infrared
I want to detect objects crossing a line. I decided to use one IR led (with a 555 circuit to generate a 38khz signal) and one 38khz receiver (TSOP1738). But after testing it on a breadboard I found out that TSOP1738 has gain attenuation, if it receives a 38Khz signal for a "long" period it will turn the internal amp gain down. I want to use 38khz signal to avoid interference with possible ambient IR lights.
How can I solve this problem?
Depending on what type of human proximity you are trying to detect, here are a couple of options that might work well:
Method 1: If you are trying to detect motion, the cheapest way is to use a photodiode or CDS cell connected with a pulldown resistor and read by one of your PIC's ADC pins. You will need to write an algorithm that senses rapidly changing values on the ADC pin, which will indicate some sort of motion. If you want a narrow detection area, you can use a tube over the photodiode as a blinder.
Method 2: If you are trying yo detect proximity by using an IR source and measuring its reflection from an object within proximity, you already have the correct setup. However, you will need to check the datasheet of your IR receiver module for its carrier frequency. I think 30Khz to 40Khz is the most common. You can use the hardware PWM pin on your PIC to constantly produce this carrier frequency on one of the IR LED pins. Connect the other pin of the IR LED to an IO pin on your PIC. Now you can send pulses or even serial data to that pin and it will automatically have the carrier frequency on it. With IR proximity, however, you will always run into the trouble of not being able to detect non-reflective dark objects. In which case, see method 3.
Method 3: Use an ultrasonic transducer for proximity detection. Typically two are used. Use one for pinging a pulse (at the resonant frequency) and use the other for detection. You will most likely need an op-amp or comparator circuit to amplify the output of the receiving transducer for greater ranges. The benefit of this method is being able to calculate exact distance to your object based on the flight time of the "ping" and its return. Flight time will be approximately 1/2 the speed of sound since the "ping" has to travel to the source (1) and back (2).
A micro-controller and associated power source and interface circuitry to hook up to the sensor and the solenoid valve driver would certainly be one way to go with this project that you have outlined. On the other hand you may simply be able to use a time delay relay to control the valve from the sensor directly.
There are various types of time delay relays. The one you want would be one that has provisions for re-triggering its time delay as long as the sensor input is active. After the sensor quiets down the time delay would be allowed to time out and trigger the solenoid. In such setup the solenoid would likely be held in this state until the sensor output became active again.
There is a nice write up regarding the types of things that time delay relays can be used for at this link. The operating mode that may be applicable for your application may be:
Best Answer
If I understand your intention with this project correctly, you want to emit an infrared beam in a continuous manner. Whenever the receiver does not detect the IR-beam you will interpret this as "object detected". Did I get it right?
The IR receiver module TSOP1738 is optimized for reception of a On-Off Key modulated IR signal with 38kHz carrier frequency. This does not mean the 38kHz can be continuously enabled, but only in bursts. After each burst the 38kHz pulses shall be off for a while to prevent the AGC (automatic gain control) in the receiver from saturation. The datasheet recommends each burst to be at least 10 pulses long, but also more pulses may be used per burst. In our case with 38kHz it means 1/38kHz= app 26.3us per pulse, then minimum 10 x 26.3us => minimum 263us/burst. The component test pattern described in the datasheet figure 8 uses app. 23 pulses per burst (600us). Of course this on-off keying also means the IR-receiver output will show a pulse train, rather than a conscious signal. An beam interrupt may be detected when the output has been low longer than the selected burst-cycle time. The decoding of this pulse train could then be done in many different ways: watchdog circuit, timer functions in a microcontroller, Retriggerable Monostable Multivibrator etc.
Some more details about dimensioning of burst timing: Burst-off time shall be equally long as the burst-on time. If we add the burst-on time with the burst-off time we can call that the burst cycle. Depending of the shortest time you expect the beam to be interrupted, you can decide how short your burst cycles can be at the shortest. To be on the safe side (Nyquist–Shannon sampling theorem), make sure to make the burst-cycles shorter than the shortest possible beam interrupt you expect. For example, if you want to be able to detect beam interrupts as short as 1ms, then the bursts cycles must be shorter than 1ms, with <500us burst-on time and <500us burst-off time. I hope this helps.