The Propulsor Zero test stand is now ready for showtime!
Like most of my projects, it started with a jaunt to All Electronics in Van Nuys.
Using that equipment, I finished up the Vibration module – an Adafruit MMA8451 accelerometer. It does a root sum of squares over time of the root sum of squares of all the accelerations in all axes. This sends back an “index of vibration” which technically has units m/s². The reason for this craziness of mathematical complication is so that a high frequency, low amplitude oscillation will produce a similar index of vibration to a low frequency, high amplitude oscillation (they are equally bad).
The mounting of this sensor was accomplished by some nice mounting tape on the bottom of the motor. The microcontroller unit (MCU) is the beautiful Adafruit Metro Mini 5V.
Then comes the AC current sensor.
The AC current sensor uses the Sparkfun 30A Inline Non-Invasive Current Sensor .
The signal conditioning circuit is the same as designated on this page. I drew a circuit schematic using the proper symbols but neglected to photograph it. Will update later.
The code is very simple and uses an Arduino library for this sensor, called “emonlib” (energy monitoring library). Again, runs with the astonishingly useful Metro Mini from Adafruit.
The split-core current transformer (“CT” as some call it) is clamped on the HOT wire of the motor. This sensor creates a magnetic field, which is modified with the passage of electric current though the wire. That modification makes an amplitude in the signal which is linear (proportional) to current, and is exactly the modification we can measure, to determine the AC RMS current.
Next came the fun task of basic wire management. Zip ties galore on this thing. Had to move the assembly inside due to rain forecast.
Next comes the LOAD CELLS. A 0.477kg (4.67N) can of Chilli will be the weight calibration device. Yum….!
After getting the mass of the can, and then placing the can on the 100kg load cell, I took measurements of the data with & without the can. Very basic math lets me see the amount of Newtons of force instead of the data straight from the sensors.
1. I measure the sensor value, minus tare value, called the “tare value”.
2. I measure the sensor value, minus tare value, with a weight, called the “loaded value”.
3. I measure the weight’s actual weight, call it the “actual loaded value”
Then for any instant, the actual load on the sensor is…
Output Load = [ (actual loaded value)/(loaded value – tare value) ]*[instantaneous sensor value]
The chips being used are the HX711 ‘s. I had previous experience with them and their spec sheet fit the bill, plus they’re super cheap, so why not? An Arduino Mega 2560’s analog pins are being completely used up by all 8 load cell modules. Quite glad that worked out!
On the software side, I’m using the great HX711 library.
There are some issues regarding the hysteresis of the S-type load cells. This is where, when a load applies itself to the cells, and then is relaxed, the value the cells retain is different from what it was prior to the load being applied, and the path the signal takes down is different than the one it takes up. It is not a closed loop response, in other words. I believe this can be compensated for in software, but I will look into that only if it becomes a problem later on.
Solid wires switched out for stranded Duponts:
Pretty rainbow colors!
All of this sensor data gets shot into a Python script which aggregates the data, and logs it to a file. Eventually it will have a GUI as well.
Now comes the difficult job of actually manufacturing a few accurate shape propulsor blades. The fiberglassed 3D prints came out with about 25% success (1 out of the 4 was acceptable). I may add more plies, or go with another approach. Any ideas out there?
I have some crazy ideas, and less crazy ideas in works.
Less crazy includes buying a desktop CNC mill, configuring it for 3D shapes as much as possible, and then manufacturing the blades with wood or Aluminum.
More crazy includes a brand new (I think?) method, I’m calling it “force forming”. It involves building a rig which applies the requisite forces (both axial and centrifugal) to a deformable material… The idea is that the material will form into the most efficient / optimum shape just by the physics itself. Essentially, the “usual way” of designing a propeller is to build a prototype, then examine how the INPUT (propeller shape) creates an OUTPUT as a result of the PHYSICS. This force forming method would be basically, replicating the desired OUTPUT, applying the PHYSICS, and therefore shaping the INPUT (propulsor shape).
Stay tuned to see which of these works out!