- Thrust of rotor engine is determined by mechanical power at the rotor and rotor’s diameter, it can be calculated with simple equation. In real life the thrust is significantly lower because of inefficiency of each of the components a) rotor b) motor c) ESC. Total efficiency (figure of merit) is a product of these partial efficiencies. Total efficiency can be estimated by hovering time. For example (measurements of my tiny whoops):
a) Tiny whoop with 0716 (17000) brushed motors looks like this . Total efficiency (FOM) is about 0.14
b) Tiny whoop (UR65) with 0603 (17000) brushless motor looks like this . Total efficiency (FOM) is about 0.10
- Efficiency (FOM) of the motor can be calculated using motor parameters (R, Kv, I0). Current, thrust, torque, rpm, efficiency are mutually interconnected, all possible values form the following curves. When motor is loaded with a propeller, the single point corresponds to this load (example). In particular, this point gives efficiency of the motor. To estimate efficiency of the propeller, rot.efficiency coefficient should be adjusted to get proper values of experimentally measured thrust.
a) For 0716 (17000) brushed motors the plots look similar to these . Efficiency of 0716 brushed motor is about 0.45 (when loaded with 31 mm prop and voltage is 4V) and propeller’s efficiency is 0.36 ( as it was seen before )
b) For 0603 (17000) brushless motor the plots are . Motor efficiency is about 0.7
(this is more or less obvious, because both motors have the same Kv (and therefore torque) and the 0603 brushless motor should be more efficient because of lower resistance of winding)
So, we have that total efficiency of tiny whoop with brushless motors are lower in contradiction with higher values of calculated efficiency of brushless motors.
In contrast to big motors (e.g for 5″ quadcopters) micro motors cannot be described with existing theory, or some of the parameters are not taken into account (like ESC efficiency). This was published also in the report where motor efficiency was shown in dependence on motor’s weight.
Firstly, I would like to know why it happens. Second, it would be nice to predict characteristics of tiny whoops. Third, there are no published data on micro motors parameters except of thrust.
My goal is not just to have some specific parameters of specific motor, but have more general picture of what’s happening and if that can be improved.
That is why I decided to build my own stand. Here is preliminary testing:
Parts are printed with 3D printer. Load cells are 1kg and 0.2 kg for thrust and torque correspondingly. Most of the interface boards are readily available: the hx711 for load cells, for voltage and current at the power source the ina219 board. Unfortunately they are not fast (1ms). This is enough for average values of current and voltage, but not enough for measurements of voltage and current at the motor (where it is good to know also possible phase shifts and waveform shapes and also voltage drop at the MOSFETs). For that purpose I’ve developed my own board based on ACS739 current Hall-sensor and couple of ADA 4522 op amplifiers. Here is the board:
Accuracy of current Hall sensors is not very good, but they are fast (1us) and do not add extra resistance (1mOhm). For better accuracy zero offset can be adjusted before each measurements, because they have some hysteresis.
For rotational speed the laser module was used together with photodiode (unknown origin). Teensy’s comparator is good enough to count pulses even without amplification. The PD is loaded with 6.8k resistor to have a bandwidth >20 kHz.
All sensors are read with Teensy 3.2 at maximal (for the sensor) speed. Each sensor’s output is filtered with low-pass filter with adjustable cut-off frequency (1Hz in the above video) The motor is controlled by MultiWii MSP_SET_MOTOR command to FC via separate COM port.
Teensy 3.2 easily do the job of data collection. Baud rate of Teensy’s virtual serial port is 12 Mbps so it is quite comfortable to exchange data between MCU and PC program.
In an oscilloscope mode current and voltage from the fast sensor are recorded in the internal RAM buffer and later can be obtained by PC program via serial link.
PC program is written with Delphi 10.2.