I'm designing an interactive table that responds to objects by reflection of infrared light off the objects to a sensor below. It is made from a matrix of around 300 IR LED and sensor pairs, with each pair located at the bottom of a 2 x 2 x 2 inch well. Each sensor can only see the LED when a hand or other object is nearby to reflect the IR light. I want the the sensor's output to give a detection level between 0 and 100% of a nearby object - so I imagine that means it has to be analogue giving say 0 to 5 V on an output. I also don't want other sources of light to interfere with the sensors so I think using modulated (say 38kHz) IR LED and sensor pairs is key. Lastly, because I need so many I would like them to be cost efficient.
Given those criteria can anyone recommend the most suitable IR LEDs and sensors? I've seen a few but some give a digital 1 or 0 only output I think (?) and others aren't modulated. So I was hoping someone could recommend the best options, thanks!
You should look into paired proximity sensors - usually paired. Also consider SMD types for higher density. I know Sharp is a leading supplier.
You didn't mention if you have an Arduino or MCU controlling it & number of available analog input pins!
Others to look into are the Vishay TSOP4P38 , TSSP58P38...with clever design you could get away with a small number of IR emitters and get actual proximity to each individual LED.
see Keyword and Part Number Search Results | Vishay
Proximity sensors cost a lot more; 300 of them would cost thousands. I know Sharp and Vishay make IR receivers but I couldn't figure out how their outputs work from reading their datasheets. I thought I would want a 0 to 5 V analogue output on a single pin but I've seen people using them in voltage dividers and other ways. I'm not sure what is the best technique or how to go about it. I know on some sites they mention some IR recievers have digital 1- or 0-only outputs which wouldn't work for me as I want 0 to 255 levels of 'touch,' for example.
My plan is to take analogue IR sensor outputs and use a DIP microcontroller with multiple analogue inputs to convert that to a 8-bit digital value. I would daisy chain enough of the DIPs for the 300 IR sensors (probably 1 DIP per 10 sensors) and using SPI send all the data to an Arduino or Teensy or similar.
My query here is: does anyone know of an IR receiver with an analogue output I could use in this setup? Or how to use IR receivers to give a varying detection level of, say, a hand over the table. I don't need a distance value.
I assume you've considered a multitouch table? Not necessarily the projector part, but the touch-sensing part alone would not be too expensive.
Vishay TSSP4P38 outputs a varying length pulse based on signal strength (proximity). You could get it at $0.66/ea in quantity 300. It'd still be a pain multiplexing all of those signals though.
Chagrin:
I assume you've considered a multitouch table? Not necessarily the projector part, but the touch-sensing part alone would not be too expensive.
Vishay TSSP4P38 outputs a varying length pulse based on signal strength (proximity). You could get it at $0.66/ea in quantity 300. It'd still be a pain multiplexing all of those signals though.
The problem would be using the camera with my wooden frame/grid of colour pixels I already have made. It's positioned just 2 inches below the glass surface with a matrix of wood separators every 2 inches. Similar to the video in my first post.
I read the datasheet and it says suitable as a 'sensor for large format touch panels' among other uses. Which seems to be quite ideal, right? I couldn't figure out how the output works though. It says the following:
The typical application of the TSSP4P38 is a reflective sensor with analog information contained in its output. Such a sensor is evaluating the time required by the AGC to suppress a quasi continuous signal. The time required to suppress such a signal is longer when the signal is strong than when the signal is weak, resulting in a pulse length corresponding to the distance of an object from the sensor. This kind of analog information can be evaluated by a microcontroller. The absolute amount of reflected light depends much on the environment and is not evaluated. Only sudden changes of the amount of reflected light, and therefore changes in the pulse width, are evaluated using this application.
The first line seems perfect - reflections from hands with analogue output. I don't understand how the pulse length works, however. If I put my hand close it would send a long pulse, and far away a smaller pulse? Wouldn't pulse intensity between say 0 to 5 V make more sense? I'm a little confused... anyone familiar with this?
kadamr:
I couldn't figure out how the output works though. It says the following:
The typical application of the TSSP4P38 is a reflective sensor with analog information contained in its output. Such a sensor is evaluating the time required by the AGC to suppress a quasi continuous signal. The time required to suppress such a signal is longer when the signal is strong than when the signal is weak, resulting in a pulse length corresponding to the distance of an object from the sensor. This kind of analog information can be evaluated by a microcontroller. The absolute amount of reflected light depends much on the environment and is not evaluated. Only sudden changes of the amount of reflected light, and therefore changes in the pulse width, are evaluated using this application.
The first line seems perfect - reflections from hands with analogue output. I don't understand how the pulse length works, however. If I put my hand close it would send a long pulse, and far away a smaller pulse? Wouldn't pulse intensity between say 0 to 5 V make more sense? I'm a little confused... anyone familiar with this?
The output sits at 5V with no reflection, and sets the output low when there is a reflection. The length of time that it sits low represents the distance to the object with longer pulses equating to closer objects.
For a device with a PWM output I would say that this method is fairly common. It's also to your advantage because it doesn't require an analog pin to read it; you just need to do a pulseIn() to time the length of the pulse. Just a guesstimation, but you could probably poll all 300 of the devices about 10 times a second with this scheme.
Also, if it did output a variable, 0 to 5V signal then you'd be facing more issues with noise in your signal lines.
Chagrin:
The output sits at 5V with no reflection, and sets the output low when there is a reflection. The length of time that it sits low represents the distance to the object with longer pulses equating to closer objects.
For a device with a PWM output I would say that this method is fairly common. It's also to your advantage because it doesn't require an analog pin to read it; you just need to do a pulseIn() to time the length of the pulse. Just a guesstimation, but you could probably poll all 300 of the devices about 10 times a second with this scheme.
Thanks for the explanation, it's a lot clearer now. I would say it's not too dissimilar to a reversing car with a beeping proximity sensor. Rather than beeping faster the closer the object, it spends more time during a beep than not beeping - longer beeps (more time at 0V).
So let's say I had just 8 sensors. I could use a CD4051 to sample each sensor using pulseIn(). And hopefully that could be done fast enough for 300 sensors to refresh at up to 30 times a second.
I have another question. Each of these sensors are placed in a 2 x 2 x 2 inch wells below a frosted glass surface. I suppose the IR sensors would detect the glass above it all the time. Hopefully a hand pressed above the glass would increase the reflectivity enough to be detected though, does this seem reasonable?
I will test it for just one sensor of course, I just wondered if anyone had any intuition or experience with this?
The way they work is the closer the sensor gets the lower the number (0-1023 - 1023 being no infrared), using a pair of these side-by-side with a reflective surface could be a way to detect distances.
Check out the product video, he talks about using that method for robotic object detection.
