The LED lighting project

[Sean]

Phew!  That was a close one 

Seriously, Tim, thanks for the correction.  It's good to know there are at least a couple of sources for low-volume buys of color temperature-binned parts.  If we can't get the temperature we want for our next batch of fixtures (I'd much rather buy off-the-shelf than have to build from scratch), I might just have to fire up the Weller.

-Sean
http://ourodyssey.blogspot.com

[boogiethecat]

Ulp me again...

Something no-one's mentioned...yet...

The "white" LED's aren't really white LEDs.  There really is no such thing!  But wait... there is...no there ain't... huh?

Well, a while back some smart guy (I think at Nichia) figured out that you could plop some phosphor on top of a blue LED and make it glow some other color than blue (the blue excites it) and brightly at that... from there, they found that a mixture of various phosphors of different colors created a goo that you could deposit on top of a blue LED to make it look white...
From there, the mixture of goo allows "white" led's to be "warm" "cold" "yellowish" etc.
If you look at any LED other than a white one with a prism, all you'll see is the same color as that LED is... ie red, blue, etc.
but if you look at a white LED with a prism you'll see a continuous band of colors just like you'd see if you look at the sun with the prism!

Just though some of you'd like to know although I'm sure it's old news to Sean and Tim (hi guys) but I can certainly see binning them (white ones) to color, as phosphor runs are likely to be slightly different from batch to batch...

Cheers
Gary

[Sean]

I am also tickled that after everyone said BobOfTheNorths lights won't work long...they still are!

Jack,

I know you don't need an answer to this, but I want to write one anyway, because I think some folks may not really understand what some of us meant, way back at the beginning of the thread.

What I said back then was not that they would not work "long," but rather that "I believe[d] ... that [he would] see LED failures in relatively short order." (Emphasis added for this post.)

The key here is the word "relatively," and it is this word that may be the source of confusion.

LEDs have an MTBF (roughly translated as "life expectancy" of between 100,000 and 1,000,000 hours, depending on type, environment, and application.  (And, yes, I know that is an entire order of magnitude spread.)  To put that in perspective, that means that a typical LED should last, illuminated continuously and operated under the conditions of voltage, current, vibration (or lack thereof), temperature and other parameters as specified by the manufacturer, for a minimum of 11 years and up to 110 years.

So if you shortened that life by, say, 99% due to exceeding specs, you could still expect the LED to last 1,000 hours, or nearly 42 days of continuous operation.  I strongly suspect that Bob has not had his LEDs running for a full thousand hours, so the jury is out on whether or not the lack of current-limiting resistors will effectively shorten the lifetime.  Bottom line is that if he ever has to replace them in his lifetime, then, yes, it probably did.

For comparison, BTW, the lifespan of a halogen bulb is from 2,000 to 5,000 hours.  I've had halogens in my bus for five years and I've only replaced two bulbs out of dozens, which gives you a sense of how many hours even a full-timer puts on lamps.  Most of us could easily compromise the design life of an LED by 95% and never know it because we just don't ever rack up that many hours.  OTOH, if you play too fast and loose with the current specs, it will only take one momentary voltage spike, say from a twitchy alternator regulator, to take out every LED that happens to be turned on.

Hope that clarifies things a bit for all concerned.

-Sean
http://ourodyssey.blogspot.com

[JohnEd]

I will go back and revisit Boogie's post on computing the resistor size.  Is it the resistor that is the current limiting device? 

Were I building a fixture that put out the equal lumen's of a 200 watt bulb, I imagine I would have some money invested.  That 'current limiting IC" would seem to make sense even if it cost $6 if it would protect "every LED in the bus" from a voltage spike.

I understand the concept of stress caused by "rapid start up".  How do you soften that shock?  A resistor isn't a coil, if you follow my meaning, so that resistive component wouldn't address that surge problem.  Right?

Thanks,

John

[pvcces]

JohnEd, what do you think would happen if you used a capacitor across the LED with a current limiting resistor?

What I don't know is how slowly the power should be applied to the LED.

Tom Caffrey

[Singing Land Cruiser]

Hey Bob and others, Christi & I are thinking about LEDs. Do you have the Pros and Cons on them. Do they put out enough lite for reading, etc...?
Cost, installation? The bset brands? Thanks, M&C

[poppi]

Well I've enjoy all this theory stuff.

One thing I did learn this month is that those little fluorescent light bulbs do not like being on a dimmer
light switch. I tried it none lasted over a month.

LEDs (a simpletons view of what has been discussed)

"it will only take one momentary voltage spike, say from a twitchy alternator regulator, to take out every LED that happens to be turned on."

 Not really too worried about it. If you think about it (on my bus) the alternator goes to the coach batteries
then to a master switch to a solenoid switch then to the house batteries. between the batteries as the
"shock absorber" and over 60 ft of wire for the resistance and capacitance then down to 12 volts the voltage
is pretty stable.

 Initial turn on spike of the unit...........tore apart my cheap led flash lights and found no fancy circuitry.
  Been using them for 2 years now No big deal in my simple mind.

Just a FWIW

  Skip

[bobofthenorth]

Michael:
The reason we did this was to reduce our power consumption prior to putting in solar panels.  We had roughly 40 of those little teardrop Halogens in the bus.  They did a better job of heating than they did of lighting and I say that only slightly tongue in cheek.  Sean claims good life out of what I assume to be the same bulbs but that most definitely was not our experience.  We've had the coach 5 years and I have bought over 40 of those bulbs in that time.  I probably have 10 left in inventory for the few of them we have left in service but still I think we have likely replaced 40 bulbs over the 5 years.  So they most definitely were not a reliable lighting source in our application.

Skip:
Thank you for the voice of common sense, a commodity that has occasionally been sorely lacking in this thread.  The only time that I have any real concern about voltage with regard to my LED modules is when we equalize.  Since we do that manually I can control what lighting is on while we do that.  I'm not convinced there is a problem even during an equalization cycle but, if there is, then it is a controllable problem.

I'll repeat my caveat: I don't know how long these creations may last but so far they have dramatically outperformed the commercial units that they are running with.  I haven't detailed all the commercial LED solutions that we have used but I can say that every one of them has failed to perform in some respect.  We also tried a 12 volt fluorescent fixture.  When it failed (fairly promptly) I replaced the tubes with some of my LED creations and just used the fixture as a housing for the LED lights.  For what I have spent on the homemade LEDs c/w what I have spent on the commercial solutions I can afford to replace my LEDs in less than 10,000 hours, a lot less in fact.

I will however be trying some of the LEDs that Tim referred to.  I have difficulty believing that they are really 80x as bright as the ones I am using but even if they are 10x then they would be worth the extra bux.

[Sean]

LEDs (a simpletons view of what has been discussed)

"it will only take one momentary voltage spike, say from a twitchy alternator regulator, to take out every LED that happens to be turned on."

Please let me clarify my own comment:  I said that IF you failed to properly limit the current, which merely requires resistance of the appropriate value, THEN a spike could take out all your LEDs.  LED's have a non-linear voltage/current relationship -- once voltage rises even a little bit above the maximum rating for the part, current draw skyrockets, and it is the current which kills the part.  Having some resistance in the circuit forces the voltage/current relationship closer to linearity.

....tore apart my cheap led flash lights and found no fancy circuitry.
  Been using them for 2 years now No big deal in my simple mind.

You can't draw parallels between cheap flashlights and an alternator-, converter-, or charger-driven system.

If you take apart one of those LED keychain flashlights, you will find no parts besides the battery, LED, and switch.  So where, you might wonder, is the resistor?  The answer is:  in the battery.  Every battery has "internal resistance."  You can easily calculate what this resistance is by short-circuiting the battery with an ammeter.  I do not recommend you do this with your house batteries!

The tiny little button cells used in keychain lights have very high internal resistance compared to their output.  In other words, even if you shorted the battery completely, the total amount of current it generates is very limited.  So limited, in fact, that it is safe to drive an LED directly from such a battery.

No such thing is true about the electrical system on your coach.  Even when running only on batteries, your battery bank is capably of generating huge amounts of current if shorted -- you probably could even weld with it.  I always remove my watch and my ring when working on batteries for just this reason.

BTW, modern LED flashlights often use driver chips.  I have a light I always carry with me that uses a single AA cell, with one of Tim's high-output Luxeon Rebel's in it.  It is a very bright light.  The only way to drive such an LED with a single 1.2-1.5v cell is to use electronics.  But to put a price tag on that, the light cost me $20.  Inexpensive ($2-$4) LED flashlights, which have to use two or three batteries to get the proper operating voltage, usually have a simple resistor in series with the LED.

-Sean
http://ourodyssey.blogspot.com

[poppi]

My apologies for not including the If portion on the quote.

  Though that brings back a quote my dad used to say

   If the dog hadn't stopped to wet the tire he would of caught the rabbit.




Take care
Skip

[Tim Strommen]

"it will only take one momentary voltage spike, say from a twitchy alternator regulator, to take out every LED that happens to be turned on."

 Not really too worried about it. If you think about it (on my bus) the alternator goes to the coach batteries
then to a master switch to a solenoid switch then to the house batteries. between the batteries as the
"shock absorber" and over 60 ft of wire for the resistance and capacitance then down to 12 volts the voltage is pretty stable.

The 60Ft of wire thing though is what can get you into trouble...  if you switch the LED locally (at the near side of the 60ft wire), you'll get an inductive "kick" (spike) as the current starts running through the wire.  The best way to counter this is to put a bulk capacitance at the switch to absorb the kick at the switch (both on/off).  We use this same concept when dealing with hot-swap backplanes in the electronics industry.

...I will go back and revisit Boogie's post on computing the resistor size.  Is it the resistor that is the current limiting device?...

Hello John,

Yes, you are correct, the resistor is a current limiting device in this type of circuit.  The is because the LEDs have a specified forward voltage (a voltage which must be overcome before it begins to conduct in volume), and after you add up all of the V_fwd of all the series LEDs there is only so much voltage left.  This is where the resistor comes into play - it has a measured resistance against the passage of electrons for a given voltage, and it's effect on the flow of electrons is calculated by using the formula in Ohm's Law:

E
---
IxR

Simple math is all that is needed to pick a resistor with the correct "Ohm" value to allow a measured passage of current at a given voltage.  However its reliance on the voltage across it is the weak point - if the voltage is not regulated before the circuit - the circuit is then subject to the fluctuations of the power supply (any line noise, surges, sags, etc.).

...I understand the concept of stress caused by "rapid start up".  How do you soften that shock?  A resistor isn't a coil, if you follow my meaning, so that resistive component wouldn't address that surge problem. Right?...

You can do this by putting an inductor in series with the current limiting resistor on the power supply side, and a large value capacitor in parallel to the LEDs - the inductor will resist the high speed change of the input voltage, and the capacitor will rob some of the current that is allowed to pass through the current limiting resistor, until the voltage matches the V_fwd of the LEDs (roughly).  The capacitor in parallel with the LEDs should add a bit of safety to the power supply - but the math for doing this is a bit more involved than just picking out a current-limiting resistor.

The easiest way to take care of start surge is to get an LED driver IC that has Soft-Start built in (and a vast many of them do, LED driver manufacturers are staring to be told that if they don't have SS customers won't buy their part...).  The rate at which the chip is ramped up to the desired current is typically controlled by picking the size capacitor you want (and there is easy math in the driver's datasheet to figure that out).


-Tim

P.S. Sean and Boogie are great guys, I've spoken with them off-line several times - I think we understand each other enough to know when we're joking around . -T

[boogiethecat]

 Hey there,
I thought maybe making a mechanical analogy of LED's, voltage, and resistors might help you all figure out what this is all about, and why current limiting is needed.

So I made this little drawing of a piston in a cylinder...





Imagine that the LED is the red part in the drawing, and it won't glow until it gets some pressure applied to it by squeezing it.  (pressure equals voltage in this adventure)
For this example let's say that the voltage rating of the LED is 2 volts.  What this means is that anything less than 2 volts across it will not make it glow.  That's all you need to know about the voltage.

For current, in this example LED current is equivalent to the squeezing force that the piston applies to the LED as it's being raised.  The LED isn't solid, but it also isn't a marshmallow. It does have a little bit of give but if you squeeze it too much (ie put too much current thru it) it will squish and be destroyed. force = current...

For the resistor in this example, consider it a spring. The more it stretches, the more current is flowing...
..and a whimpier spring would be equivalent to a higher ohm value resistor...


Ok.Now consider drawing A and suppose that you start raising the piston up- for the sake of arguement let's say that instead of raising it with something weak like your fingers (equivalent to a little lithium coin cell in this context) it's being raised by a very strong thing, like a hydraulic system that would rather break things than be stopped. (equivalent to your bus's battery bank that would rather melt things when given the opportunity ..a short for instance)

So when the piston is at the bottom (0 volts) nothing happens. 
As you raise it (ie raise the voltage across the LED) still nothing happens until the piston finally reaches the LED and touches it (the voltage finally becomes equal to the LED's rated voltage)
  Now the LED starts to glow. 
Looking at drawing B you can see that the LED is glowing (some current is flowing) but it is getting a little squished because the voltage is slightly more than 2 volts, although not enough to break it.  As you further raise the voltage as in drawing C, you find that without too much more increase in voltage, the poor LED has gotten squished (equivalent to way too much current flowing thru it) and it's now a DED instead of a LED!! (ie it doesn't work anymore)
It didn't take much more than 2 volts to kill it because the hydraulic system squeezed it between the piston and the stop with mucho force.  Equivalent to even a small, momentary  voltage spike in your bus...

So as you can see, the difference between a happy LED and a dead one is a very small amount of voltage. This is exactly what happens when you simply add LED voltages up and hook em to a battery. They may work but it doesn't take much to kill them and there is little control of the current through them.

RESISTOR TO THE RESCUE

Now consider adding a "stretch me" kind of spring in the middle of the piston shaft, which is equivalent to adding a resistor in series with an LED.  When pressure  starts getting applied to the LED , the spring starts to stretch and keeps the hydraulic system from squishing the LED. Likewise a resistor adds some "springiness" to a circuit and limits how much current can flow.  Like the piston system with a given sized spring, it will eventually stretch far enough that it can't limit the pressure anymore and the LED will get squished.  Same with a fixed resistor, it will limit the current effectively but if you raise the voltage high enough, it will let enough current through to kill the LED. 
In drawing D, again the voltage isn't enough to make the LED glow, and the spring isn't stretched at all (no current flowing).
In drawing E, the voltage is now up enough to make the LED glow, and the spring is a little stretched because the voltage is a bit higher than the LED's published voltage of 2V.  In fact it's high enough to kill the LED if the spring wasn't there, but the spring stretches, gives the LED some slack and keeps the hydraulic monster from squishing the LED. (ie the resistor keeps the extra voltage from causing too much current to flow, and keeps the LED within its safe range of current)

Finally in drawing F, we've raised the voltage to 4 volts.  Well enough to destroy the LED but since the spring is in there and it's stretched, (more current is flowing but not enough to kill the LED) everything is happy even at 4 volts.
  So now as you can see, the difference between a happy LED and a dead one -when a spring is included in the system- is a fairly wide range of voltage instead of a very narrow range of voltage.
That's the magic of a simple resistor!

Last, in this analogy if you consider more resistance to equal a whimpier spring, you can start to grokk how it all works.  More ohms/resistance = dimmer (but safer) led's....not as bright as with fewer ohms but a wider safety range voltage-wise.  That's where the compromise comes in... with a resistor you can safely work over a 2:1 range and have reasonable brightness at either end of the range.  Beyond that range, it's probably time for electronic limiters which can give you hundreds-to-1 voltage range.  But on a bus, 12 volts plus or minus 6 is just fine with a simple resistor, and you probably won't see nearly that much swing anyway...

Oh heavens I hope that helps...
Cheers
Gary

[poppi]

Gary, (boogiethecat)

   Thanks for the time and effort you put into your explanation.

I got to tell though I would never put individual springs on a piston.
I'd go with an acumilator and regulator inline so I wouldn't have to worry about uneven pressure
  and keep it constant

Skip

[boogiethecat]

Heh!  I guess we need accumulistors for this job!!! 

[Tim Strommen]

Don't you mean capacitors? 

-Tim