Infrared Communication

Does it have to be an LED, or could you use a collimated IR laser diode? What kind of range are you looking for?

Use a parabolic mirror on the receiving end to gather more of the light energy and focus it in your receiver.

Also, you can pulse the LED at higher current than you can run it constantly, so high energy pulses will gain you extra distance.

If you are sending data, there are a variety of noise mitigating data encoding techniques that allow you to squeeze good data out of a poor signal.

How would you pulse high current when you are transmitting data over a 2.3Ghz carrier? Would you have circuit ideas?

I don't think you can pulse an LED that fast. Are you sure you're not talking about an RF transmitter?

Your carrier may be in the GHz range (normally RF energy carried on a cable or transmitted by wireless using an antenna "cut" for the wavelength), but your original data signal may not be. Data is normally sent in packets through a modem (modulator/demodulator). Data is often multiplexed onto a carrier frequency through various means, and there is a lot of confusion between data speeds and carrier frequency.

Common modem speeds for packet (especially radio) are 300, 1200, 2400, and 9600 bps (bits per second). Many are familiar with PC modems which operated over audio carrier frequencies (usually 20-30 KHz max) on DC telephone lines, allowing 14.4K(bps), 28K, and 56K, with some pushing even further by using compression and other digital techniques. Satellite and cable modems push these limits even higher with the top speed for the Internet and other telelcommunication systems ranging from 300K into the Megabits range. The fastest modems push the limits to the 100's of megabits, and usually are used in conjunction with Microwave RF transmission. Wikipedia says "Some microwave modems transmit more than a hundred million bits per second. Optical modems transmit data over optic fibers. Most intercontinental data links now use optic modems transmitting over undersea optical fibers. Optical modems routinely have data rates in excess of a billion (1x10⁹) bits per second."

So are you actually pulsing at 2.3GHz (or Gbps or 10⁹ bps), which would put it in competition with the fastest intercontinental optical modems, or is that the RF carrier frequency, with the actual data rate being somewhat lower?

Also, use a parabolic reflector (as on a flashlight) at the transmitter to focus more energy in parallel lines for less divergence. Unlike incandescent bulbs, you won't capture additional energy from the base of the LED, but most LED's have a dome-shaped lens that allows energy to be scattered over 180° or so, side to side, so the reflector will redirect that side energy to the front or top of the LED, provided you can get the LED at or near the focal point of the reflector. Since incandescent bulbs stick up a bit more, this may require you to modify a standard flashlight reflector, which can be done fairly easily by sanding the bottom until it fits down lower over the LED.

Of course, all this assumes an exposed LED, and not one behind a window in an IR transmitter. If this is the case (pardon the pun), you will have to remove the window and/or a section of the transmitter housing to expose the LED. It might also be as simple as drilling/cutting a large hole to allow the reflector to be set down inside the case. I would start with the plastic reflector of a AA Mini-mag flashlight by Mag Instruments. They are relatively inexpensive and fairly accurate, as they are designed to vary the beam by moving the reflector in and out, from flood light to narrow beam.

If that helps, but you need a little more, try a larger reflector from a larger flashlight (C or D battery), which will capture more of the diverging energy and redirect it toward the front of the transmitter (top of the LED).

IR Communication is very good

Do you have such succeed case in this?

Your question provides little to go on. Since you specify an LED we would presume you have a basic design and wish better performance. There are two good ways to improve performance without impacting signal to noise ratio and both have been mentioned. Optical gain, if that is feasible in your application, but not useful if you are using fiber optics. Increase transmitted power. Maybe you need a laser diode. If your data rate is as high as you seem to indicate then you need to choose a form of modulation that is suitable and seriously, it may be that that the answer to your question is beyond the scope of this forum. Get some up to date books on the subject, they aren't cheap, and do some serious research because this is not a common topic. Good luck with your project.

I searched amplitude modulation +led +ir and came up with 38,000 hits.

so some must be already doing something like you want.

http://www.google.ca/search?hl=en&safe=off&q=%22amplitude+modulation%22+%2Bled+%2Bir&btnG=Search

Any question that just says "more" is unanswerable until we know the starting point.
I know of at least two contributors to these fora with solid expertise in this area, so if you state your problem fully you might get some good answers

I've designed a transmitter for indoor use with 96 IR LEDs, thinking that the distance would increase, but I still get 7 metres before going into noise mode. It's a simple circuit based upon a 6W NPN transistor in class C switching on the 96 LEDs. To power up the LEDs I use a power supply with o/p 2A @ 12V. All the LEDs are working but the distance stays the same.

The O/P stage is a matrix of 8 LEDs in series, by 12 rows of them, each row separated by a resistor to supply.

I've read books on pulsing the LED at 4 off to 1 on pulses, and at the same time increasing the current supplied to the LED.

Would you have any schematics?

You can transmit 100 feet with a single LED and a 3 volt lithium battery (but not at 2.3GHz) . Check my article in the June 2005 issue of Circuit Cellar "Short-Range IR Communications System". You can find it online at: www.circuitcellar.com

The design described in the article is for signaling a switch event from a remote that is powered from a lithium coin cell with a battery life of up to 4 years depending on usage. We are all shooting in the dark here anyway since so little has actually been said definitively by the original poster with regard to the application.

I had >200-metres (outdoors in good weather) at 20MB/s with a single LED in 1970. But it was necessary to voltage drive the LED to turn off between pulses. I suspect the components I used then are no longer available. But you should now be able to use current drive up to about 50-MHz with modern LEDs - if you chose ones that are designed for the purpose. You will also need a suitable low-noise receiver, designed for your specific bandwidth.

That should have been "Mb/s" and "1970's" - sorry about the typing.

At 2.3-GHz, you will need to extract charge as well as drive it. In addition, extreme care would be needed to actually get the current into the devices if you drive 96 components in series at this frequency. Are you certain that much of the charge is actually reaching the devices? You might find that you get better results with fewer devices. Even if the potential reaches the pins of each device, I think that you will be fighting a losing battle with most LEDs, because the internal RC time constant can be quite long when the device is forward biased, as the internal charge storage will be in the 10-ns region. The easiest way to increase the range will be to use a lower carrier frequency if this is at all possible - even if it means converting so that you can transmit at some fraction of the input carrier and regenerating the carrier frequency at the receiver. IR-LED operation becomes reasonable below about 100-MHz, and this also has the advantage that you can use a larger-area detector and/or have lower detector noise. You don't give data-rate, but if the system is relatively narrow-band, there will still be advantages from resonating out the capacitances both in the transmitter and in the receiver.

Fyz

As IR is one end of the light spectrum, and if it is only at the LED end you need the range increased, and you cannot change this LED for another, then as with visible light if the radiated angle is reduced then it's range would increase, this may be achieved with a simple lens, as would be done with visible light.

Otherwise if part of a transmitter receiver application, if possible reduction of extraneous interfering light would improve the signal to noise ratio therefore range.

This is of course you cannot change your components.

To quote steve-o's question again, what range do you need? Without knowing this, it will be hard for someone in this forum to advise on a solution, as there seems to be many approaches that would work, to a greater or lesser degree.

Wndrtch

Simple answer use several Led's and some suitable lenses to produce a stronger beam. Pulsing led's above 100Khz is going to be very difficult.

the max practicle range to expect without special optics would be about 1/2 mile. The main limiting factor is aligning the transmitter and receiver, may be a laser could help out then. If it is indoors it will be easier, outdoors involves weather, any water ie: rain mist fog snow will reduce distance. In daylight heat haze can also limit distance. There are too many variables involved to guarantee sucsess.

I think LEDS are useable to far far higher frequencies than that.

http://www.clearmesh.com/downloads/wp_led_121906.pdf

In addition they can go to many many Ghz with external modulation via Kerr shutters or other optical means

Hi aurizon

300MHz* non-super-radiant LED is possible, but only with some loss of efficiency, because the radiant carrier lifetime in these diodes is multi-nanosecond (unless there have been very recent developments that I have missed). But 10's of MHz is economically achievable without loss of efficiency using suitable structures.
*The maximum mentioned on your website reference.

Regarding Kerr cells: LED output is highly multimoded. Running such multimoded Kerr cells at GHz would take massive power, which pretty much removes any advantage that LEDs could have had in the first place

My view of the sensible solution to this part of the problem: lasers (or super-radiant LEDs) if possible (high temperatures would still be an issue here).

Regards

Fyz

well, you would use a polarizer to constrain the led and then the kerr cell (which does not take much power, being voltage sensitive it does not pass current and as a wide plated capacitor does not even pass much AC current, and a kerr cell is capable of very high speed operation and ultra low space if made up as 1 mm cube. You can also use non linear crystals as optical rotators that are also voltage driven.).

In addition, I was making the point that the limit by another of 100 Khz was far too low.

These guys set the limit of LEDs at about 270 MB/s, and give some other background

http://www.fiber-optics.info/articles/LEDs.htm

Mostly, I think, we are agreeing, but not [yet?] about the power requirements of the Kerr cell.

Regarding the LED article: the 270Mb/s already includes deliberate overshoot and undershoot in the drive current, which will absorb a little extra energy. The waveforms shown and the speeds quoted (both including undershoot) seem consistent with a carrier lifetime of about 8 ns, which is about as expected. Voltage drive could still be faster - but requires devices with low series resistance and also special assembly to minimise strays.

Regarding the Kerr cell, what N.A. can it handle? And what would be the capacitance and the half-wave Voltage?

As a sort of background reference, I suspect that nitrobenzene (yuk) would need about 10-kV when operated as a 1-mm cube, and although the capacitance would be rather small (~300 fF?), driving it would take about 15 W/(Mb/s), i.e. 1500-W at 100-Mb/s. Pockels cells using LiNbO3 BaNaNbO3 GaAs etc can be rather better, as you can take advantage of the intrinsic asymmetry, but even here you would be looking at Watts rather than mW if you wish to modulate multimode radiation at high speed.

I don't have easy access to good data on these electro-optic materials, so my numbers could be entirely wrong, in which case I will be happy to be corrected.

Fyz

I am rusty. I used Kerr cells in q switching pumped ruby lasers in grad school in 1964. seemed fast at the time. Rotation is a function of path and I had 1 cm cells that I calibarated and switched with a pulse and the laser went into a crystal double which made UV which was used in boundary layer flow modelling for man made heart plumbing, which had a tendency to start clots where the shear was excessive.

Now I am retired with memories, but there will indeed be a capacitive current, which at high voltage might heat some of the usual kerr rotary organics, like NB. I think there are others. I did not use pockels cells, they may well be lower in power loss than the kerr type?

The power requirement with random data is in the drive circuits rather than loss in the cells. This was not a great problem for mode-locking (the capacitance can largely be tuned out) or Q-switching (single pulses).

In addition, as these were typically single-optical-mode systems, it was possible to use long cells with small cross-sections. The CV^2 dissipation of a Kerr cell is roughly proportional to the area of the cross section - length changes the operating Voltage but not the CV-product. The reason for using long cells was to reduce the operating Voltage so that the device could be operated in air without issues of breakdown, but it also reduced the proportion of the driver power that was wasted in driving stray capacitances. Personally, I was always rather nervous of the carcinogenic reputation of Nitrobenzene, so I was more comfortable when such optical perfection was unnecessary, so KD*P could be used (now known as DKDP, I think - and the material quality is much improved). KDP also has a lower CV^2 product, typically by a factor of about 4 as I remember. Later materials with decent transverse Pockels effect allowed much lower CV products to be obtained, but only at the expense of reducing the number of modes they could handle.

Regards

Fyz

Yes, lots of nasty organics. In those days we bathed in it, figuratively speaking, as many of the bad aspects were not well documented in those days. These cells were sealed with a built in expansion bellows and had no apparent leakage. The cross section was 1 cm and the path was 4 cm, and with single pulses there was no apperent heating of the cell, but if you used these as a 50 Mhz modulator I think the AC current may well cause heating and density gradients with might aggravate you by making losses. Yes, KDP was what we used. After I left Hummel et al went on to be of great value to the artificial heart plumbing business.

transverse Pockels, yes, tech has marched on since my days. I have never used those. We might have used them, but the storeroom had the Kerr cells, and grant parsimony dictated that we use them, we paid enough for our ruby rods, as I recall. $4000 for a 4 inch rod 1 cm thick, ouch. But those were the laser glory days and the provisioners raked it in.