going out when the dimmer is set low is lessened.
4 leds in each string will work fine, but as you said, you'll have to add 2 leds to get the strings the same. 4 strings of 5 leds each will also work, but the brightness regulation will be very poor. The more leds in each string, the greater the chance of current peaks through each string, due to spikes, etc, as someone else mentioned.
Just my opinion, but I think 3 leds in series, with a series resistor, is about optimum to drive from nominal 12 v, given the normal variations in battery voltage (11 to 14 typical).

J Tiers
03-03-2006, 08:04 AM
<font face="Verdana, Arial" size="2">Originally posted by Evan:
White LEDs are really improving lately. In a couple of years they will very likely be practical for most lighting situations.</font>

Well, they still have a cold light that will suck your eyes right out of their sockets.... Like "cool white" flourescents.

They are cheaper, probably, but not more energy-efficient than other lighting sources. They SEEM efficient because they focus all their light in a very bright narrow cone. But in actual output vs power in, they fall short of many flourescents, and not necessarily just the new type compact flourescent.

03-03-2006, 08:32 AM
"Well, they still have a cold light that will suck your eyes right out of their sockets.... Like "cool white" flourescents."

Not any more.

<font face="Verdana, Arial" size="2">
Luxeon Warm White

Lumileds Luxeon Warm White LEDs are the first commercially available low CCT and high CRI warm white solid-state light source. Luxeon warm white LEDs deliver superior color rendering and color temperature previously associated only with incandescent and halogen lamps. Luxeon Warm White LEDs enable an entirely new world of lighting applications.



J Tiers
03-03-2006, 11:58 AM
The reduced number of LEDs and use of larger dropping resistor is a good solution for matching the dimmer curves vs the other lights.

It does waste power, but that isn't much of an issue in a car, where that waste is a drop in the bucket compared to all the other huge energy wasters.

<font face="Verdana, Arial" size="2">Originally posted by Evan:
Not any more.


Yeah, they came in and demo'd those for us at work a while back. But there was some issue yet, either they are more expensive, not available, or??? Don't recall, might also have been an efficiency difference.

In any case, LEDs are STILL not very efficient. But they do work well for some places. Especially places where you never want to have to change a light bulb.

Brake lights are a perfect application.... since they don't blow out, the restricted dispersion cone is perfectly acceptable, and you can intensity modulate them with PWM to change current draw / relative brightness.

They will probably get fixed to where they make sense. And for focused task lighting, the tiny hot-spot of light isn't as much of an issue.

But a regular compact flourescent still blows them out of the water for general purpose light output, and still saves big on power, lifetime, etc. I have 10 year old CF bulbs that are on a lot of hours per day and still work acceptably.

And you can buy the CF anywhere for a few bux. The LED is lots more expensive.

03-03-2006, 12:35 PM
"They will probably get fixed to where they make sense."

That's why I said about two years from now.

<font face="Verdana, Arial" size="2">
XLamp 7090 LEDs in development have demonstrated maximum luminous flux of 86 lumens and 70 lumens per watt at 350 mA. This represents a 43 percent increase in brightness compared with the maximum luminous flux of white XLamp 7090 power LEDs currently in production.

“These performance results indicate that the LED industry is at least two years closer to achieving the holy grail of 150 lumens per watt,” notes Steve Johnson, head of the Lighting Research Group at Lawrence Berkeley National Laboratory.

70 lumens per watt is right up there with good fluorescents and 150 surpasses them. With lifetimes up to 100,000 hours and much smaller size, which means less garbage, they have a bright future.

03-03-2006, 01:08 PM
You mentioned that you are planning LED lights for your boat. Is this for running lights? I've been looking at this and concluded it will require an array of lights to get the proper field, 112 degrees for port and starboard and 136 for the stern.

I'd be interested what you are doing re this just to prevent inventing the same thing twice.


03-03-2006, 04:36 PM
Running lights...

Rough conditions on make LEDs good choices for trailer lights ( I see them FREQUENTLY on trucks now) and for any external boat light.. mast / spreader lights..

I think that there are reasonably cost effective lights out there now.... but...
(a) I find the illumination numbers published "lacking" and (b) I am tinkering with controls to make them addressable/verifiable...


03-03-2006, 05:33 PM
One place where LEDs rule is in Fault Indicators used on underground primary cables. Their small current draw allows a very small long life battery to be used, plus they don`t burn out. I don`t believe any other light would be practical for that particular purpose. Traffic lights seem to be as bright if not brighter with the LEDs also, which isn`t a bad thing for us "Mr. Magoos".

03-15-2006, 01:04 PM
Mad Scientist
A voltage regulator is probably all for the good but don't forget that now you should protect the voltage regulator from spikes.

J tiers
The assumption so far seems to be that LED's are relatively safe fron voltage spikes in the forward direction because they just pass the spike through and the resistor limits the current. You still have the problem of a spike in the reverse direction. LED's have a relatively low peak reverse voltage rating and the resistor will not limit current until after the LED has broken down on reverse voltage then it's too late. So it would be wise to provide some sort of reverse spike protection


03-15-2006, 01:18 PM
A 1n4007 should do.

Mad Scientist
03-15-2006, 03:01 PM
Yep. Also a small capacitor (.01mf) across the power leads at the LED"s.

J Tiers
03-15-2006, 04:39 PM
And that is only if you have a credible reason for expecting such a reverse polarity spike......

MOST circuits will not have any reason to expect one.

And, if any unidirectional part (diode, SCR, etc) is in series, IT will take the reverse voltages, and there is no need for added parts.

[This message has been edited by J Tiers (edited 03-16-2006).]

03-15-2006, 11:59 PM
Things with battery/starters, motors, "generators" - (cars or boats, etc.) are subject to spikes... wall power or plain old batteries are pretty benign... Some bright white LEDs are 0.55 or so... some blue LEDs are 0.15 or so...
Harder to get, but real numbers anyway...

-- jerry

03-16-2006, 03:09 AM
The 317 is a 3 terminal device. The lowest parts count regulator circuit using it is a current regulator. If you connect a single 47 ohm resistor between the adj terminal and the output terminal, that's your current regulator. No other parts or connections needed, just the input voltage to input pin, that resistor between the adj terminal and the output terminal, and output current is taken from the adj terminal, not the output pin. The output terminal goes only to one end of the resistor and nowhere else.
With the 47 ohm resistor, the output current is about 30 ma. That's about right for an led, or a string of leds. If you directly parallel strings of leds, you can change the resistor value to suit. For example, if six strings of leds were paralleled and driven directly from the 317 regulator ic, then the resistor value would become 6.8 ohms (giving a 180 ma output). Brightness control would be had by varying the value of the resistor.
Note that the regulation is done by the 317, and merely adjusted by the resistor. This is not the same as a single resistor in series with an led, or a string of leds, where brightness will still vary with changes in battery voltage. Using the 317 in this simplest of circuits, the brightness won't change with changes in battery voltage unless the battery voltage drops below the level required to operate the regulator ic. In the case of the 317, this 'dropout voltage' is about 2 volts, so the led string can't require more than 10 volts if operated from a 12 v source.
If using white leds, you can't make each string longer than 3 leds, otherwise none will light unless your battery voltage rose to 15 volts or more. For colored leds, you could have 5 greens in series per string, or 6 reds per string. Blue limits you to 3 per string, similar to whites.
This elegantly simple circuit will protect against voltage spikes up to about 50 volts, and adding a single diode, the 1n4007 as Evan suggested, will protect against reverse voltage spikes.