Underwater IR Webcam

So Connor and I set out to waterproof a webcam the other day...

We decided to just coat the whole thing in a thick layer of silicone, this basically is just a test for a future project we plan on doing.

It worked out fairly well, though no long term underwater testing has been done yet...



Here, our beautiful workspace:

After coating it with silicone we got this wierd artifact (the semi-circle in the top right corner). If anyone could shed any light on why this appeard, I'd love to hear it:

The "finished" IR underwater Webcam:

Flex Sensors by Flexpoint

I have just been playing around with some sensors by Flexpoint. And I have to say, that I really love them...

While the company appears to focus on custom design, they produce standard sensors in three sizes, 3", 2" and 1". The actual variable resistor within the sensor is a bit shorter. Check the image for approximate actual measures:

Other flex sensors I had previously worked with where quite noisy and corse. Also they where usually pressure sensitive as well which in some cases can confuse readings. Another thing which I was sort of afraid of was that they had some type of memory, similar to the conductive polymere I have been working with.

I started playing with the three inch sensor. This is what I figured out:

• The signal is as clean as I would expect from a high quality potentiometer
• Using a voltage divider by Phidgets and an Arduino Duemillenove I was able to get a resolution of 800 points
• There is a slight memory effect. If the sensor is bent very strongly it does not go back to baseline
• The sensor is not strictly biderectional as about 80% of the possible readings are produced by flexing it in one direction. However there is enough resolution to do coarse bidirectional readings.

I decided to figure out the sensitivity at different levels of flexion (is that a word?) and came up with a chart that I wanted to test:

(U --> Umfang --> Circumference)

Basically I am using the sensors length as a measure and am wrapping it around an object 4 times, 3 times, 2 times and identical to the sensors length. That corresponds to distances of ∞,  2.26,  2.1, 0.777 & 0 inches (I dident do 1,33)

I used a phidgets voltage divider set to 27.4k and the analogue in of my Arduino for these measurements. These are the values I got:

• Baseline ~ 725 when first connected,  710 after beeing bent strongly and released again
• 4x -- ~ 550
• 3x -- ~ 450
• 2x -- ~ 290
• Circle ~ 80

I also drew a linear scale, moving the two ends closer together by half an inch at a time:

• 3"    --  ~ 710
• 2.5" --  ~ 420
• 2"    --  ~ 280
• 1.5" --  ~ 180
• 1"    --  ~ 120
• 0.5" --  ~  80

While there is a slight memory effect (i.e. my reasings are lower after having flexed it once) the readings where constant and repeatable after that.

I measured it by marking the distance I want to bend (or flex) to and holding one end tight moving the other end to that mark:

The sensor is glued to a strip of plastic for a more rigid setup. This is to minimize rotation or uneven flexing of the sensor. Which brings me to the main drawback of the 3" sensor: As it is so long, the flexing can happen at different places and it is also prone to rotate, which severely influences the readings. For very precise sensing, it may be of benefit to use shorter sensors.

Here are some images of the 3" sensor in action:

 

The 2" sensor seems easyer to work with than the 3" sensor, as it is less prone to multiple bends or rotations which give readings which are hard to interpret. My favorite, was the 1" sensor. To be honest that one really surprised me. I got the largest range of readings from it, and becouse of its small size, the readings are very clear (its almost inpossible to bend it in two different ways at the same time, and it is really easy to avoid rotation.) here are some images of the 1" sensor in action:

Low-Tech Sensors

My friend Connor introduced me to Active Surplus in Toronto. Needless to say, my mind was blown by that crazy store. Except for the fact that its not the cheapest place to buy stuff, its every hackers wet dream :-)

Anyway. I found these really neat tilt switches and wanted to build something cool with them.

So I did :-D

Take a look:

I love the design of these... its so simple and pretty :-)

RaibowCoat - ShadowCoat with some Color :-)

Take a look :-D



And some stills:

And my arduino and breadbord, which powers all of this:

ShadowCoat v 0.02

Before I start rambling, take a look at this video. It’s a demonstration of how the prototype I built can be used as an interface device. I am here demonstrating how it could be used as a mouse. It’s sort of cool, though it sort of takes something really cool and turns it into something banal. While this works, and I know I can get it to work a lot better still, I am hoping to find more intuitive and natural ways of interfacing with computers using this type of technology. (ah, I have already started rambling. Whatever. Watch the video-clip.)(and hear some more rambling)

While my original demo worked quite well, it didn’t have any cable management, so you can spend quite some time figuring out which cable leads where etc. Also the material was quite flimsy and I mounted the Velcro upside down which isn’t helping the material either... also, the pressure fitting was sort of annoying etc. etc. Lots of little things which could be improved.

So on my week off of university, I decided to build a new prototype. Things I wanted to include/change where

- Feedback
- Cable Management
- Tilt of wrist
- Smarter positioning of Velcro
- Sturdier material 


Halfway through my work my sewing machine died on me. Well. It didn’t really die. It just sort of protested against sewing cables and electrical equipment to clothing. Guess this isn’t the future it envisioned. But it still is pretty, have a look:

Also I realized that by sheer luck I did a really good job on my first prototype. I had to figure out, often by trial and error, lots of little things which I, by pure coincidence, got right the first time I did it.

But I prevailed :-D. And while I got about half of what I wanted done, it’s a good start and I can definitely build on it, it is much more expandable/upgradeable than my first attempt.

The white thing you se protruding from my arm is an LED strip, so I can have color feedback to my movements.

One of the things which really baffled me was the voltage divider. First time round it just worked, so I didn’t spend much thought or time on it. This time I was not so lucky. So, here is how it should be done:

It is my experience, that the resolution is best when the resistance of R2 is as close as possible to the nominal resistance of the corresponding sensor. I think it might help this project to include opamps for better resolution. I hope to be testing that soon.

I also rewrote all of my code. For the first time in my life I found myself actually appreciating how powerful object oriented programming is. It improves the simplicity of my code drastically, makes it much easier to add sensors, to scale and map them and to just generally have some useful signal flow.

I also figured out that timing is really crucial. I have never really worried about how long it takes to execute a function, but I realized that this is the reason to 90% of errors and bugs I had in my programming, as it can go haywire when the timing between serial and java as well as the communication between objects is out of sync.

Anyway, for the first time I have some code which actually works well enough, that I feel like sharing it. I will figure out the proper format of doing so as soon as I get around to it. In the meantime feel free to contact me at paul dot strohmeier at gmail.com if you would like a look at it.

*


OK, some more random pictures, just because I took them :-D

 

4 Voltage dividers (schematic of individual one can be found in a pic further up)

 

Stretch Sensor material from Images

I did this at my grandparents place. This is my gradfather workshop, where I did some soldering.

It’s hard to see, but in case you are interested in how I connected the sensor material with the wired, this sort of gives a clue. I basically just wrap the wire ends around the sensor. If I had a working sewing machine I would pin this to the canvas using criss-cross stitches. As my sewing machine preferred not to support this line of work, I just put electrical tape around it, before sewing it to the shirt by hand.

Anyway - that’s it for now. I really hope I will have time to incorporate the leds and other stuff, but I am afraid it might take a while before I find time to continue work again...