Hello. I have a some questions about driving our epitex LEDs to achieve the largest energy delivered. I was wondering if you or someone else could tell me how I can drive the LEDs beyond the max forward current? The max forward current is 100mA for my LEDs, and the total output power is 48mW (e.g. for 3mm plastic LEDs I just ordered L810-33UP). I am attaching a PDF of the specification sheet of this LED. It says in the spec sheet that the max pulse current is 1000mA, and the duty cycle should be 1% for a pulse width of 10us. In our application we are turning the LEDs on for about 5ms so it is my understanding that the total energy delivered over this time would be 48mW*5ms=0.24 mili joules, when pulsing it would be 480mW*1% duty cycle*5mS=0.024 mJ. If my understanding is correct than it would make sense for me to stay at the max forward current of 100mA, and not pulse my LEDs to obtain the largest amount of energy.

I was wondering if I could pulse my LEDs in such a way that I could obtain more than the 0.24 mJ of energy over the 5ms? For example, could I pulse the LEDs at 1000mA for 5 ms as long as I waited 500 ms before I turned them on again, so the duty cycle would still be 1%? In this way I would have 480mW*5ms=2.4 mJ of energy delivered. I guess what I am wondering is if the 10us pulse width is the max, or I can increase it as long as the duty cycle doesn't go beyond 1%?

Currently I am using the Allegro 6277 LED drivers to drive my LEDs at 100mA, see link. (http://www.allegromicro.com/en/Products/Part_Numbers/6277/6277.pdf). Although I recently purchased different LED drivers that should be capable of driving the L810-33UP LEDs. The new LED drivers are made by ON Semiconductor NCP3065PG, see link. (http://www.onsemi.com/pub_link/Collateral/NCP3065.PDF). I was wondering if you could tell me how to wire this LED driver to ensure that the L810-33UP LEDs I purchased will deliver the maximum energy over 5ms and still not cause the LED to break?

Thank you in advance for your assistance, and the LED and driver spec sheets should be attached.

Kevin

 

Key Specs:

Pulse Forward Current IFP 1000 mA Ta=25°C

‡Pulse Forward Current condition: Duty=1% and Pulse Width=10us.

In other words you can not exceed 10µs at 1A, and the max freq for this kind of drive is 1Khz, per datasheet.

Or you can run it continuous at 100ma (a very high number for LEDs). Even then, if the forward dropping voltage is on the high side, it will be out of spec for power dissipation (170mw).

So what will it be, pulsed or continuous?

It is generally not a good idea to run any electronic component at 100% of its ratings. It is a good way to dramatically shorten the lifespan.

What is the application?

 

Thanks for the quick reply Bill! I am using individual LEDs to illuminate the eyes in sequence, and I also have a CCD camera that acquires an image when each LED is on. As I mentioned currently I am using a continuous 100mA solution for about 5ms. There is a long wait before the LED is turned on again (~ on the order of 100ms). As to whether I would like to use a continuous or pulsed option, I would like to obtain the solution that drives each individual LED such that I can obtain the maximum energy over a short time (e.g. about 5ms)? So I don't have as much eye movement when I take an image.

I have been using similar LEDs for several years that are driven at 100mA, and have never had a problem with them burning out. As you said the spec sheet says I cannot exceed 10µs at 1A, but what about another current larger than 100mA but less than 1000mA? I don't understand the current and pulse width relationship that exists between having the LEDs continuously on at 100mA, and with a pulse of 10us at 1000mA? For example, if I pulse them at 5ms can't I safely increase the forward current beyond the 100mA?

Thanks again for your insight.

Kevin

 

I would make an inquiry via the manufacturer's technical support team.
Since they make the devices, they should know what it is capable of.

That way you can have a truly authoritative answer in the least amount of time.

 

You can use a photodiode as a detector (hook it up to a scope) to help you optimize things. A plain old LED can also be made to work as a detector; it's just not as efficient. I took an el cheapo Radio Shack photodiode and hot-melt glued it into a chunk of PVC pipe. I installed a BNC female jack on one end. This is a very handy light sensor; I can hook it up to my DMM and go outside to measure sunlight or use it in a dark room. It also works well just connected directly to a scope for measuring transient stuff -- or hook it to an averaging system digital voltmeter for quantitative work in the presence of noise.

If you control the measurement geometry, then you can make quantitative comparisons amongst the LEDs and operating parameters (e.g., current pulse width, height, etc.).

Definitely start with the datasheet values as guides and contact the manufacturer as SgtWookie suggested for more guidance. If you can afford to sacrifice some units by testing them to failure, you can make some estimates about how much above the rated specifications you can run things -- as Bill implied, if you're willing to trade early failures for better performance. But realize that, often in the real world, the benefits gained by increasing an operating parameter often have a benefits curve with a negative second derivative.

Another thought is to run some units in parallel and collect their light into some e.g. solid acrylic tubes (light pipes) or some optical fiber. You might be able to get the illumination levels you want without running the LEDs at such strenuous currents. You could bend some 1/4" or 3/8" acrylic rods with a heat gun to an appropriate shape and wrap them with black electrical tape (or heat shrink tubing) to minimize spurious illumination and put them over the LEDs' output.

 

Using more LEDs as someones dad suggested is an excellent idea if possible. LEDs are more efficient in terms of light output per W at lower powers. It probably varies from LED to LED but you might find that running 3 or 4 at 100mA will give the same light output as 1 at 1000mA. I'm just guessing on the figures.

 

You can also put them in series, which also increases the immediate efficiency.

The chips you show also substitute for a commercial device called a buck puck, but in this case the current is low enough it isn't needed.

I've written an article for beginners on LEDs. It doesn't apply well to your flavor of LEDs, but the basics don't change, and I gather you are not a beginner.

LEDs, 555s, Flashers, and Light Chasers

I'm currently in the middle of a rewrite on the forum version.

 

Thanks everyone for your advice. I have been using LEDs on and off for several years, but I defiantly think that I could be using them better. I have a few specific questions/comments about some of your replies and a few general questions:

1) Bill_Marsden: I read your article and it looks really good. In particular one point jumped out at me. In chapter you say "If you are dealing with a stable power supply a resistor is good enough." Well it just so happens that I THINK I am using a stable power supply. I am using a 5 V DC power supply that plugs into the 110 AC outlet, so I think it should be constant? In addition to having the LEDs be as bright as possible, another one of my greatest concerns is that I am illuminating ~30 different LEDs at different times and I need them to have the same irradiant intensity as possible. That is why am using the Allegro 6277 LED drivers that I mentioned before. I thought I was "matching" the current for all the outputs. After reading the application notes for the ON Semiconductor NCP3065PG LED drivers and your article it seems to me that I should be using LED drivers only if I have a battery that will supplies my power? It seems that I could use just one resistor and as long as I only turn on one of the LEDs at a time then each of the LEDs output should be the same? Can you see any potential drawbacks to doing this (e.g. the LEDs would output different levels)? Another problem I have is that I don't know how to use the ON Semiconductor NCP3065PG LED drivers I just received? With the Allegro 6277 drivers I just need to place a resistor at one of the pins to specify the current, and then wire each LED to one of the outputs on the driver. I was thinking I could do the same with this driver, but it doesn't appear so? It looks like I will need to do more work. Any suggestion how I can get it to work quickly?

2) someonesdad and Markd77: I too would love to couple several LEDs into some type of light pipe to increase the irradiant intensity. Also another aspect of my application is that I need to have the smallest light source size. This is why I am now working with the 3mm plastic LEDs when before I used the 5mm LEDs that were brighter. It would be even better if I could decrease this size more and obtain irradiant intensity. I purchased some plastic fiber optic cable and a fiber optic kit from http://i-fiberoptics.com, and I purchased some LEDs from Epitex that have the die closer to the surface of the LED (supposed to work better with coupling to fiber optics). I am not sure if I was doing it right, but I was only able to get about 1% efficiently that I could get with having the LED in free space. All I did was take images with a white piece of paper with my CCD camera and compared the pixel intensity. I didn't do anything fancy when I tried to couple the LEDs with the fiber. All I did was drill two holes at different ends in some Plexiglas with one hole that was the width of the fiber and the other was the width of the LED. In this way I was trying to "touch" the LED die to the fiber to get the best coupling. Obviously I don't want to have couple~100 LEDs to get the same intensity that I could get with one in free space. Can you or someone else give me some suggestions on how I might better couple my LEDs together? I read your comments about the acrylic rods, but I don't even know where to start...

3) SgtWookie and someonesdad: I also contacted the manufacture to see if they could provide the more details that I need to know about driving my LEDs beyond the max continuous forward current. Unfortunately, they haven't responded and I am not holding my breath for a response that will be helpful... As someonesdad suggested I am now planning on sacrificing some of my LEDs to determine how much I can drive the LEDs beyond the max forward current. I am going to pulse the LEDs for about 5ms and turn them off for at least 500ms (i.e. 1% duty cycle) with a 5 V DC power supply and supply a series resistor with a value to ensure that I get ~50mA, 100mA, 200mA, 500mA, and 1000mA. I will then use my CCD digital camera to acquire images during each pulse. Therefore, I should be able to monitor how the performance degrades over time by examining the average camera pixel value over time. I already sacrificed one LED by continuously driving it at ~400mA, and after several minutes I could visually see that it was becoming dimmer.

Thanks again for everyone's suggestions/comments.
Kevin

 

Without knowing exactly what power supply you are using I can only guess, but most wall wart types are not regulated. There are no standards with them, so there is also a chance that it is.

OK, lets assume it is a regulated power supply. Your LEDs have a nominal value of 1.55, but they can go as high as 1.65. This means you can only have 2 LEDs in a chain, for a total of 3.3V (3X in a chain is 4.95V, which is uncontrollable). If you use a 82Ω resistor and the LEDs are 1.65Vf then you will have 20.7ma, but if the LEDs are 1.55Vf then you will have 23.1ma. Both are borderline acceptable, but it illustrates the problem with tolerances, 91Ω is better.

Increase the power supply voltage, you will radically reduce the number of resistors.

Something you may not have thought about, TV cameras take 30 pictures per second. If you pulse these LEDs you may have some strange artifacts come out of it, like the image slowly fading in and out.

 

Thanks again Bill. I currently am using "switching" power supplies that plug into the 110V outlet. Is there one that would perform better, and how do I know what to "look" for? The load on the power supply should be the same for each LED, so any output voltage differences would be from the 110V coming into the power supply.

I am not currently placing any LEDs in a chain, and I will continue to do this unless in the future I can efficiently couple them into fiber optic or acrylic tubes. I only turn on one LED at a time. If I understand I can expect an error: 5V-1.55V=3.45V with a R=36 then I=95.8mA, and 5V-1.65=3.35V with R=36 then I=93.1mA which gives a ~3% error in the current. I also noticed that Vf could be up to 1.8V so I=88.9mA and an error up to 5%. My question is that if use LED drivers is this error reduced, and if so by how much I could expect it to be reduced? If I continue to use my LED drivers I have to use at least 4 of them because they only have 8 outputs each. Therefore, I would expect an error from there if they each had there own resistor, and their is 5% error in the resistance?

Also my CCD camera is triggered, so it will only expose an image when one of the LEDs on.

Thank you.
Kevin

 

With most electronics 5% is considered negligible. If you have a component where it is that critical you back off on the specs, or make it adjustable to tweak it. Even with a driver you will need something to tweak the specs, since it will be using the same 5% resistors.

I've never heard about syncing a CCD camera to the light source. I'll be interested how that works out.

 

If you can afford to sacrifice some units by testing them to failure, you can make some estimates about how much above the rated specifications you can run things -- as Bill implied, if you're willing to trade early failures for better performance.

Sound advice. Some years ago, in the early days of Blu Ray development, I got to visit Nichia, inventors of the blue LED and blue semiconductor laser. They had a customer who was driving their delicate early prototypes at many times the rated power, but with a very narrow pulse, and Nichia were happy (and a little surprised) to report they were getting away with it. Much can be gained from going out on a limb, followed by thorough validation, as even the smartest OEM can't predict every application for their product.

 

Thanks Bill. I am thinking that I will start to simply use a resistor instead of the LED drivers. The other part that gives me some error is the orientation differences between different LEDs, and I think that this is more than the 5% difference that could arise from the different forward voltages. It is my understanding that a "switching" power supply is regulated? If this is indeed the case than the only differences between my LEDs would be the difference associated with the different LED forward voltages.

I am syncing my individual LEDs to the CCD camera exposure by using the CCD camera's external trigger option. I am using a USB data acquisition device with timed digital outputs to turn on/off the different LEDs, and it also is used to send the trigger signal to the camera at the same time. The DO signal switches a reed relay that completes the circuit to turn the LED on/off. The DO signals also go through a set of OR logic gates and if any of the DO signals are high it will send the acceptable signal level high for the camera to expose. I have the CCD camera set up so it will only expose the image when the trigger signal is high. There are probably better ways, but this method has been working for me.

I am currently trying to test the limits of the LEDs by sacrificing some of them as suggested. I am trying several different currents 30mA, 100mA, 300mA, and using short pulses of 5ms. I then take several succesive images (e.g. 1000), and then I report the average pixel value as function the number of images. I am currently waiting 500ms between each image, so hopefully it will "cool" off before I take another image. I am still working on this, and you can see from one of the attached images that with a 12 ohm resistor the average pixel value starts to drop off after several images.

I also just recieved the attached spec sheet for my LEDs that according to the OEM should specify the max forward current I can use for a specific pulse width (upper right figure in PDF). Although I am having difficultly reading it. I was wondering if anyone could make sense of it... For example, if I have a 1ms, 5ms, 10ms pulse what is the max current I can use and how long do I need to wait before I can turn that LED on again? Thanks again for everybody's help.

Kevin