Since light is a product of Lumens, brightness, and temperature, there is no exact formula that is really usefull. The best I have seen is to use a phototransistor and measure resistivity. Empirical observation says that using 18 7200 mcd blue-white led's is about 40watts in comparison. It is closer to 7000k in color, than an incadescant bulb of 3200k. My example was bunched into a 3/4 copper fitting. I don't know how it would look strung out along an edge. I had a spacing of .10 inchor touching bulb to bulb. I would inmagine the spacing could be as much as 2 inches. Flourscent lamps vary from 2800 to 7200k. The cheapest are the higher k temperatures. Yellow, green, blue and IR can produce different effects that you may want to consider. One thing interesting about led lighting, by varying the voltage or the repeat rate on a strobed array, will change the brightness. I did not experience any appreciable heat from the array.

.

LED lighting . . . . candelas, lumens, color, spectrum etc. Light is funny stuff to try and quantify => "how bright is it ? "

Color is an odd thing => LED's produce a VERY narrow spectrum of light, each material has a peak emission point. So, a "red" LED can make various shaded red objects look "differntly". True sunlight has ALL colors in it. So, while the physics books say that red, green, and blue light product white; they are correct. However; that is FULL SPECTRUM red, green, and blue. Mixing light from red, green, and blue LED's will make some things look pretty good; and others look absolutely terrible / unreal.

Enter "white" LED's => a misnomer at best; they are actually blue LED's that are coated with a phosporous material => basically captures the blue light and re-emits it as "white" light. White LED's available today run the gamut; some are pretty darn nice . . some tend towards the yellow => warmer; and some tend towards blue => cooler. Another interesting thing about them is how to adjust their brightness. You CAN'T vary the voltage across them; they all have a very definite forwared voltage. String them up with a resistor and you CAN vary the voltage to the combination; => however the LED itself still has the same voltage across it; in conjunction with the resistor you are actually varying the current THROUGH the LED. The best way to "dim" an LED is to duty cycle it => switch ON and OFF very rapidly => varying the current through them seems to accomplish the same thing => buy varying the current through some LED's also changes their "color" or spectrum.

Most LED's are rated so many millicandelas, or candelas. Along with that is a beam width => some are narrow, some are wider. So how does this translate to "illumination" . . which is after all what we're intending to accomplish ?

Candelas are brightness, in a given area => the beam area. Lumens is brightness in ALL directions. A good example is a "laser" pointer. The beam is very intense; in fact dangerously so. The candelas of such a device are VERY high. However; it is also limited to a VERY narrow beam. Take that same "amount" of light; and spread it in ALL directions; and you couldn't see squat. Lots of candelas, but very little lumens.

Bottom line: How well a light illuminates something is a function of it's brightness AND it's beam / dispersion. Your picture would look like crap / be hard to see if you strung LED rope lights that emitted light in ALL directions => some would be coming directly at you => glare / blinding you. Some would go directly where you want it . . some would go off into the room in all directions. To properly illuminate such a thing would require careful choice of the LED's in terms of their light output, and their dispersion. Don't know how to go about doing that quite frankly . . . I can't quite clearly picture HOW you're trying to illuminate the picture => around the edges ? ? ? illuminating JUST the edges ? ? ? . . .

LED rope lights are probably intended to disperse light in "all" directions; though I'm sure they fall far short of that . . . and they may illuminate the edge portions of your picture just fine but do poorly towards the middle . . . kind of a crap shoot.

While I'm a big fan of LED lighting, and have been for many years; you may be better off doing modern fluorescents. Right now; good LED's produce the same amount of lumens per watt as good CFL's => however; the cost per lumen is about 100 times as much . . . .

Hope that helps some . . .

Bob

Bob Brown,
I'd already considered the color of the LEDs. The piece to be illuminated is a fantasy piece, so a blue-white would look fine. In fact, I'd considered just going with blue LEDs since I love blue, but DH does not like that idea. I don't need any effects, just a soft glow through the body of the glass.

Here is a similar effect - I plan to email these people to see if I can get help from them since they seem to to a lot of this type of installation:

rcmjr,
I've considered fluorescent, but the problem is that I would end up with a very wide border around the item which I'd like to avoid. And then there is the problem of allowing access to replace the bulbs. Certainly with LEDs that would also be a problem but not as often and the narrower margin would be worth the extra cost.

I need a thin linear source of concentrated light that can be directed into the edge of the glass. I do NOT want the surface of the item lit, just light traveling through the glass itself, similar to a fiber optic cable.

The only installation I can find that would eliminate the need for access to replace bulbs is fiber optic edge cable, but I've got more than just the expense as a problem. NO ONE seems to be able to tell me where to get it or how much light output I can get to see if it would be enough.

I've looked at high output rope lighting, but I'm concerned about heat build up. But rope lighting does have advantages for what I want to do. For rope lighting I have found channels intended to direct the light out one edge and channels that would reflect and amplify the light in one direction.

I think I need a frame that would leave a channel for the light to reach the glass edge with seals to prevent light from leaking across the surface of the glass.

Here is the item I want to light:

If you look at the gallery of well lit carved sandblated art glass below, you can see how much better my piece could look with the right lighting effect. This is not the artist that did my piece, but I plan to call this guy tomorrow and see if I can get any pointers!

Here is a link that might be useful: Permanent Reflections

.

Took a quick look at your link . . . I understand better what you're trying to do . . .

FWIW, I'll guess there may be some neon etc involved in some of the pieces shown in the link . . . not a great deal of light, but varied colors and well dispersed.

Can't add any more info; but I'll be curious if you end up finding out anything about how those in the link were done . . .

Good luck . . .

Bob

I'm not an expert, but I think you may be on the right track with LEDs. Their great strength is that they can be very small. Rather than using ropes, I'd be more inclined to try making my own system with long, slender PC boards into which I'd solder 3mm white LEDs. That way I could choose the beamspread and color from a bare LED supplier. I might fit the boards into a plastic or wood channel.

Something else to consider is that you'll probably want a filtered DC power supply to power the LEDs, to avoid the flickering effect you get on AC. Depending on the number of LEDs, even a 800ma to 1a wall wart might work.

Another possibility might be slimline T2 fluorescent fixtures. The lamps are only 1/4" in diameter. Maybe something like this would work for you. These fixtures have a 1" square cross-section - no thicker than some picture frames. They're available in various finishes and in custom lengths from 12" to 48". I can imagine making a frame with these embedded, or maybe even using the fixtures themselves as the frame.

Everybody,
Thank you for your suggestions!

I called the Permanent Reflections artist. He gave me some hints but would not give me specifics since he claims they were "trade secrets". But the first type of lighting he recommended was LED. He did say he has used neon and used to use fluorescents. Then he suggested talking to a sign company since some of the products and methods he uses are taken from edge lit signage.

So I called some of the sign companies in town and finally found one who had a little bit of a clue but wasn't sure of the engineering for the project. He referred me to the LED company he does business with - Sloan LEDs. Sloan will engineer signs for companies. I spoke to one of their sales people who talked to one of their engineers. They called back while I was out this afternoon so I did not get to talk to them, but they said they have some ideas to run by me.

The local sign guy said if Sloan will engineer it, he will construct and install for a percentage of the cost of materials. At this point I have no idea how much they are talking but considering this is an art piece by a nationally known artist who is now limiting his new pieces and who no longer does this type of piece, I'm willing to pay a good amount to display it properly!

I'll come back and let you know what we work out.

Anne

Here is a link that might be useful: Sloan LED

Hello,
Wiring led's directly point to point is not too difficult. If holes are drilled into the frame to allow the led's to point towards the edge, you could probably space them about 1.5 to 2 inches apart. This is definitly a custom lighting job. If a strobe circuit is used, the lift expectance can be quadripled. But the electronics would also be custom. If the segments overlayed 30% the loss of a segment would probably extend the life of the lighting even longer. Changing the voltage will affect the brightness. I played with a design for police flash lights. The higher the frequency and voltage, the brighter the array became. Using a fixed voltage and continuous frequency will reduce the life of the leds.

Mixing yellow with white may give a bright but subdued color. Find an electronics student and tell them what you want. Buy the led's from a online electronics surplus. You probably don't need the highest output lights available, and could use some very inexpensive bulbs. Radio Shack sells bulbs, but the cost would be in the $100's for parts, wheras surplus suppliers would be 1/4 of the cost. The overall thickness of the bulbs is about .21 or 1/4 inch, and the lay down dept would be about .4 or 7/16 inch. This could be put into almost any frame. The life expectancy of a simple design would probably be 10 years, a strobed design probably around 20 years. The cost to leave the light on all of the time would be less than $10 a year.

.

Looks like the Flex lite at the Sloan site may be just the thing for your picture . . . . . a ready made solution. Maybe just get a "custom" one that is long enough to go around your picture completely . . .

Bob

Bob Brown,
I understand theorectically how to put together the system. The problem is that at the moment I cannot do the actual work (I'm going into surgery on the 23rd) and finding someone to do it locally from my descriptions/suggestions has been impossible.

That is why I am willing to work with this man from the sign company. If he gets an engineered diagram from Sloan LED, he will do the work. Yes, it will cost me more, but it will get done in the time frame I need.

rcmjr,
I'm not sure which one they are going to suggest. I've got to leave and I missed the people I need to talk to out in Californis. Maybe they will email me the ideas they have as I suggested on their voicemail.

Anne

Hello,
Making a custom design is basic electronics 103. It is on the same level as learning the transistor, which is just after learning basic ohms law. The difficulity is in finding the best price for the LEDS. Anyone creative can wire the lamps into a frame and hot glue them into place. Determinimg the voltage you want to use, also determines the number of lamps on a string. Like christmas lights. Interlacing the lights is a prudent design, so if a string fails, you do not loose too much light. In this way you could in theory loose 60% of your lamps before needing to replace them. Using a counter strobe will lenghten the life span of the bulbs. Varying the frequency can adjust for brightness.

I would not imagine that a company that sells light strings is willing to do a drawing for this purpose. Although an electronics student at your local jr. college may build the light for a reasonable price.

Bob Brown,
The company that I contacted is charging nothing for their services - but they do expect that the sign company that sent me to them will buy the LEDs from them to build it. I expected that and find it a rational way to do business.

But at this point I am reasonably satisfied (I may not be when I get the price, LOL). They emailed me images of similar installations that they have designed and the effect is exactly what I want. One of the examples is shown under different external lighting conditions so I know how it will look when the LEDs are the sole light and when other light sources are on.

Anyway, a third party company makes a track designed to set mini-LEDs in the bottom, then hold glass above or to the side of the LEDs. One of the installations shows a slab of etched glass held by the track above the LEDs - the track appears to be the sole support.

Aside from the track we'll need to construct a frame of molding around the piece to cover the track - that is the easiest part for me to get my GC to do.

I'll update as I get more news.

Anne

Varying the frequency can adjust for brightness.

I may be mistaken, but I don't think that varying the frequency will vary the brightness, just the perceived flicker.

The reason you can pulse LEDs and get high apparent brightness without overheating them is that the eye's persistance of vision "fills in the gaps." What's perceived is the on-time intensity. However, junction heating is proportional to the average current. That's a function of the duty cycle anyway, not the frequency.

Unless I misunderstand the principles, to vary brightness of pulsed LEDs you still have to adjust forward current. But as I say, I could be mistaken.

.

Adjusting the brightness of LED's . .

For many typical applications like indicator lights etc; limiting current is the way to dim them. The losses are quite small in typical, low output LED's that are used for indicators on your stereo, TV, telephone, whatever . . .

For Luxeons, Piranhas, and others "serious" LED's that are meant to put out LOTS of light, duty cycling, not current limiting is the way the mfr's recommend. There are a couple reasons why it's the best way to do these higher output LED's. First one is that the spectrum ( especially in white LED's ) is impacted => the "color" of the light will change. Secondly; the losses are much less.

Consider a typical LED run at 12 v => figure the current limit of say 20 mA. The LED drops ~ 1.5 volts so resistor required is ( 12v-1.5v )/ .02 amps => about 525 ohms. Losses are about 21 mw in the resistor. Now do the same thing for example; with a Luxeon. Assume same input and same voltage drop; but the current is now ~ 350 mA. Resistor required is ( 12v - 1.5 v ) / .35 amps. Thats about 30 ohms. Losses will now be 3.7 watts in the resistor.

The way to do it is to drive the LED at it's full current level; but to switch it ON and OFF very rapidly. This can be done by a fixed pulse at a varying frequency, or simply with a PWM. Either way will do the LED right; but losses in driving the higher output LED's will be MANY times less . . .

Bob

I thought that PWMing LEDs wouldn't work because they don't have the time constant that incandescents do. They turn on and off very fast. I've read that with the eye's persistence of vision, they appear at full brightness even if the duty cycle is 50% or even 10% - that's the trick behind some automotive brakelights running LEDs at a peak that's several times their rated continuous current.

The junction heat generated is proportional to the average current, but the light output impression is seen as peak brightness. (If you ignore the really horrendous strobing effect. ;-)

Now, if the frequency is high enough, do they have enough of a time constant in turn-on to make PWM work? How high does the frequency have to be? Based on appearance, I'd guess that the auto brakelights are in the 100 - 200 Hz range.

Or might you use PWM with a series inductance to effectively reduce the actual voltage applied?

I'm a EE dropout and I'm running up against my wall of ignorance here. Help me out. ;-)

.

Davidr;

You are correct that incandescents DO have a "persistence" that LED's do not have. Watch an LED brake light; you'll see it's instantly ON . . . and OFF. Very different than an incandescent one.

Switching multiple LED's on and off "very" quickly is referred to as multiplexing. Turn one on while the others are off, and go "round robin" 'til they've all been on . . then start over again. This has multiple benefits => lower power consumption, and FAR fewer lines to drive them all.

The persistence that really applies to such situations is the persistence of the human eye => I've got lots of products out in the field with 4 digit LED displays, all running at 25% duty cycle => only one digit on at a time . . but the "round robin" is fast enough so that the persistence of the human eye perceives them as ON constantly.

Speed of multiplexing => fast enough to avoid it looking like it's flickering . . that's about the only real concern. With older fluorescents you wanted to avoid 50 / 60 Hz and it's multiples to stop the "strobe" effect from happening. Other than that; no real boundary on speed of switching. Once you've gotten above the speed the human eye can perceive; there is little gain in going faster => switching losses increase but there is no gain in visibility.

You are correct that you can pulse LED's at higher than rated current; usually a mfr will spec some value and a duty cycle you can use. In reality; those numbers are based upon THERMAL considerations of the LED => heat is the biggest degrader / destroyer of LED's !

You CAN'T reduce the voltage across an LED => they have a pretty fixed voltage drop just like other types of diodes => if you get below it; they fail to conduct and therefore light up.

Hope this helps . . . if you want more info, and a REAL slick / simple animated GIF that will demonstrate persistence of the human eye; send me a real email address for yourself to my member mail and I'll send along further explanations of whatever you'd like plus the GIF . . .

Bob

Thanks for the answers. One thing that I'm still puzzled about is whether (and how) it's possible to vary the perceived brightness of an LED. AFAIK, LEDs average the temperature but not the brightness. What am I missing here?

Sorry, I meant to write :

"... whether (and how) it's possible to vary the perceived brightness of an LED using PWM."

.

Davidr;

Take a 1 millisecond "window" . . that's a frequency of 1000 Hz, or 1 kHz. During that 1 msec window; turn ON the LED for one microsecond, and then OFF for the balance of the 1 millisecond. Do this constantly . . . . your eye will perceive a very dim light . . . perhaps none at all. Now turn that 1 microsecond into 10, . . or 100 microseconds of LED on during the 1 millisecond period. You will now see some light. If you turn that 100 microseconds into 999 microseconds; you will now see the LED at essentially full brightness.

PWM in it's simplest form; is nothing other than varying the ON and OFF ratio . . . . done at a rate that is faster than the eye can respond to gives the eye the illusion of dimmer / brighter . . remember that during the ON of the LED; it is at / near it's rated current; and therefore correct "color" or spectrum. You're simply turning the thing fully ON and fully OFF; at a rate that is faster than the eye can discern . . and gets perceived as brighter or darker . . . . .

Make sense ? . . .

Bob

Thanks. Actually, I understand how PWM works, but I appreciate the review.

Here's where I'm still confused though. As I wrote above, I read some time ago that the reason that pulsed LEDs can be overdriven to appear bright is that the eye's persistence of vision doesn't fade significantly, and LEDs turn on almost instantly. You can have 10% duty cycle (and 10% junction heat) but the eye is fooled into seeing that 10% of the time the LED is lit as essentially continuous. To the eye, it appears as bright as if it were operating all the time.

To give an example, suppose we have an LED spec'ed at 20ma forward current. We hit it with 200ma, but PWM the signal so that its duty cycle is only 10%. The junction heating is supposedly equivalent to the average current - the specified 20ma - so the LED isn't damaged. But to the human eye, the LED appears [almost?] as bright as it would if it were getting 200ma continuously.

It seems as if you're suggesting the opposite - that either our vision doesn't respond very rapidly when a light source turns on, or else that the response time of an LED is such that the brightness ramps up after power is applied.

I suppose that the latter must be the case, that there is some time constant when power is applied to an LED. How long does it take for an LED to reach full brightness?

As a related question, I note that the pulsed LEDs I've obsereved seem to be operating at a relatively low frequency, perhaps 100-200 Hz. Given a 10% duty cycle (for example), is there a PWM frequency at which the eye perceives 10% duty cycle as 10% brightness, instead of the [nearly] 100% brightness we get from pulsed LEDs? What is that frequency?

Thanks again for your patience with my questions!

.

Do you dim LED's with frequency or duty cycle ? Well, in a word; yes. Can you do them at 100Hz, or 10 kHz; well . . yes.

I don't have any numbers on eyeballs . . in terms of their response times . . . but there obviously is a certain minimum time for the eye to respond to light; and there is minimum persistence time as well.

LED's are virtually instantaneous in both "starting" and "stopping" giving off light. There is a little perstistence in the phosphorous used to create "white" LED's; but it's very small and can be ignored. I'm also not sure if all the rods ( or is it the cones ? ) in your eyes respond at the same rate . . . or not . . . . different rate for different colors I think . . .

I remember one project I did where I had 4, seven segment LED digits to run . . . had to multiplex them to reduce required # of wires . . . With 4 digits, each was on for 25% of the time . . and I believe I ran them ON for 4 msec. While they weren't as bright as they would have been if they ran 100% of the time; they were VERY readable in all but the brightest of light conditions => bright outdoor sunlight. I tried them at 5 msec each; and there was some perceptable "flicker" to them. So, running them at 4 msec each, the LED's were at 25% duty cycle; and a frequency of ~ 62.5 Hz. If that frequency were a little different; say 60 Hz; then . . . under old fluorescent lights they'd have flickered TERRIBLY . . under newer / electronic fluorescents they'd have done fine. So, you can even do it at less than 100 Hz; though one needs to avoid known "trouble" frequencies like 50 and 60 Hz. By the way; these were run at their NORMAL LED current; they were not run higher due to shorter ON time. This was done mostly to keep consumption down . . and they still look pretty darn good. No complaints at all from customer. You don't HAVE to run LED's higher than their constant rated current in order to mux or PWM them . . . . but if absolute maximum brightness is a concern then you should.

Just as another example that all this is inter-related => take and LED and run it at rated current. Run it at 20% duty cycle say . . . do that at 0.1 Hz and you've got a blinker ! . . full ON for 1 second, and full OFF for 9 seconds. A blinker or flasher as far as your eye is concerned. Now keep that same duty cycle; but increase the frequency . . . and at some point; the LED will appear to be "steady" ON. I'll guess that'l be somewhere in the 10's of Hz . . . but it will appear constantly due to persistence of the human eye. Now lets say you've decided to run higher than rated current to get more light . . usually there is a spec on the current level; AND the ON time and duty cycle. Yes, this all relates back to heat in the LED => there is a certain max current; and a max amount of time that it can be present. Exceed that and you've potentially damaged the LED and/or reduced it's lifetime.

It's a combination of frequency, duty cyle, current level, and persistence of the human eye, and the light level you need to have for a particular application that determines how and what you do. They ARE inter-related with each other . . . .

Hope this makes sense . . .

Bob

You may find the answer to your question in the Rope Lighting Forum...

Here is a link that might be useful: Rope Lighting Forum