Wednesday, 7 November 2012

Smart tex again

I apologise for once again using my blog as a placeholder for my notes and planning stuff.

It's just quicker typing it out here than writing it down on paper, and if i saved it as a normal document I'd just lose it.


For the innovation project I will be working from the work I did in last year's innovation project, where I made potentiometers from knitted conductive fabric. I also made knitted pushbuttons. All of which worked, but the potentiometers were a wee bit sketchy. Though, funnily,  they're the part of the project that was picked up by someone else, and was used as the basis of a synthesizer with a knitted textile interface.

Simply a 1D potentiometer, with a single variable. Pressing it against a conductive surface at various points along it's length caused the resistance value of the circuit created to vary, thus making the circuit do different things. It was all analog, my good man Tom is all about the analog.

This year I want to expand on this, by creating a 2 dimensional resistive touchpad. Regular resistive touchapds work by having two conductive sheets which are pressed together by a finger or a stylus or something. Then the resistance is measured from various places to tell where exactly on the pad the contact is being made. It is an established technology with lots of uses and applications and stuffs. I'm going to try doing it with knitted fabric, because I'm super special and that.

So, this should consist of three things, which need to be explained in more detail

  1. the interface, made out of fabric
  2. an analog electronic part, for basic processing of the signal and that
  3. a computer for to take the input and make it do something on the screen

So, the fabric interface works like this. Simple huh? It works fine for push-to-make buttons

The conductive layers will be made up initially of a regular single jersey fabric knitted from that silk steel yarn I have from last year. The non-conductive part will be made of some kind of slightly open knitted fabric made from some kind of fibre that doesn't give me any greif by felting or absorbing and significant amount of moisture. Like some kind of artificial filament, maybe polyester or nylon or something. I'm sure there's nylon to be had. The point is that it has holes in it so by pushing you can cause the conductive layers to contact each other. Otherwise they are held apart.

I think it would help if the whole thing was held reasonably taught. For two reasons. Firstly it'll be easier to use if it's nice and flat and all rumpled up like last night's knickers. Secondly, the natural resistance of the overall cloth varies when it moves if it's slack. We've found that holding it out as tight as can be causes it to be much more stable, which is to be desired.

The analog processing part I don't know about, I may not actually need anything there at all. Oh, hang on, I'll need a pull down or pull up resistor in there somewhere to prevent floating points from occuring. Nobody wants that. Actually, maybe I won't need them, because it's a potentiometer doodah rather than a push-to-make. Hmm, I need to think about this.

The computer is a Raspberry Pi which I have in my house and will take into college. It will take input from various points along the fabric and use this to attempt to determine the point which is being pressed. In theory, in practice I expect it to vary quite a lot from a single point if I use raw data. I'll probably have to throw some computation at the problem in order to reduce the "noise". Not sure if that's an appropriate use of the term noise (like in a signal). It's like this, if there's an overal resistance of 1000 ohms across the cloth, and I short it halfway, then I should get a resistance of 500ohms, which the script on the Pi will interpret accordingly. However, what I would expect would be for the resistance to vary between 200-800 ohms in this case potentially due to what I know about this stuff. I am being wildly pessimistic here btw, I'd probably expect it to be much better.

Aaaaaaanyway, there's probably algorithms out there I can simply cut and paste to tidy up the signal I'm receiving from the cloth and make it more useable. I could probably just take 100 readings in a fraction of a second and average them out.

Computation is cheap these days.

So, right.

 I need to collect the following, to do the following with

To make the fabric, I need

  1. A cone or two of the conductive stuff
  2. A suitable non conductive yarn
To get the program working I need

  1. A tutorial on using the GPIO pins to read analog values
  2. Instructions, libraries or code for cleaning up scratchy signals
  3. Perhaps a cut and paste program I can just rip that's made for prototyping touchpads. There must be something out there

The yarn is easy, I'm going to try getting the rest today. I will post links to whatever I find in the comments

No comments: