I Built THIS to Find the Best Motor & Prop for My Organic Wing

Ajouté : 2026-05-09

[00:00:00]In the last video, we finished something really special. We use artificial intelligence to search for the optimal shape of a flying wing using a fully parametric geometry and an evolutionary optimization algorithm. After all the simulation, optimization, and design work, we finally built that ultimate organic flying wing and took it to the sky. Honestly, it flew even better than I expected. Stable, efficient, predictable. But even then, one big piece of the puzzle was still missing because no matter how good the wing is, the airplane is only as good as the system pushing it through the air. So, in this video, we'll build a thrust bench from a scratch, measure real motor propeller performance, and answer a question simulations alone can't fully solve. Which propulsion system combination is actually best for this airplane? Stay tuned because the data might reveal something unexpected.

[00:01:00]Whenever I have no clear idea what propulsion system to use, I usually start with a first estimate in ecal.

[00:01:08]It's a really useful online tool because it lets you quickly explore different combinations of motor, propeller, battery, and airframe weight. By testing a few reasonable setups, I can get an initial sense of what should work for this kind of flight I'm aiming for. At this stage, I'm not looking for the final answer, just a solid first guess.

[00:01:30]But of course, tools like these are still only blackbox models. They are great for narrowing things down. For example, selecting a brushless motor, but they cannot tell us the full story with perfect accuracy. That's why some time ago I developed my own tool to choose the optimal setup. So let's pause for a moment because we first need to define the thrust requirement from the physics of the airplane itself. Now imagine the aircraft in a steady level cruise.

[00:01:58]From the detailed aerodynamic analysis we performed in the previous videos, we know there is a specific target flight speed at which the wing reaches its maximum aerodynamic efficiency.

[00:02:11]In this equilibrium condition, the longitudinal force balance becomes very clear. In the vertical direction, lift must exactly balance weight. While in the horizontal direction, thrust must exactly balance drag, which means the drag at maximum aerodynamic efficiency directly gives us the minimum thrust required to sustain flight.

[00:02:35]So now we have a target. The question is no longer how much thrust the airplane needs, but which motor propeller combination can produce it in the most efficient way. Let's first take a look at the motor side. From an analytical point of view, a brushless motor can be described using just three key constants. KB which links voltage to RPM, the internal resistance which represents electrical losses in the windings and the no load current which is the current needed to keep the motor spinning with almost no load. With just these three parameters, we already have a very useful first order model of the motor that described torque, shaft power, and efficiency. All of them as functions of RPM for a given applied voltage. Taking a look at their behavior versus motor speed. Torque decreases almost linearly as RPM increases. Shaft power is low at low RPM, rises to a maximum in the middle, and then falls again at high RPM. And finally, efficiency also peaks somewhere in between after being low at both very low speed and near no load speed. Let's now move to the propeller side. Unlike the motor, the propeller is not driven by electrical quantities but by aerodynamics. Its behavior is usually described through two main dimensionless coefficients. The thrust coefficient and the power coefficient. The key parameter on the propeller side is the advanced ratio. This basically measures the relationship between how fast the aircraft is moving forward and how fast the propeller is spinning. And once those coefficients are known either from measurements, simulations or manufacturer data, we can estimate the propeller's thrust, torque, and efficiency at any flight speed and RPM.

[00:04:26]Just like with the motor, these equations give us a family of curves. As RPM increases, the propeller generally produce more thrust, but it also demands more torque from the motor. And somewhere in between there is usually a sweet spot where the propeller converts power into thrust most efficiently. So now we have both sides of the problem.

[00:04:48]If you remember the motor torque curve, we can plot it together with the propeller torque curve on the same graph. Their intersection naturally defines the equilibrium operating point of the whole propulsion system.

[00:05:01]Everything else follows directly thrust, power, and both propeller and motor efficiency.

[00:05:09]A well-matched system is one where at the equilibrium operating point, the propeller produces the required thrust while both the motor and the propeller operate close to their peak efficiency.

[00:05:20]And that is exactly what my code does.

[00:05:23]The inputs are the motor constants usually estimated from the manufacturer's specifications, the battery voltage, target flight speed, and required thrust. But there is one more requirement that matters a lot.

[00:05:37]Maximum static thrust. Since my airplanes are usually hand launched, they need that initial push to get safely into the air. As a rule of thumb, I aim for a static thrust of at least around 80% of the aircraft weight. Once all those inputs are defined, the code searches through a huge propeller database with more than 200,000 data points to find the setup that best satisfies the mission. The result is a short list of the most promising candidates. On this graph, each point represents a different propeller. Higher means better efficiency and further left means lower power. So, the best options are near the upper left. But these are still just predictions and small differences can change the winner.

[00:06:23]That's why we absolutely need to build this So before building the real thing, let's step into the digital model and take a closer look at how this thrust bench actually works. The most important elements here are the load cells because they are the core of the entire measurement system. Each one is rated for 5 kg of actual force. I use one load cell to measure thrust and the way it works, it's quite direct. The whole upper structure including the motor mount is attached to a carriage that is slides on two linear rails. So the thrust force is transmitted cleanly along a single axis. Proper alignment is crucial here. So having a solid 3D design becomes really important and that makes this a perfect moment to thank the sponsor of this video on shape.

[00:07:21]For a complex and precise project like this, a proper cut model makes all the difference. A lot of you have asked what cut software I normally use and for quite a while now I've been using On Shape. It's basically a cloudnative cut platform which means I can design, iterate, and manage assemblies directly in the browser from pretty much any device. What I really like about it is how easy you can adjust parts, refine mechanisms, and keep the entire assembly organized as the design evolves. Sharing designs, reviewing changes, and collaborating with other team members is also incredibly convenient. So, if you're working on your own engineering projects and need a solid cut tool, definitely check out Own Shape. And if you're an engineer, you can get up to 6 months of the pro version free through my link in the description.

[00:08:12]Okay, going back to our design. To measure torque, I use two additional load cells symmetrically and at the same distance from the motor plane. By combining both readings, I can estimate the reaction torque generated by the motor propeller system. With this architecture, the bench will be able to capture the full picture thrust and torque first and later we'll see how to measure RPM, voltage, current, and power as well. At this point, the mechanical concept is clear. So now it's time to leave the digital model behind and start building the real thing.

[00:08:52]The whole system is built around a rigid frame made from 20x20 aluminum extrusion and off the shelf brackets, giving the bench the stiffness and almost Legoike modularity needed for precise and repeatable measurements.

[00:09:26]I always try to make as much as possible in house. However, for the interface parts where precision really matters, I'm using custom CNC machined 7050 aluminum components made by an specialized company. And honestly, they look amazing.

[00:09:43]The thrust load cell is mounted to the main interface part using two M5 screws, and the whole assembly then slides neatly into the vertical aluminum extrusion profile.

[00:09:57]Then I added two more aluminum profiles which act as the mounting base for two custommade L brackets that also support the linear rails.

[00:10:26]Regarding the torque load cells, they are mounted symmetrically on either side of the motor mount.

[00:10:34]Then a base plate ties the entire assembly together in a stiff configuration.

[00:10:40]This arrangement allows the reaction torque from the motor propeller system to be transferred cleanly into the sensors while the rails keep the whole moving assembly properly aligned. And then slotted holes on the motor mount make it easy to accommodate a wide range of motor sizes.

[00:11:00]I also 3D printed several mounting brackets and added brass threaded inserts. These inserts are important because they give the plastic parts durable metal threads so you can repeatedly assemble and disassemble components without wearing them out. As the brain of the whole bench, I'm using an Arduino Giga Air1. It has a powerful dual core processor, plenty of RAM, and supports high-speed data acquisition, which is essential when reading multiple sensors at the same time.

[00:11:33]This 3D printed bracket holds the three HX711 modules, one for each load cell.

[00:11:39]These modules amplify the very small signals from the load cells and convert them into a digital signal that the Arduino can read accurately. Everything was then carefully connected to the Arduino and I was finally ready to test that each load cell was providing a measurable signal using the Arduino's serial monitor.

[00:12:03]A brushless motor needs an electronic speed controller or ESC to operate. It basically takes the DC power from the battery and convert it into a three-phase signal that drives the motor. And the good thing is that the throttle signal can be generated directly from the Arduino.

[00:12:21]The bench wouldn't be complete without knowing how fast the motor is spinning.

[00:12:25]So to measure RPM, I use an infrared reflective sensor. A small piece of tape on the rotating case changes the reflection once per revolution, creating pulses that the Arduino counts and converts into RPM.

[00:12:40]And finally, this module is used to measure the electrical side of the system. It provides realtime voltage and current readings, allowing me to calculate the input power during each test. Then came a lot of cable management, rooting the wires cleanly, securing everything in place, and making sure nothing could move, vibrate, or get close to the spinning propeller during a test. Which brings us to the most important part, calibration. To make sure the load cells were giving accurate measurements, I used non-calibrated weights and applied them directly to each sensing direction. For the thrust sensor, I hung weights along the thrust axis and recorded the raw signal.

[00:13:23]And for the torque sensors, I hung known weights at a known distance from the rotation axis, creating a known torque.

[00:13:30]By comparing the row readings with the known forces and torqus, I could generate the calibration factors needed to convert sensor data into real physical units. And with the bench complete and calibrated, there was only one thing left to do. Put it to the test.

[00:13:49]And believe me, I did a lot of testing.

[00:13:53]For the motor, I selected five different brands, ranging from a very cheap, lowquality option all the way to the most expensive one I could find. And for the propellers, I used the recommendations from my code as a starting point. I ordered every propeller I could find locally that matched or came close to the suggested diameter and pitch. So, now we have a pretty interesting comparison.

[00:14:16]recommended sizes, different manufacturers, and very different quality levels. All tested under the same conditions.

[00:14:31]I tested every combination of the five motors and eight propellers, repeating each setup twice. For each run, the Arduino increased the throttle in 10% steps. And at every point, it recorded data for 10 seconds, averaging the readings to ensure a stable and consistent measurements. At every step, the system exported the measurements while in real time, I could check that the thrust and torque values made sense.

[00:15:00]I ended up with a huge CSV file containing data for every motor, propeller combination. So now the real question is what does all this data actually tell us? If we look first at pure maximum static thrust, the trend is pretty clear. The larger diameter propellers generally produce more thrust. One combination clearly stands out. The AXI motor with a 10x4.5 propeller gives some of the highest static thrust. But once we add the minimum thrust requirement, several setups are already good enough. So which one can provide that thrust with the least power? This graph compares different propellers on the same motor and the trend again is quite clear. For the same static thrust, the 10-in propellers generally require less electrical power. Now this does not directly translate to forward flight because propeller behavior changes with advanced ratio, but it still gives us a useful picture of which propellers convert electrical power into thrust more efficiently. So for the propellers, we have two clear winners, the 10x4.5 options.

[00:16:07]But what about the motors? To compare them, I plotted the static efficiency in g per watt for the same propeller across all five motors. And this shows quite conclusive results. The T- motor is consistently one of the most efficient motors across the useful thrust range.

[00:16:25]So again, we have a clear winner. And what surprised me most is that the result does not simply follow price. The most expensive option was not automatically the best one. In my original setup, I was already using the same 10-in propeller. So, the main change here is the motor itself. So, assuming this static trend translates reasonably well to forward flight, this motor change alone should mean on paper around a 15% increase in flight time, which is honestly huge. In the next videos, we will get some real flight test data to see if this modification can push the endurance of our flying wing to the very limit. If you want to build your own thrust bench and test even better combinations, you can find the full assembly files on my Patreon.

[00:17:12]And if you want to print your own NX2 and find out the maximum endurance yourself, there will be a 15% discount for the next 10 days using this code.

[00:17:22]And as always, stay curious, stay creative, and keep building the future.

[00:17:27]See you in the next one.