Theoretical Max Transfer of 38kHz IR

[Newz2000]

I'm trying to get a better understanding of *how* IR remotes transfer data. I understand that it works and have followed the instructions for a test circuit, but I'm curious to understand what's going on. My goal is to understand how fast you can transfer data that way.

Here's how it seems to work to me, please correct me where I'm wrong:

No IR light shining on the receiver, therefore it causes a logic 0
When the IR LED shines on the receiver at regular, 38kHz intervals the receiver realizes that something is trying to communicate with it and makes a logic 1.
Gaps in the 38kHz pulses cause the signal to drop back to a logic 0.

I hope I've got that right. Assuming the above, then before a change from logic high to logic low can occur, a few pulses have to be present or not present.

I know in audio sampling the sample rate has to be at least 2x higher than the highest freq it can sample. I suspect something like that is true here, but it is likely more than 2x even in ideal circumstances.

I was contemplating the feasibility of using IR Transmitter/Receiver as a wireless 2400 or 9600 baud connection. It's so cheap for the hardware and there would be so few components involved... but would serial communication routines work?

By the way, my goal is something stupidly simple, I want to send messages to an LCD display, although if that works, it would be cool to put the display on a non-autonomous robot and control it remotely.

I've seen one way serial communication in some BS projects but I've never used it... it seems like it would be quite limiting to not be able to communicate back to at least acknowledge receipt of the data. I'm guessing that is weakness of IR transmitting.

[Chris Smith]

The easy way to remember a coded signal is that it is like a ticker tape, the info per se is not the whole deal because you are reading both info and at a given speed.

So first the KHZ must be up to speed, [paper speed] and only then within this given speed and paper length, do you punch in a code into the ticker tape for changes or signals.

The whole ticker tape is a code, both made up from speed, length of paper, and the speed at which the tape spews forward with given codes written in the short pieces of measured paper.
Blip blip, blip and each blip has a whole code, and the paper is X in length. No more, no less.

So first the paper comes out at X MPH, then in the given space of time, a individual Morse code is written on this tape length which means the paper is only X inches long per message, and only when both parameters are met does the recognition mechanism start to read and correspond.

[L. Daniel Rosa]

I've looked into this some time ago. It seems that the best of the receiver modules need 6 pulses to lock in the signal. More common ones take 10. There is also some latency with turn on and turn off, which may vary by PN, (IR) signal strength, and manufacturer. Under the best of circumstances, you may hope for 3167bps. I would plan on 300.

[rshayes]

The data rate depends on the signal to noise ratio and the encoding and decoding method. With a high optical signal to noise ratio, such as a diode laser and matching dichroic filter, it might be possible to reduce interference and background noise to the point where an on-off signal was reliably detected. In this case, one bit per pulse would be possible, possibly with the addition of start and stop pulses, and the data rate could be nearly 38 KHz.

In the real world, this optical parts are too complex and expensive to be practical. A keyed carrier at 38 KHz can be isolated from background sources by electronic narrow band filtering. This will require a pulse train to provide a specific frequency. Depending on the filtering and detection method, pulse trains as short as 4 pulses might be possble. The 10 pulses suggested by Dan Rosa might be more practical, since this could probably be detected with filters of moderate Q (around 10) and using a simple envelope detector. This would a data rate of about 3600 bits per second.

Quadrature amplitude modulation (QUAM) might allow data rates several times this if the signal to noise ratio is good and the complex decoder can be tolerated. Actually, QUAM can transmit several bits for each half cycle of the carrier if the signal to noise ratio is high enough, but don't expect this in a simple, cheap system.

Frequency shift keying is another possibility that migh get you to the 5 to 10 KBit range.