I was contacted by a friend on
behalf of a company that sells performance parts for ATVs and dirt bikes about a
product that they had hoped I could design for them. Always up for an
interesting design challenge, I happily agreed.
The task at hand
My goal was to build a device that would prevent water from
entering the running engine of an ATV in the event that the vehicle was flipped
over or submerged in deep water or mud. If water is allowed to be drawn into
the running engine through the intake manifold a hydrolock condition will
result and the engine will likely be destroyed. To prevent a hydrolock, I would
need to detect the presents of water in the intake manifold and block it's path
to the combustion chamber. I would also need a way to shut down the engine when
water is detected. To accomplish this, I decided to use a microcontroller to
read data from a sensor and then actuate a servo. When the microcontroller detects water in the system, it sends a signal to the servo to close the valve. While the valve is closing, another signal is sent to a relay that grounds the ignition module of the engine causing it to stop. These simple methods meant
that I could use mostly off the shelf components for the electronics and 3D print the main housing. I did, however have to build a custom control board to mount the microcontroller and the passive components to. Below is a few images and a video of the control board being milled in my CNC.
+
Performance
Ideally, you would like a device of this nature not to interfere with the natural operation of the engine. More specifically, it can't restrict or disrupt the path of fresh air that the engine needs to be able to run. The easier an engine breathes, the more efficient it operates. I setup a simple air flow simulation in Solidworks to see if there would be any noticeable disturbances in the way that air flowed through the device and the results are promising. As you can see from the video, the air moves at a fairly constant speed without any turbulence or vortices.
Unforeseen Issues
In testing the prototype, I found that an unacceptably large
force was applied to the servo when the valve was closing. The vacuum being
drawn from the engine wanted to close the valve door faster than the servo
could complete it's stroke. To address this, I designed a spring loaded mechanism
into the valve door linkage. This allowed
the door to close rapidly and the servo to catch up afterword.
Spring loaded linkage assembly |
For detecting the presents of water, I designed a type of electrical
continuity sensor. An electrical connection was formed between two metal screens
when water flowed between them. This had the advantage of being able to tune
the devices sensitivity; I didn't want it to activate over just a few drops of
water, only when it was being flooded. This had a drawback in that it also
picked up stray RF interference, one could see the spikes from looking at the
output data. Sometimes when you put your hand near the sensor, it would pick up
the electrical signal from your body and activate the valve. I'll admit, this
was kind of amusing at first, you would reach to pick up the device and just
before your hand got close enough to touch it, suddenly it would activate and
startle you. I decided to fix this in software by implementing a smoothing algorithm
that compared readings and averaged samples.
Conclusion
Overall, the device works great. It satisfies all of the design requirements including detection, activation and engine stop. The only area that I feel could use some more work would be the valve door seal. Currently, a section of rubber surgical tubing is used as a sealing gasket between the valve door and the valve seat. The tubing is glued around the inside perimeter of the door and compresses and conforms to the valve seat when the door is closed. An potential issue is the adhesive bond between the ABS plastic door and the rubber seal. These two materials require a type of adhesive that is not locally available. The only adhesive that I have had any success with is Superglue (cyanoacrylate) but I have doubts about it's long term durability. The best solution would be an extruded rubber seal that has had it's ends joined together but that would not be practical for a prototype.