The LED lighting project

[bobofthenorth]

Update time:

After considerable reading and consulting off-line with some people who are capable of explaining the subject matter it appears that my "design" will fail.  Time will tell when it will fail, maybe tonight for all I know.  I opened this discussion by admitting that I didn't know what I was doing and I still don't although the list of what I don't know has got a little bit shorter.  Since there never was any design talent applied to this project the fact that I came up with a bad design should surprise nobody, least of all me.  I'm prepared to run the risk of imminent lighting failure in exchange for economical short term lighting but I am also working to actually design a circuit board which will incorporate some resistors.

The great thing about internet forums in general and this one in particular is the free exchange of knowledge.  Sometimes you have to wade through the chest thumping and loud proclamations of "you can't do it that way" to get to the core material.  The mark of intelligence is the ability to speak to your audience and that is particularly obvious on forums.  Nevertheless there is a tremendous amount of intellect (and sometimes imagination) brought to bear on a particular problem if people are willing to share - witness Ace's electric thread and Doug's or my engine threads. 

The frustrating part of forums is the chest beating and scare mongering that sometimes occurs.  In my case I'm replacing halogen bulbs that run hot enough to literally kindle a fire.  Suggesting that my LEDs somehow pose an incremental safety risk is just silly.  That kind of foolishness always causes me to discount everything else a particular poster may have to offer right out of the gate, sometimes at my own peril.  Stay tuned for an update on the imminent failure of my current modules, or just watch out your window for the fireball depending on your inclination.

[boogiethecat]

My offer still stands to send you some 10 ohm, 1/2 watt resistors. If you include just one of them in series with each of your boards, I think that your failure rate will probably go close to zero.

Kind regards
Gary

[Tim Strommen]

...I opened this discussion by admitting that I didn't know what I was doing and I still don't although the list of what I don't know has got a little bit shorter...

Hey Bob,

    You might see my name coming up a few times on this board whenever LEDs are discussed.  I'm a geek, and work in the electronics industry (currently working as a contractor at Nv*dia, job is ASIC Qualification Engineer).  I'm interested in your project mostly because you went through the trouble of building a PCB for the LEDs.  While I do agree with Boogie about the minimum safety added by a current limiting resistor in series, I don't like them.

This isn't to say that I'm disagreeing with Boogie, I prefer an alternative in LED driver ICs.  One of the problems I have with current limiting resistors is that they aren't very smart - if the voltage changes, it doesn't keep the current constant to the LEDs which can cause visible dimming, color shift, or shutoff of your LED string.  Driver ICs watch the current drawn by the string and adjust the voltage it is supplying to the LED string to keep the current constant.  Even among regulators, you can have differences - switching regulators tend to be more complicated than linear regulators, but they also can have much higher efficiency and better stability over a wider range of system voltages.  A way to describe a linear regulator is like a variable resistor, while a switching regulator is like a light switch that gets the supply around the area you want it to be, then relies on the attached circuit to even out the supply.

Take for example this part from Linear Technologies.  It claims a 94% efficiency around the 24-30VDC range.

The complexity with LEDs is that yes, they need a certain forward voltage to start conducting - but once they do start there is only the the size of the PCB traces, the die-connection-wires, and the silicon's thermal runaway characteristics that will limit the current (with an abrupt "pop" I might add, or at least the escape of some smoke).  Since you don't want to be an Engineer, I can bring up the water analogy for this example - but for LEDs it gets more fun...

Imagine that the doped electron gap in the LED die (the part of the silicon which determines which side is +/- or Anode/Cathode respectively) is actually a small ditch.  On one side of the ditch is a kid with a hose, on the other is a bucket.  Now, depending on how much that kid opens the valve on that hose only so much water will come out.  If he doesn't open it very much, most of the water drops right down at his feet and barely even gets into the ditch (nowhere near the bucket...).  If he opens it a bit more, the water might spray halfway across the ditch, but still not make it into the bucket.  If he opens it just the right amount, the water goes all the way across the ditch, and makes it into the bucket.

By crossing the "ditch", electrons in the LED jump the gap in the die and this act will emit light at a specific wavelength due to the chemical "doping" the LED manufacturer used for the process - for white this is usually a form of blue - with a white phosphor coating (like in a florescent tube) that reacts to that color and emits white (it can be a mix of RGB, or blue and yellow, but this is becoming less common).

Now here is the fun part...  Yeah okay, now you got the water across the ditch, but the valve can open more - but if you overshoot the bucket, you're wasting water.  For LED's this waste is heat, and as every person who's been around electronics can tell you - heat is bad m'kay?  That heat in the die has to go somewhere, and the LED manufacturer has only designed the LED to be able to get rid of so much at a time (this gets worse if the temperature around the LED is high, as there isn't as much difference in temperature to "lose" heat to...).


Now I understand where Boogie is coming from...  Do you have kids?  What profanities would you be yelling out your back door if you told your kid to fill the bucket and he was spraying it all over the place - especially in California where they are talking about water rationing or jacking up our rates?  All of a sudden it becomes important to hit that bucket huh...  Just the same, seeing your kind of circuit drives us engineers nuts and we don't know what to do with ourselves .


Well, an LED driver IC is like a computer controlling the valve on that hose, with a sensor to see if the water is landing short or long, and can correct the opening of the valve base on what it measures.  A current limiting resistor is like a valve with a fixed opening, so if the city cranks up the pressure or it has a sudden drop, your water may miss the bucket (and may not even cross the ditch causing the light to go out, or will completely overshoot the bucket causing the LED to burn up).  But either of those devices is really better than just taking what the city provides and blasting it across the ditch blind.


Back to the Linear part I mentioned above - it would make your circuit a bit more complex than the board you designed, but the fact that you showed that you are willing to play around with this and are bold enough to build boards, shows me that you could learn a little more and successfully put some form of driver IC on the board to get a nice reliable piece for yourself.

-Tim

[bobofthenorth]

Now we're getting somewhere and you have rifled in on my concern about the fact that a resistor in the circuit may regulate the current but it still doesn't limit it with fluctuating voltage.  IC sounds expensive - maybe that's just more evidence of my lack of knowledge but couldn't I accomplish more or less the same effect more economically with a zener diode?  The whole point of this exercise is to come up with a light module at a reasonable cost.

Thank you Tim for a clear explanation and for taking the time to understand what my concern was.

[Tim Strommen]

Hey Bob,

   No problem.  The Linear part and all of the required auxiliary parts can be had for less than $8 per fixture.  The Zener solution is one way, it's like a linear regulator of sorts but is more for regulating the voltage of a load, not the current.  There is an additional complication that I spared you from, in my previous description - LED's have a "dynamic resistance".  Therefore, the voltage required to "jump the gap" can climb slightly as the die gets warmer.  This is why current regulation is the way to go for LEDs in illumination applications (really in any application, but more so in illumination).  By giving a regulator free-reign to control the voltage, so long as the current stays correct, this takes care of that problem.

I did a quick search, the Linear part can be had in single quantities for $6.38 at Digikey.com.  The other parts are a few cents each.  After that it's just a matter of building a board around it - the circuit for driving your LED strings should be about the same physical size as two of your LEDs (and remember, since it doesn't put out light - it can be on the back of the board to save space).

-Tim

[Sean]

Now we're getting somewhere and you have rifled in on my concern about the fact that a resistor in the circuit may regulate the current but it still doesn't limit it with fluctuating voltage.  IC sounds expensive - maybe that's just more evidence of my lack of knowledge but couldn't I accomplish more or less the same effect more economically with a zener diode?  The whole point of this exercise is to come up with a light module at a reasonable cost.

Thank you Tim for a clear explanation and for taking the time to understand what my concern was.

Everything Tim's written is, of course, correct.  But, FWIW, virtually every commercially available LED fixture for this type of use utilizes a simple resistor for current limiting.

Resistors are a few pennies; driver chips are several dollars.  Where the latter really shine is in applications where the source voltage varies within a wide range.  For example, to deliver a relatively constant brightness flashlight given battery voltage that cascades downwards from, say, 3v all the way to 1.5v -- a 50% drop.  Or to build multi-volt fixtures, such as marker lights than can be used on both 12v and 24v nominal systems.  Or, as Gary wrote, to extract the most output with the least amount of energy, such as cell phones, music players, and the like, where miniature battery longevity is a paramount concern and the extra price of the part is relatively minor in comparison to the total package.

When voltage varies within a narrow range, such as in a nominal 12v system (where voltage typically ranges from about 12.2 to 14.0, or only about 15%), a resistor is a perfectly acceptable solution, and it's definitely more cost effective.

If you choose your LEDs and resistors properly, the modules will work across this entire range, and there will be almost no noticeable difference in brightness.  In addition to saving a few bucks on every module, you'll have less complexity in the solution (and less board real estate as well).

Again, if you take apart almost any off-the-shelf 12v LED fixture (or bulb-replacement module), you'll find resistors, not driver chips.

FWIW.

-Sean
http://ourodyssey.blogspot.com

[bobofthenorth]

Again, if you take apart almost any off-the-shelf 12v LED fixture (or bulb-replacement module), you'll find resistors, not driver chips.

Sean, shame on you.

Allow me to paraphrase:
"Again, if you take apart almost any off-the-shelf RV you'll find truck chassis not Prevosts or Neoplans."

[boogiethecat]

Sean, I think we both need to walk away from this one & move on... you can take a horse to water....
Gary

[Sean]

...
"Again, if you take apart almost any off-the-shelf RV you'll find truck chassis not Prevosts or Neoplans."
...

Bob, I'm a practical guy.  I had gotten the impression somewhere along the line that you were, in part, making your own modules to save some money, in addition to the other benefits.  I was simply pointing out that there is a place where driver chips are cost effective solutions, but this application is not it.

If what you want is the whizziest solution, irrespective of cost, then by all means go ahead and get the driver chips.  They are certainly effective.  But many other folks are following this thread, and, with all due deference to Tim, some folks are looking for simple and inexpensive, and I felt it needed to be pointed out that, in this application, resistors present a cost effective solution that is very nearly as good.

The comment regarding construction of off-the-shelf fixtures was for the benefit of the many who may not realize how they are made.

-Sean
http://ourodyssey.blogspot.com

[Van]

Bob,these are the folks that helped me out with my project

http://www.hitechledproducts.com/product_info.php?cPath=26_38&products_id=196&osCsid=2b4b9094e985a6682a644841b7e6f2eb

[bobofthenorth]

And you are right Sean but I couldn't resist the analogy.  And I still have to ask in the context of this community - is it too radical a concept to look for an elegant solution that is also cost effective?

[Tim Strommen]

Hey Bob,

     If you want "elegant", one of the key criteria is "simple".  Thus if you are seeking elegant, go for the resistor.  If you want efficient/stable go for the switcher or another regulator.

I believe that Sean and I both took different approaches with your question (one of the great things about this board, is you get so many different right answers).  I was running under the assumption that the reason you were going with LED fixtures was the desire for low-power-draw/high-overall-efficiency in place of a hot/high-power-draw halogen - I had made a little leap based on the fact that you had gone into building circuit boards for your project.  Sean has stated that he was under the assumption that:

...I had gotten the impression somewhere along the line that you were, in part, making your own modules to save some money, in addition to the other benefits...

He also undoubtedley read the part of your post where you stated that you were not an Engineer, nor had any intention of trying to become one.


Now I have nothing but greatest respect for Sean and Boogie - and the resistor is one of many right answers.  For each assumption, our respective solutions are correct.  They do compete, but this is actually one of the parts of Engieering that most people miss - the balancing act.  There is always a cost/performance(reliability) trade-off, and a design-timeline/complexity trade-off, and you have to know your limits.

Something you should know about me is that I spend a lot of money on my projects because I want them to be "just-so".  It's a bit of pleasure for me to voluntarily remove some of those trade-off constraints that are typically found in commercial engineering (this is after all MY damn bus!!!).  The benefit to me is that I know a lot about how things work and can make them do exactly what I want them to, with any parts combination I want, upfront-cost be damned.

Commercial products are constrained by these limitations however - and yes like Sean said, you will find resistors in the better-selling fixtures, because they are cheaper to build and can be sold for a better margin.

Take a look at your application - are you running these lights from solar on a deep-cycle until they are dead in order to squeeze out as much power as possible?  If yes, then a switcher or some other form of driver IC is for you,. If however you are plugged into shorepower and running a genny, and never run your batts below 95%, don't waste the time with having to learn good circuit design, just use a resistor (the math is way simpler, which would satisfy the "not make you an engineer" criteria).


The whole "shame on you" thing?  Even if it was in good fun, after I read it (mind you with exactly zero knowledge of your personality - which is a hazard with a BBS post) I was a bit put-off.

With respect to the "chest thumping" and "you cant' do it that way" comment and the:

...Suggesting that my LEDs somehow pose an incremental safety risk is just silly.  That kind of foolishness always causes me to discount everything else a particular poster may have to offer right out of the gate, sometimes at my own peril...

Look when someone tells you not to stick your hand in a blender and you yell at them because you know it's not plugged in - it doesn't mean that they are spitefull and just trying to posture, they may be thinking about training you not to put your hand in the blender (next time you may forget and it might be plugged in).  I find it's best not to make a habbit out of dodging bullets - I'd rather be off the range.

-Tim

[bobofthenorth]

.... and when someone tells you not to put your hand in a mixing bowl they're just being silly.  When somebody says something stupid they can expect me to call them on it.

You guys need to lighten up considerably.  Anyone here who couldn't read the humour in the "shame on you" comment needs to step back and assess why they couldn't see it.

Tim you have opened my eyes as to some of the broader possibilities. My goal is 2-fold:  One to keep the cost under what it would cost me to buy LED lighting, if what I wanted was even available.  Two to let us live off the grid for extended periods of time.  Since your first post I've done a lot of reading about ICs and current limiting diodes and all it has accomplished so far is to open my eyes to how little I know.  What I have gleaned out of my reading so far is exactly what you just said - that the IC will be the most efficient way to drive my LEDs - I'm not yet sure whether the difference in theoretical efficiency is material in our situation.  And I'm also not sure what kind of an IC solution is possible without a bunch of support circuitry (read $$$).  It also appears that there are other simple current limiting circuits that I might be able to implement for relatively little cost.  In the meantime I've got 47 ohm resistors in front of most of my famous modules with a couple left as test cases.   

[Sean]

And you are right Sean but I couldn't resist the analogy.  And I still have to ask in the context of this community - is it too radical a concept to look for an elegant solution that is also cost effective?

No, it's not radical.  But, hey, didn't you write this:

...  But sometimes things can get over-engineered and it is better to go the simple route. ...

? 

... to let us live off the grid for extended periods of time.  ... the IC will be the most efficient way to drive my LEDs - I'm not yet sure whether the difference in theoretical efficiency is material in our situation.

Bob, it's actually pretty easy to calculate exactly how much energy will be "wasted" by using resistors instead of more advanced methods.

Since we care (presumably) only about energy consumption when on batteries, and not when on shore or generator power, we start with an average system voltage of around 12.4 volts.  In practice, your battery voltage will run from about 13.2 down to about 11.8 (loaded).  With the three-led strings we discussed, with a 200-ohm resistor in series (which is worst case -- the four-LED strings with smaller resistor would be more efficient, if they worked at the lower voltage), you will be dropping 2.5 volts (12.4 - 3x3.3) in the resistor, for a power consumption of .03125 watts.

Multiply .03125 times the number of modules you will have illuminated at any given point, then multiply by the number of hours used, to get watt-hours.

For example, if you have 20 of these modules on for five hours each day, that would be 20x5x.03125 or 3.125 watt-hours.  On a 12-v battery bank, that works out to just about one quarter of an amp-hour; half that on a 24-volt system (less whatever inefficiency is involved in the 12-to-24 conversion).  Remember, too, that this is an upper limit on the "waste" -- driver ICs also impart operating losses to the system, just smaller ones.

In two full weeks of use at that rate, that's only about three and a half amp-hours.  Of course, plug in whatever numbers make more sense to you -- I'm just guessing at your usage rate.

HTH,

-Sean
http://ourodyssey.blogspot.com

[bobofthenorth]

UPDATE

I've now had three out of eight of the Lights of America (professionally engineered) 110 volt LED bulbs fail.  Meanwhile my little modules continue to light the bus every evening and (so far) haven't let out even one little puff of smoke, let alone lit us on fire.  And that includes the ones with no resistors as well as the ones with the 47 ohm resistors and the ones with the 10 ohm resistors.  In addition one of the bulbs I bought from Sailor Sam's (at marine pricing) is intermittently going dim.  Usually if I slap it lightly or turn it off and back on it comes back.  So the store bought bulbs are currently failing at a rate of roughly 1 bulb per month.  I'm not smart enough to calculate MTBF but that seems high to me.