Hi all,
I'm trying to build a high power, high frequency LED driver for an optical communications project. My goal is to be able to power some high power LEDs (~1 amp each) at a high frequency (~30MHz or more) for data communication. The idea I had was to create a constant current source (most likely a simple circuit using an LM317 regulator), then try and modulate the brightness of the LEDs. Right now, I'm trying to figure out a good circuit for modulating the LEDs that quickly, as well as any advice on how to combine the constant current source circuit with the modulation circuit. I've seen several ICs available from MAXIM and other companies that act as high power LED drivers with switching frequencies up to 2MHz, but this is still lower than what I need. Any advice would be greatly appreciated.

Thanks, Gavin

 

Hello,

Just take a look at the ronja project:
http://ronja.twibright.com/

The project uses optics for a datatransfer link.

Greetings,
Bertus

 

I took a look at the site and read through some of the documentation for the project. Although the Ronja project does fulfill my goal of transmitting data, the project is much more complex and involved than what I'm looking for. The site explains how to build a complete optical data transmission system. I was hoping to only build the LED driver that would allow someone to transmit information a short distance (something that could potentially be used in an office building).

 

Can you define "short distance"?. Why is using 10's or 100's of watts of white light superior to IR or radio? Don't forget laser diodes and optical fibers.

We might notice that 30MHz is not much bandwidth for data transmission and multiple users. Will this compare with Wi-Fi?

 

I remember a something at work that went considerably higher than this. Solid state lasers are naturally polarized (LEDs aren't, far as I know). They ran it through a device that had a lense that polarized and depolarized, which alternately block and passed the laser light. Don't know if this would apply.

 

Yeah, that's a Kerr cell, or used to be.

 

Thanks for replying. By a shorter distance I mean anything from a few inches to a few feet. This is part of a LED lighting project at my university. I do understand how using lasers would work better and I realize this project would use a lot of power, however, the idea behind this is to demonstrate one way LEDs can be used. If I get it to work, then I'll proceed to look into other cheaper, more energy efficient alternatives. Right now though, my research is based on LED communication, thus I need to use LEDs and not lasers or other means. I do appreciate the alternate options though. If you do have any suggestions for using LEDs, please let me know.
Thanks again

 

Cool project. Since you will be using the room lights to transmit the information, you will need a much lower switching frequency. At 30Mhz, reflected light and light from different distances will corrupt your data.

If you switch to a lower frequency < 1MHz the project hardware will get much easier and the data transmission will be much more reliable too.

Good luck with a cool sounding project.

 

Well, you could polarize the LED light...

Seriously, I worked in the telecom industry all through the evolution of lightwave, which you are recreating. They started with LEDs, moved to simple solid state lasers , went to high precision very pure lasers, then started stacking different colors (each a different laser) in a fiber optic cable. Through all of this the fiber didn't change (literally in the case of ocean spanning cables).

What kind of results have you gotten just turning the laser on and off? Another possibility is to do what ECL does, don't actually switch it completely, just vary the intensity between two levels, the LED stays on full time. You could recover the digital data later from the stream.

At this point I wouldn't worry about coupling at all. 30Mhz is high, but doable.

Also, what kind of reciever are you using? A phototransistor? Whatever it is, it has to have the freq responce you need too.

 

@ Bill Marsden

I believe we will be using a phototransistor. So far in the project, I've just been told to work on the LED driver itself, so I haven't looked at the receiving end yet. I actually was hoping to do what you said (leave the LED on the whole time, just vary the intensity level). I figured that way I would need to do some work on the receiving end to recover the data, but that would be fine. What kind of circuit would you recommend for varying the intensity like that? Even if it didn't get the high frequency I need, I would like to try something out and get a better understanding of how it works.

@ THE_RB

I'll try and give your idea a try and go (use a much lower frequency around 1MHz). If I did do that, what type of circuit would I use to modulate the light? I built a PWM circuit with a 555 IC, but the max frequency out of that was around 130kHz. I was hoping to find a newer version of the 555 IC that could output higher frequencies, but I'm not sure if that exists.

 

You can get that kind of bandwidth at an amp if you make a constant current source and use it to drive the emitters of an emitter switched pair. An emitter switched pair is much like a differential amplifier, but you drive it with a signal that saturates the switch in both of its states.

Be sure to use RF transistors such as those used for mobile radios -they operate in the 12 to 28 volt range and some can handle an amp. You may well have to use a cascode on the transistor that provides the output pulse. The other transistor can connected directly to the power supply, so no need to make it cascode.

Kind of like this, but with different parts:

http://cappels.org/dproj/ledsw/Fast_LED_Driver.html

 

This strongly resembles ECL logic, which is what I was refering to earlier. Whoever told you that you don't need to work on the reciever is not doing you any favors, how are you supposed to measure and test the transmitter? They are matched, one works with the other. I strongly recommend coming up with something to measure the signal from the LED, or pestering the other guys to give you a prototype. Else, you are designing in the dark, and only hope it works.

Dick's design looks like it will work, but it is a bit overcomplicated. You could also shunt the current away from the LED with a transistor to modulate it. That is, put a transistor in parallel with the LED in question, when the transistor is on it will take the lion's share of the LED current. A simple regulated power supply and resistor is all you really need to turn an LED on.