Several years ago I purchased a small video camera that was shaped like a flashlight for $50.  It had VGA 640 x 480 30 fps video quality and would record up to 2 hours on a single charge.  I used it to record several bike rides around the county on numerous bike trails.  The videos provided a way to recall the conditions and experiences which I used to plan future trips.

A couple of years later I purchased some 808 key chain cameras.  I liked the small form factor.  They fit nicely on my helmet and on the handlebars of my bike.  The video quality was still the same.  Around this same time was the release of the 808-16 camera.  This had much better video quality with 720p at 30 fps.  This quality and size difference made my earlier flashlight camera obsolete.  However, it suffered a setback due to its limited battery run time of 30 minutes.  I was able to offset this with the use of an external 2ah battery pack.  This became my first generation rig and it set the standard for rigs to follow.

Gathering as much of my surroundings was a goal early on with the 808-16.  I used a fish eye lens that was originally targeted for smart phones.  With some hot glue, I managed to attach these lenses to the 808-16 housing.  The resulting video had noticeable distortions and dead space that I attempted to minimize with some adjusting.  Not long afterward, replacement 808-16 sensors with fish eye lenses became available.  This reduced the bulk and improved the quality of the video.

In addition to new sensors available on the market, there were sensor extension cables.  These would connect between the camera housing and the lens sensor unit.  This meant that the camera could be placed more discreetly on the helmet.  The audio of the top mount rigs was always horrible while traveling because the wind would wash out any background sounds.  Now it was possible to capture the audio by placing the camera housing under the visor and keeping the lens sensor mounted above.

One of the biggest challenges with all of these rigs was weather.  Rain and moisture are a big concern that I was able to mitigate using a number of methods.  The first and second generation rigs only had issues with rain water obscuring the lenses.  I attempted to fix this by 3d printing a housing, but the added bulk wasn’t a good trade off, so that was scrapped.  As a result I didn’t pursue extending the lens sensor at this time.

After a season of operating these rigs, the next generation of cameras became available.  These were another leap in quality with 1080p 30fps video.  The Mobius cameras had astonishing video quality compared to the 480p camera I had first started out with.  As with the 720p rigs, these also offered fish eye lens built with the sensor as well as extension cabling for out of body mounting.  It was at this point that I decided to integrate the camera inside the bicycle helmet.

Digging trenches through the helmet was done using exacto blades and a small steel pick.  I was able to etch out a cavity to hold the camera circuit board using this same method.  The lens and sensor were placed in the front of the helmet.  I used a drill to make the opening for lens to fit through.  Once the sensor, wiring, and circuit board were in place, I used hot glue and Sugru to hold and protect the items from moisture.  The power for the camera was supplied solely by an external rechargeable unit that mounted on the back of the helmet, similar to my earlier rigs.

In addition to installing the 1080p inside the helmet, I decided to install a modified 720p unit.  This unit had the sensor near IR filter removed and in its place a visible light filter.  The light filter was nothing more than a cut piece of floppy disk media.  Having both a visible and near IR camera on the same rig allowed me to make comparison observations of rides.  The last challenge to this setup proved to be too much.

I knew heat was an issue for the Mobius.  The problems of overheating were well documented on the RC forums and other various blogs.  My attempt to keep things cool involved tacking a copper pad on the chips using thermal adhesive.  My bench tests worked fine and everything seemed to be fine.  It wasn’t until I actually wore the helmet that problems began.

The one thing I noticed early on was that the camera kept shutting off.  At first I thought that my power supply was the issue.  It wasn’t until I could not restore power to the camera that I began to realize it had been a heat issue.  This was confirmed when I viewed some of my video.  Moments before the camera would shut off, the color in the video began to distort.  It would suddenly loose all color or just become a magenta tinted image.

How could the bench test and field test have such different results.  The answer was the heat from my skull was countering the cooling effects of the copper heat sink on the chip.  Worse yet, it was a thermal conductor that allowed my heat to pass into the chip.  This nailed the coffin shut on this rig.  This video gives you an idea of how much heat dissipates off of human skulls.

The Icarus effect had been reached.  This was the end of any further development.  I had ideas of placing a plethora of sensors and controllers in and on the helmet.  There would be dust, smoke, gas, and wind speed sensors.  I had intended to integrate a GPS module and a microSD card logger.  I had also considered placing a ESP8266 module to do war driving.  Of course these were just ideas, but when the camera failed due to heat, these ideas went to the funeral instead.

Those ideas were in part responsible for the camera failure.  The project snowballed with one idea after the next until it became the impossible dream.  That became the eulogy.

This is a cautionary tale for any project development.  Keep it simple, make it easy, start out humble, refine over time, don’t rush it, these are all terms we have heard before but they apply still.

What does this mean for the bicycle helmet and embedded electronics?  It really isn’t the end, just a new beginning.  Most likely the helmet and electronics will not be merged to the same degree they had been.  Instead the helmet will remain unchanged.  Any attempt to quantify will be designed and built so it can be added or removed from the helmet.  The electronics should operate independently from the helmet.  They should be designed in a way that they can be operated without the constraint of placement.  It should be possible to place them in a back pack, in a bike bag, or on the handlebars.  But the most important part is they should be able to return to the bench for further testing and refining.

More on the Mobius 2