True, but it doesn’t give a separate normal current, and most things in the “Basic Characteristics” table is defined at 100 mA current. I’m inclined to think that it’s acceptable to use with a 100 mA constant-current supply.
That does not mean that the device can be driven continuously at that current. Most often LEDs are driven pulsed and you have to take into account the duty cycle, etc. There are ICs which are designed to drive arrays of LEDs in a pulsed mode. Simple approximations may work if you are lucky but you may be overlooking issues. All I am saying is that there are many more issues involved than just adjusting the current. That is why I am saying that this is not so simple as it may appear at first to non-professionals.
The safest and most efficient way is to use LED driver chips. But they’re kinda difficult to use. The next best option was given by scr4: use constant current sources.
Because an LED has a negative temperature coefficient, an LED “likes” to be powered from a constant current source. Depending on the source voltage and the value of the resistor, a series resistor may be a *good *approximation of a constant current source, or a *bad *approximation of a constant current source. If it’s a bad approximation of a constant current source - which is often the case when you have a small-valued series resistor - you run the risk of smoking the LEDs due to thermal runaway.
So it’s always better to power LEDs with a true constant current source. The simplest is an LM317 wired as a constant current source; a resistor goes between the output and the adj pin, and the value of the resistor is 1.25/I.
Here’s an example:
Source voltage: 32 VDC
Desired Current: 50 mADC
The resistor would have a nominal value of 1.25/0.05 = 25 ohms. We choose 24 ohms since it’s an EIA standard value. The resistor will dissipate 65 mW, so a 1/8 or 1/4 W resistor would work fine. The actual current would be 52.1 mADC.
There needs to be at least 3 V between the LM317’s output and input pin. (Assume 3.5 V.) And the resistor has 1.25 V across it. Thus the maximum voltage available to the string of LEDs is 27.25 V. If we assume a worst case forward voltage drop across each LED (1.7 V), this design could power 16 LEDs wired in series.
Well, in that case it’s fortunate than I’m a professional electronics engineer.
Me too 
In most datasheets I have seen there is language explaining what “Absolute maximum ratings” are and it contains language along the following lines:
As an engineer I have always understood that the “absolute maximum ratings” were not to be the normal operating conditions.
I would not drive those LEDs at 100 mA DC and if I was even anywhere close to that I would make sure I knew what I was doing very well, including thermal dissipation.
I would recommend a pulsed drive which allows much higher peak power while keeping down the average power.
I am not a designer of LED drive circuits but I have visited several factories in China as a purchaser and I have seen first hand some of the pitfalls so that even some professionally designed products did not pass our tests. It is not as simple as throwing a few LEDs and resistors together.
Actually that has been my understanding as well, but I’m confused about this one because it doesn’t list any other recommended/normal operating current. If you were using this device and trying to obtain as much output as possible without risk of damage, what operating current would you use? (Assuming proper heat dissipation, and a pretty good constant-current source)
It is a bit odd that they list 100mA in the Absolute Maximum Ratings, and yet give several other specs later at 100mA current levels, implying you’re supposed to run it at that current (not necessarily at DC, though). They do give a separate higher peak current, and even a “surge” current limit in that same section, making me wonder just what the 100mA limit is doing in the Absolute Maximum Ratings section.
Anyway, it looks to me like Figure 3 shows that at DC, you shouldn’t run more than 30mA through it. But the only way to approach the 170mW power limit they spec is 100mA at 1.7V, though.
It looks to me like they’re intended to be driven with a pulsed 100mA, relatively accurate current source from an LED driver chip. I’d avoid this LED if you’re using something simple like a series resistor and a DC voltage, unless you stay way under 100mA.
Wow! I’m glad I posted here… now I’m daunted.
Maybe my best best is to buy LED Ribbon, but 9’ with the power supply and shipping is around $250 US. Compare that to free LEDs and some sweat equity, and I’m still motivated to giver a try.
If I should go less than 100mA, does that mean, say 80mA? Assuming so and sticking with 24 strings of 6, does that mean the only change is replacing the 18Ω resistors with 22.5Ω? (do I round up to 23?)
Thanks again to all who have replied!
I missed this when leaping to 80mA.
Drat! I fear they might not be bright enough at 30mA. I’d hate to string together a couple hundred solders just to find out it doesn’t work.
What’s the biggest risk if I boneheadedly go with, say 75mA? Is it heat? Longeveity? Death?
ETA: What of these pulsed LED drivers? Are they something a n00b like me should look into?
I am not even going to attempt to advise you on this project because I do not know enough about it and there are just too many variables. The acrilic, the camera, the dimensions, etc, will all affect the intensity of the IR you need. You are not following a precise recipe which will lead to a guaranteed result but rather you are developing a general idea and this means you need to understand things in depth so you can build, test and correct along the way.
Note that the LEDs you mentioned are for surface mount (SMD) and are probably not what you want. You can probably better handle through hole devices.
I would recommend you study the datasheet and understand all the data and charts. That will give you invaluable knowledge.
Note that the amount of light emitted is roughly proportional to the current so you can just use more LEDs as needed.
The specific acrilic and other characteristics will affect the system. Very specially the camera, the filter, etc. As I say, I really do not know enough to give you any concrete advice.
I’m only after advice on LED wiring - the rest I think I have down solid enough - at least enough to experiment meaningfully.
I am truly appreciative that a couple of electrical engineers have provided insight. If I may prevail for just these questions:
-
given my revealed ignorance, should I even think about LED driver chips and pulsed power thingies?
-
If I use a simple DC/resistor and exceed 30mA (say 75mA) what’s the worst that can happen (aside from wasted time with all of those solders) - is it the heat, and if so is the danger fire or merely component failure? If the latter, are we talking instant poof, or instead of lasting 50,000 hours they will only last 10,000?
I wouldn’t pulse them - it’s likely to create problems (“beating”) with your video.
Running at greater than the recommend duty cycle will shorten the LED’s life, but in my experience you can run at 2x the recommend power for on the order of minutes before you kill the device. Still, increasing the resistor values is easy to do, and if you allow a decent margin (which I didn’t do on my first quick glance at the data sheet), you should have no problems.
What makes you think 100 mA will produce enough illumination, but 30 mA will not? Radiometry is tricky business, and it takes a very detailed calculation to estimate the required power to this accuracy.
It’d be a lot simpler to just start out with a few LEDs, make some measurements (e.g. with the camera’s exposure set very long) and determine how many more you need.
If you exceed the rated current, the LEDs will die quickly. Don’t even think about it.
As for pulsed LED drivers, I don’t think it would gain you anything for this particular application. The average power each LED can take is still the same. PWM is more energy-efficient than a linear regulator, but if you’re running off an AC adapter that’s a moot point. The only thing that can make a big difference is if you can pulse the LED in sync with the camera (i.e. the LEDs turn on only when the camera is accumulating an image).
I don’t know anything, but even before posting was fearful these LEDs would be too dim. Having to dim them just increases that concern - but I’ll give it a try and it will either work or fail. If they fail, I’ll spring for the LEDs others have used successfully. At least now - thanks to everyone here - I am in a knowledgeable enough position to give it a try. (enough to be dangerous, I know…)
If all goes to plan I may have some results to report on the table itself, which I’ll do in a separate thread.
I would definitely look into using an LED driver chip if you were building commercial units. But if you’re just building one or two units, I probably would not use a driver chip. I would also not use a simple series resistor. With just one extra component (an LM317), you can build a true current source, which is much better (IMO) than a simple resistor. See my previous post for more info.
As for current level, I believe it’s primarily a heat related issue. No one can define an absolute threshold current level that you must stay under; reliability theory is based on statistics and random variables, not absolutes.
If I were you, I first do some experiments to determine what current level you need.
In this application note by Vishay it says
Correction: The LEDs I actually have are these ones (VSMG3700-GS08) Many differences (not the least of which is they’re 850nm not 950) but they appear to me to be essentially the same electrically so I’m confident the advice provided here is applicable.
Update: After digesting the information everyone here generously supplied, I figured I would start with 20 sets of 7. I ordered a mess of resistors (25 each of 3.9Ω, 2.2Ω, and 1.5Ω).
While waiting for the resistors to arrive, impatient ole me, I rummaged around my garage and found a 3.3Ω resistor (which should draw 36 mA). I soldered a string of 7 plus this resister then fired it up.
WOW are these things bright! I have night vision specs. Wearing these, covering its IR emitters with my hand I expected to see 7 points of light. Instead I saw a blazing bar of light as long as the string. The room I’m in is in the basement and has no windows - completely dark. The string was so bright, using the specs I was able to read a book from the other side of the room, with the spine of the book facing the LEDs! (well, actually the focal length of the camera in the specs is such that at arm’s length I couldn’t focus on the words sufficiently to actually read, but the contrast of ink and page was more than sufficient).
But oh the heat! I couldn’t touch the wires they were so hot! After 20 min, they dimmed down to 7 points of light - barely glowing.
Based on this test both successful (they should be bright enough) yet unsuccessful (they didn’t last very long) - when the resistors arrived I combined a 3.9Ω and a 1.5Ω (to draw 27mA). Knowing these things are brighter than I expected and thus I would need fewer around the table, I soldered’em at 2½" apart instead of the 3/4" in the first test. Finally, I set up the string in front of a large house fan on full blast. When I powered this up, I couldn’t tell by eye if they were dimmer (through the night vision specs). They were still at the same brightness after 13 hours.
I then tried at 3.9Ω by itself (30mA) and it ran for over 12 hours with no fan. It was slightly warm to the touch. Woo hoo!
Long story short: Using the information you all provided, I have successfully wired all the LEDs I need. Over the weekend I finished constructing the surface of the table, positioned the projector, computer, and the modified PS3 Eye, and all tests have passed. My finger touches register brilliantly on the PS3 Eye video capture stream. All that remains is to set up touchlib, which I will work on over this week.
Once again, thank you all for your valuable assistance! It contributed to a (so far) successful project.
What is the voltage of your power supply?
Based on your numbers, it would appear the voltage drop across the resistor is much lower than the voltage drop across an LED. :dubious: This means the voltage across each LED is regulated more so than the current. I’m glad they seem to work, but I am very surprised they didn’t cook due to thermal runaway. (Keep in mind that you can never really predict the voltage drop across an LED.)
As mentioned above, LEDs like to be powered from a constant current source, not a constant voltage source. A voltage source of around 12 V in series with a 3.3 ohm resistor is not a very good current source; the source impedance is too low.
I don’t mean to sound like a broken record, but IMHO it would really be better to use an LM317 and resistor (configured as a constant current source) rather than just a resistor.