Variable Pitch Propeller Attempt

Inq

Elite member
Edit: This tread originally started out asking about a request for centrifugal weight based variable speed propellers. After some initial CAD work, I realized swinging weights, and 3D printed ball bearings were just not going to fit inside the spinner of the desired 1/12th scale plane. As the thread progresses, I go back to the more complex, but smaller spinner packaging of a computer controlled servo pitch design.

I've used various Google searches including the title that don't seem to find the hits I'm looking for. Invariably I get millions of hits for constant speed propellers using the hydraulic line through the drive shaft that came about around WW2 and is still used today. I'm looking for details on a propeller design that was used I believe around the 1930's. It used weights at the end of offset arms to twist the propeller blades as a function of rpm - as the rpm's climb, the centrifugal force would spin the weights out causing the blade to increase pitch, causing more drag and thus slowing down the rpms. Tuned just right, it would maintain a relatively constant RPM. Here is a picture of one on a Stearman...
Super-Stearman-PW-450-HP-top-view.jpeg


I've taken a stab at the problem using first principles, but I'd really like to get a reality check from existing designs. I'm looking for anything that might be of a more "how to design" type article.

Background - I looked at the Constant Speed Propeller design from an electronic viewpoint - (1) getting either mechanical linkage through the turning shaft (swashplate, helicopter design) or (2) putting the servo in the spinner (power and control data to the servo through the spinning shaft). I went so far as to even look at wireless charging to power a Bluetooth/MPU unit to control/power the servo. It spiraled down to China. It just doesn't seem feasible at the 1/12th scale I'm attempting. So I've turned to 1930's tech to tackle the problem.

Any/all references, pointers, or comments are desired.

Thanks,
Inq
 
Last edited:

L Edge

Master member
A number of years ago, I explored the idea of 4D.
vp.jpg


It amounted to installing a motor with a hollow shaft where one end was hooked to a servo, ran thru the motor, where the metal segment (above pic) was attached to the motor, and the shaft allowed the movement to rotate the blades.

Planes had to be super light to allow blades to fully reverse in order to perform 4D. Bought 4 units of cheap plastic (expensive then) and started to have fun. With vibrations(even balanced unit) it didn't take much for blades to ripped out of the hub and ball joints fail.
The stress generated by blade changes did not hold up to the design of cheap plastic. Even now, I bet the few pilots that do 4D find the planes have short lives.

Wish you luck!!!

 

Inq

Elite member
A number of years ago, I explored the idea of 4D.
View attachment 233631

It amounted to installing a motor with a hollow shaft where one end was hooked to a servo, ran thru the motor, where the metal segment (above pic) was attached to the motor, and the shaft allowed the movement to rotate the blades.

Planes had to be super light to allow blades to fully reverse in order to perform 4D. Bought 4 units of cheap plastic (expensive then) and started to have fun. With vibrations(even balanced unit) it didn't take much for blades to ripped out of the hub and ball joints fail.
The stress generated by blade changes did not hold up to the design of cheap plastic. Even now, I bet the few pilots that do 4D find the planes have short lives.

Wish you luck!!!


HOW COOL IS THAT?! :cool: I've never heard of 4D before. Definitely a couple of orders of magnitude above my skill level.

Your prop above... very nice, simple design. But... you say its possible to get electric motors with hollow shafts. I'll have to look for those.

Thanks.
 

L Edge

Master member
HOW COOL IS THAT?! :cool: I've never heard of 4D before. Definitely a couple of orders of magnitude above my skill level.

Your prop above... very nice, simple design. But... you say its possible to get electric motors with hollow shafts. I'll have to look for those.

Thanks.

E-flite - EFLM1210HS Horizon Hobby has hollow shaft motor.
 

telnar1236

Elite member
I've used various Google searches including the title that don't seem to find the hits I'm looking for. Invariably I get millions of hits for constant speed propellers using the hydraulic line through the drive shaft that came about around WW2 and is still used today. I'm looking for details on a propeller design that was used I believe around the 1930's. It used weights at the end of offset arms to twist the propeller blades as a function of rpm - as the rpm's climb, the centrifugal force would spin the weights out causing the blade to increase pitch, causing more drag and thus slowing down the rpms. Tuned just right, it would maintain a relatively constant RPM. Here is a picture of one on a Stearman...
Super-Stearman-PW-450-HP-top-view.jpeg


I've taken a stab at the problem using first principles, but I'd really like to get a reality check from existing designs. I'm looking for anything that might be of a more "how to design" type article.

Background - I looked at the Constant Speed Propeller design from an electronic viewpoint - (1) getting either mechanical linkage through the turning shaft (swashplate, helicopter design) or (2) putting the servo in the spinner (power and control data to the servo through the spinning shaft). I went so far as to even look at wireless charging to power a Bluetooth/MPU unit to control/power the servo. It spiraled down to China. It just doesn't seem feasible at the 1/12th scale I'm attempting. So I've turned to 1930's tech to tackle the problem.

Any/all references, pointers, or comments are desired.

Thanks,
Inq
Looks like a cool concept. One issue I would expect is that IC engines and brushless motors behave very differently in response to load. An IC engine throttle controls the amount of air/fuel, and thus the power output (this is not actually a constant power curve since the power also scales with RPM and due to weird internal dynamics). Therefore, the RPM is variable at a given throttle setting. For a brushless motor, the throttle attempts to set the RPM, with the ESC attempting to provide enough current to hit that RPM, meaning that if you went with this concept, I would expect it to result in a prop that just increased in pitch with throttle and drew excessive current. If you want to pursue this concept, I would definitely recommend a brushed motor, since the ESCs for those do not attempt to match RPM. Thus, the ideal brushed motor power curve is constant for a given throttle setting which would seem to match pretty well with you goals here.

In terms of the prop, not something I'm familiar with, but I would imagine balancing it to avoid excessive vibrations would be a huge pain.
 
  • Like
Reactions: Inq

Inq

Elite member
Looks like a cool concept. One issue I would expect is that IC engines and brushless motors behave very differently in response to load. An IC engine throttle controls the amount of air/fuel, and thus the power output (this is not actually a constant power curve since the power also scales with RPM and due to weird internal dynamics). Therefore, the RPM is variable at a given throttle setting. For a brushless motor, the throttle attempts to set the RPM, with the ESC attempting to provide enough current to hit that RPM, meaning that if you went with this concept, I would expect it to result in a prop that just increased in pitch with throttle and drew excessive current. If you want to pursue this concept, I would definitely recommend a brushed motor, since the ESCs for those do not attempt to match RPM. Thus, the ideal brushed motor power curve is constant for a given throttle setting which would seem to match pretty well with you goals here.

In terms of the prop, not something I'm familiar with, but I would imagine balancing it to avoid excessive vibrations would be a huge pain.

Although this is a totally embryotic mental exercise at the moment, it stems from the desire to eventually make WWII warbirds with scale propellers. For instance a 1/12th scale plane I like will have a 12" diameter 3 blade propeller that I'll never find off-the-shelf. This... on a plane with only a 38" wingspan. To really optimize it for take-off and buzzing the field, it needs two drastically different pitch propellers... thus... this thread.

I also play with and haunt a microcontroller/robotics forum. I'm not beyond using throttle position as an input to an on-board microcontroller representing a desired current setting and have the microcontroller monitor battery draw and rpm and let the microcontroller set both ESC setting and propeller pitch setting.

Just for S&G, I've CAD'd up a first cut at the blade. At the moment, it uses a Clark-Y along the entire length and has a 40º twist. I've been researching some better foils using multiple shapes along the span. I'll optimize the propeller to eek out for top-end speed and rely on the pitch control to handle the mundane duties like... take-off, landing, accelerating and climbing. OR... I might throw a blade or two in the house and get shut-down by the Admiral.

Can anyone guess what plane this comes off? This blade is 125 mm long and with a 50 mm spinner, it'll give a 300 mm diameter / 3 blade prop for a 1/12 scale example. Disclaimer - This is not a working prototype. It is merely a geometry check using vase mode.
Blade.jpg
 

telnar1236

Elite member
Although this is a totally embryotic mental exercise at the moment, it stems from the desire to eventually make WWII warbirds with scale propellers. For instance a 1/12th scale plane I like will have a 12" diameter 3 blade propeller that I'll never find off-the-shelf. This... on a plane with only a 38" wingspan. To really optimize it for take-off and buzzing the field, it needs two drastically different pitch propellers... thus... this thread.

I also play with and haunt a microcontroller/robotics forum. I'm not beyond using throttle position as an input to an on-board microcontroller representing a desired current setting and have the microcontroller monitor battery draw and rpm and let the microcontroller set both ESC setting and propeller pitch setting.

Just for S&G, I've CAD'd up a first cut at the blade. At the moment, it uses a Clark-Y along the entire length and has a 40º twist. I've been researching some better foils using multiple shapes along the span. I'll optimize the propeller to eek out for top-end speed and rely on the pitch control to handle the mundane duties like... take-off, landing, accelerating and climbing. OR... I might throw a blade or two in the house and get shut-down by the Admiral.

Can anyone guess what plane this comes off? This blade is 125 mm long and with a 50 mm spinner, it'll give a 300 mm diameter / 3 blade prop for a 1/12 scale example. Disclaimer - This is not a working prototype. It is merely a geometry check using vase mode.
View attachment 233635
The idea with the microcontroller could work. You just wouldn't truly have a constant speed propeller if you did that since the ESC only targets speed. But if your goal is just to have a scale-looking prop, it could potentially work. Honestly, the easiest approach might be just controlling the prop pitch with a servo driving something running through the shaft like L Edge was showing.

That's a cool looking blade. Maybe from a spitfire or BF 109? Yeah, the Clark Y airfoil is way too thick for a prop, but it looks like a high quality print. Do you think you could get the orientation of the layer lines different? Right now, I'd be worried about it cracking along them and launching a blade at high speed. Please be safe when testing these.

Edit: thinking about it, this type of prop would never truly be constant speed. The microcontroller idea is a great one and could definitely work
 

Inq

Elite member
The idea with the microcontroller could work. You just wouldn't truly have a constant speed propeller if you did that since the ESC only targets speed. But if your goal is just to have a scale-looking prop, it could potentially work. Honestly, the easiest approach might be just controlling the prop pitch with a servo driving something running through the shaft like L Edge was showing.

You are totally right (y)... part of the morning before my last post, I was still trying to make a centrifugal design work. First, the blades have to have near frictionless movement even under 10K+ rpm and heavy thrust. Otherwise, balancing the centrifugal forces working one direction and aerodynamic foil pitching working the other can't hope to control the blade pitch accurately if friction is involved. The only way to achieve that was with some kind of ball bearings around each blades shaft. Second, even assuming frictionless blade pitch movement, I feel some kind of fine tuning would be required. Either, some kind of spring mechanism or some adjustment of the weight's location. Anyway, there was just no way for me to cram all that within a 2" diameter spinner. Point being even though I had started re-evaluating with @L Edge design suggestions in mind, that didn't come out in what I wrote.

I now thing the only way to squeeze it in a 2" diameter spinner is to have the control servo (and wiring) outside the spinner working some mechanism inside the spinner. @L Edge suggestion is a working solution. A servo should be able to handle a little friction and would be controlled by the micro-computer.

Edit: thinking about it, this type of prop would never truly be constant speed. The microcontroller idea is a great one and could definitely work

I've given what you said some thought and you're right. (y) I don't guess we really should care about it being constant speed. We're not concerned with over-revving an ICE engine. Instead of putting some kind of LED/photocell to measure rpm, the wiser idea would be to monitor electrical current. Say... if the motor/ESC can handle 30 amps, the on-board micro would measure the current and adjust the ESC and pitch until the desired amount of current is achieved (throttle sets max current from 0 to 30 amps). Got to do some head scratching to map which gets adjusted first... more ESC "throttle" or more aggressive blade pitch. But it sounds like a nice juicy science project.

That's a cool looking blade. Maybe from a spitfire or BF 109?

Although, my favorite plane is the Spitfire (late IX) I can't see ever getting good enough to land a truly scale model. Just way too easy to ground loop a plane I would spend hundreds of hours making. :oops: So, my second choice... Focke Wulf 190 D9 or Ta 152C.
uu652Zm.jpg


It would be frustrating for me to go to a huge amount of trouble to make a scale plane with retracts, flaps, wheel doors, even panel seems, war-damage, wear and tear painting and it is sporting a scrawny, 9" two bladed propeller. Even a proper size 300mm (~12") off-the-shelf propeller would be glaring on this plane. Not to mention any motor capable of turning such a prop would be way overkill for a plane with a 38" wingspan. And that's... before adding this big honk'n paddle prop!

I plan to try this propeller using a D3536 1000KV motor. It's only rated to handle about a two bladed 10x6. I imagine using a proposed three bladed paddle prop, at full throttle static and taking off, the pitch might be down around 12x2... but it'd pull stumps and accelerate like a demon. At full speed, it might be pulling a 12x10 or better. I just think it'd be a really cool project and look fantastic as a static display as well.

Yeah, the Clark Y airfoil is way too thick for a prop, but it looks like a high quality print. Do you think you could get the orientation of the layer lines different? Right now, I'd be worried about it cracking along them and launching a blade at high speed. Please be safe when testing these.

As I said in the disclaimer... it was merely to put my hands on it... not to usable. I was getting tired of looking at...
Blade.png


I would never rely on Z layers taking the loading! I would have been concerned about even doing a 3D printed propeller with the primary loads being printed in the X/Y. But...
The printed 6x4.5 on my test stand with an Emax 2822 1200kv.
helped to convince me. For a 125mm blade, I probably will use Nylon - stronger and far more tolerant of a hanger rash, transporting and even ground strikes.
 

Inq

Elite member
I was actually just going over an old post on the forums here. @willsonman was building a servo-driven variable pitch propeller for a P-51 that has yet to be finished (for good reason).
Might give you some insight that you can apply to your scenario. https://forum.flitetest.com/index.php?threads/winter-build-2019-2020-top-flite-p-51-0-60-size.60646/

Thanks... when I was searching, I didn't find that one. I go through it and see what the snag was. Should be an interesting read.
 

Inq

Elite member
That's a real shame. I've heard of people building up to a sensitivity to epoxies.
 

Pieliker96

Elite member
Say... if the motor/ESC can handle 30 amps, the on-board micro would measure the current and adjust the ESC and pitch until the desired amount of current is achieved (throttle sets max current from 0 to 30 amps). Got to do some head scratching to map which gets adjusted first... more ESC "throttle" or more aggressive blade pitch. But it sounds like a nice juicy science project.

As I understand it, the fundamental goal of constant-speed props is to tune the pitch of the prop to maximize its efficiency at the particular airspeed the airplane is flying. Measuring ESC input current and voltage will tell you how much power you're putting in, but not what you're getting out, and you need to know both to get efficiency. Using the current-based control is flawed: a valid solution for static thrust at lower power settings would be to increase pitch as much as possible and stall the blades. That loads up the motor but doesn't produce meaningful thrust - or, at least, much less than could be achieved with a finer pitch at a lower amp draw.

1673237448163.png


"Advance Ratio" J (the x axis on the chart) is directly proportional to airspeed V, inversely proportional to propeller angular speed n, and inversely proportional to prop diameter D. Beta is the blade angle, which can be set based on the measured advance ratio to maximize efficiency. Of course, maximum efficiency does not produce maximum thrust.

There's a ton of theoretical stuff out there about variable-pitch and constant-speed props but I don't think it's necessary to test one. The following is how I would figure out how a variable-pitch prop performs:

1. Build a flying testbed with the following sensors:
-Load cell between the motor and airframe to directly measure thrust in-flight
-Pitot tube
-Current and voltage sensor between the battery and ESC
Both the motor and pitot tube should be in aerodynamically undisturbed places as to get as close as reasonable to freestream conditions
Note: strictly speaking, current and voltage measurements are only necessary for the efficiency measurement, though they can also be used to enforce ESC current limits / motor power limits

2. Flight-test the testbed to collect data
-Devise a test matrix of various ESC PWM settings and propeller pitches, with multiple trials for each. Multiply the number of PWM settings by number of pitches by number of trials, this is the total amount of data collection passes you will have to do. This can get big quick so a good idea is to start with coarse steps to get the general trend then fill in the gaps with finer steps if you have time.
-Use a known safe constant-pitch setting for normal flight
-For each point in the test matrix, bring the testbed airplane close to stall speed and engage the desired ESC PWM setting and propeller pitch. Log thrust and ESC input power as the airspeed builds or falls, and allow it to reach a steady speed if possible. Do the same from the maximum level flight speed.

3. Process data
-Output power is equal to thrust times airspeed. Input power is equal to current times battery voltage. Output over input is equal to efficiency, a product of ESC/Motor/Prop.
-Make multiple 3D plots of thrust vs. ESC PWM and airspeed for different prop pitches, efficiency vs. ESC PWM and airspeed for different prop pitches.
-Fit 3D surfaces to each thrust and efficiency plot for each prop pitch setting, get their equations

4. Devise a controller
There are many ways to implement this. The first that comes to mind for me is "Proportional Thrust Control" - i.e. the throttle stick controls the percentage of the maximum possible thrust that is available at the measured airspeed. This would be achieved in practice by:
-Controller knows the equation of the 3D surfaces and the current airspeed
-Controller looks at the thrust 3D surfaces and finds the maximum thrust possible at that airspeed
-Controller knows throttle stick position and interprets it linearly from 0 to 100%
-Throttle stick position serves as a multiplier on the maximum thrust possible to determine the desired thrust
-Controller uses thrust 3D surface to calculate possible combinations of ESC PWMs and prop pitches that satisfy the desired thrust. It then cross-references those with the efficiency surfaces, and picks the combination of ESC PWM and prop pitch (interpolating on the latter) to produce maximum efficiency. This could also be done by finding minimum input power among the acceptable combinations. (Note: This could all technically be done with a single 5D hyper-surface-data-fit-object and its system of 5 equations, but It's easier to visually tune and verify numerous 3D surfaces and interpolate the remaining variables.)

Further additions could include checks to make sure ESC PWM is at minimum (motor stopped) when the throttle stick position is at 0%, adding filters on the airspeed and/or ESC PWM output to damp noise in the system, a kill switch to guarantee motor shutdown, a switch to disable active control and revert to a known good pitch setting, possibly a switch to feather the prop, and enforcing current/power limits.

I have done something similar to this in the past, measuring thrust in-flight with an EDF and a variable outlet area nozzle instead of a motor and prop combo. I ended up stopping after the data collection phase due to the effects being small compared to the noise of the load cell. I'd be willing to donate a flight controller, GPS, compass, and pitot tube of your choosing to the cause if you're interested, as @CampRobber graciously did for my exploration of the variable are fan nozzle. I feel the need to pay it forward, and think there's potential here for interesting data to be collected.

1673244402301.png


1673244451090.png
 
Last edited:

quorneng

Master member
Inq
I agree entirely about a scale looking prop. The German 'paddle' blades on the FW190 look fantastic and better than the 'needle' blades favoured on most allied planes!
You are mistaken that a big prop has to use a lot of power. It all depends on the rpm! You only need to rotate it fast enough to give the thrust you need. If you can find the right power motor with the right kV you can turn a true scale prop and at a suitable speed.
Variable pitch is only really required if you expect to exceed 200 mph. Unlikely in a small model. Even the Spitfire managed initially with a fixed pitch 2 blade prop and achieved close to 300 mph! The trouble was the fixed pitch to achieve this speed dragged down the rpm of the Merlin at take off such that it was only producing about 2/3 of its full power. A brushless electric motor however does not suffer the same problem.
Its all a case of finding a motor with the right power and at the right rpm for the prop.

There is a down side even if you can find a suitable motor. A scale prop for the same thrust will require more torque than a smaller one turning faster. The extra torque can create control issues at take off and at slow flying speeds close to the stall.

An example of a scale prop.
The Antonov AN2 biplane has always used a big 4 blade prop.
CockpitView.jpg

When I wanted to build an RC model I had to find both a suitable prop, then a motor that could drive it using the sort of power level I required. The motor is a bit heavier than I would have liked but by running on 3s LiPo rather than its intended 4s the power is still quite adequate.
05Jan19c.JPG

A scale size prop on a 1/14 scale plane.
 

Inq

Elite member
@Pieliker96,

I've read some of your topics and know you to be quite technical in your aerodynamic usage. My degree(s)/career were in structural analysis and I only had the one obligatory Fluid Dynamics class. I'm quite comfortable at the level of using Reynold's number, L,D,M = (Cl,Cd,Cm)ρV^2S/2 type work and use Martin Hepperle's JavaFoil and JavaProp for most of the heavy lifting... and to output DXF of the foils. ;) So if I sound like I'm arguing below, it is probably my lack of background and would really ask that you elaborate or reference articles appropriate for me to study.

As I understand it, the fundamental goal of constant-speed props is to tune the pitch of the prop to maximize its efficiency at the particular airspeed the airplane is flying.

For real planes (Lycoming, Continental, Merlin, MB, Jumo, etc) a constant speed prop is used to:
  1. keep the engine from overrevving
  2. when cruising to allow the pilot to get the most fuel efficiency. This is typically done by getting the engine at its maximum efficiency RPM by adjusting the pitch at the desired throttle setting and leaning out the fuel mixture until gas exhaust temperatures reach a do not exceed. I would also assume that WWII bombers and fighters cruising into Europe would also use these same techniques to optimize fuel efficiency.
  3. Climbing, max speed, dog-fighting - it's all about getting the maximum thrust at the current speed. Frankly, I'm not sure how this is done in a real-world sense. The pilots I know are jet pilots and that is all done in software and the pilot just shoves the throttle into afterburner and efficiency is ignored. I would imagine for WWII pilots, prop pilots, it's either seat of the pants or pre-defined settings the engineers specified.
Back to us using brushless motors...
  1. is irrelevant.
  2. unless someone is designing a solar glider and going for duration records, I think electrical efficiency is secondary to (#3).
  3. most of us want great take-off, climb and top speed. Eeking out the max efficiency might extend our flight time from 10 minutes to 10.5 minutes, but I don't think people care about efficiency when they shove the throttle stick to the hilt.
"Advance Ratio" J (the x axis on the chart) is directly proportional to airspeed V, inversely proportional to propeller angular speed n, and inversely proportional to prop diameter D. Beta is the blade angle, which can be set based on the measured advance ratio to maximize efficiency. Of course, maximum efficiency does not produce maximum thrust.

I want to come back to this after I've studied your comments and references. I see merit in it already, but my first gut reaction is it mainly applies to #2 above and is of a secondary concern. But like I said, I want to educate myself on it more before commenting more. That is not to say I am ignoring this, I just think JavaProp and JavaFoil are taking that to limits better than I can do by my calculations.

The following is how I would figure out how a variable-pitch prop performs:

1. Build a flying testbed with the following sensors:
-Load cell between the motor and airframe to directly measure thrust in-flight
-Pitot tube
-Current and voltage sensor between the battery and ESC
Both the motor and pitot tube should be in aerodynamically undisturbed places as to get as close as reasonable to freestream conditions
Note: strictly speaking, current and voltage measurements are only necessary for the efficiency measurement, though they can also be used to enforce ESC current limits / motor power limits

I think you have been very thorough in your testing outline. With my hobbyist electronics/MPU programming background, I can see how to do most all of the test rigging of the plane you've described. I can do the motor voltage, current, pitot tube, RPM, and run the telemetry back to a ground station live. All the above data points can be done for < $10. I still have one concern... I'm still a little fuzzy on possible interaction of a 2.4 GHz WiFi module along with the RC 2.4 GHz TX/RX. I've read both sides... they won't interfere and others saying it'll crash your plane for sure! :unsure: No worries either way, I can always log it to a SD card and post process it back on the ground if WiFi is an issue. The one thing I don't have a handle on is how to instrument a load cell between the airframe and the motor. I'm not willing to invest hundreds of dollars for that data point. Do you have a low-buck solution for that?

Although instrumenting the plane is something I'd want to do, I don't see it as a path to a control solution. I see it mainly as a verification tool. I don't see mapping as a viable solution of control. You've basically described the method of how they map various sensors in car engines to control fuel injection / crack timing to eek out the maximum efficiency / performance, EPA the exhaust and keeping the ICE motor from detonating. That would make a solution very one-dimensional. You make any change and you'd have to do the whole thing over. Put the power unit (CPU/ESC/Motor/VariPitch Prop) on a different plane, add weight to the plane, use a different motor, use a different propeller blade and you'd have to do the whole mapping over. I think I would tackle the problem with a more AI type mentality. It might be as simple as using a PID algorithm and give it goals and let the routine make adjustments based on conditions and input settings. That way any changes in plane, motor, prop are figured out by the controller.

All that being said, I would think that the first order of business of the controller is to keep the motor and ESC from destruction. I'm sure it's obvious setting too much pitch (at any RPM or forward speed) will drive the current to exceed limits. Even driving at a limit for too long may overheat one or the other. Thus, a temperature sensor on the motor and ESC might also be prudent.

I have done something similar to this in the past, measuring thrust in-flight with an EDF and a variable outlet area nozzle instead of a motor and prop combo. I ended up stopping after the data collection phase due to the effects being small compared to the noise of the load cell. I'd be willing to donate a flight controller, GPS, compass, and pitot tube of your choosing to the cause if you're interested, as @CampRobber graciously did for my exploration of the variable are fan nozzle. I feel the need to pay it forward, and think there's potential here for interesting data to be collected.

That is very generous of you. I think that can be tabled for some time. I think the very first order of business is the mechanical side. I am willing to design, engineer and analyze it and use the full capabilities of my 3D printers using PLA, ABS, Nylon, PC, CF/PF and supplement that with fiberglass and/or carbon fiber tow / epoxy. But I'm not willing to get into the expense of 3D printing / machining metals as was described in the @Tench745 link above for a P-51 rig. If I don't feel this can be done within those constraints, I'll share my results and stop the project. Throwing a blade is not an option. The safety margins will be large!
 

Piotrsko

Master member
Could be too mechanically complicated, but I have heard of CSprops in the past. How about starting with a "almost constant speed fixed pitch" prop? They work by using an airfoil with pitch up properties so that as the reynolds numbers increase the prop increases pitch by flexing.

Cheapo coaxial toy helicopters have variable pitch and insane rpm that might be a good starting point
 

Inq

Elite member
Could be too mechanically complicated...

:ROFLMAO: Oh! I'm absolutely sure you're right.

Aeroelastic Tailoring - I'm familiar with the analysis techniques required to do those designs using CF/Epoxy using Orthotropic material properties. It's actually fairly easy with the right analysis tools. The centrifugal axial forces can cause a blade to twist and as you mention a pseudo constant speed prop could be made. Using isotropic materials like plastics it can only be done with the plan-view shape of the blade. At which point I can't make a scale looking blade. The whole premise for the exercise.

All that aside - I'm doing it more for the mechanical/computer controlled complexity. I'm a little off - I consider this the fun part. Flying the plane will be more for proof-of-concept than any real fun I'll get out of it.
 

Pieliker96

Elite member
For real planes (Lycoming, Continental, Merlin, MB, Jumo, etc) a constant speed prop is used to:
I think I should've been more precise and used "variable pitch" instead of "constant speed" here. I also made the internal assumption in that statement of constant RPM - so "particular airspeed" should be "particular advance ratio". Internal combustion engines complicate things because their efficiency is highly variable, whereas I understand brushless motors and ESCs to have pretty much constant efficiency.

most of us want great take-off, climb and top speed. Eeking out the max efficiency might extend our flight time from 10 minutes to 10.5 minutes, but I don't think people care about efficiency when they shove the throttle stick to the hilt.
There's a flight time factor but also a performance factor. Maximum efficiency for a particular advance ratio necessarily (assuming powertrain efficiency ahead of the prop is constant through all RPM) means maximum power put down to the air, therefore maximum thrust. I do care about inefficiency if my prop is stalling on the ground, for example. Of course there's confounding variables I didn't consider here: namely, that changing pitch changes to load on the motor for a constant power and so changes the RPM and advance ratio. In my limited testing I've found that ESC PWM input is pretty much linear with ESC input power, at least with an EDF doing a static thrust test.

The one thing I don't have a handle on is how to instrument a load cell between the airframe and the motor. I'm not willing to invest hundreds of dollars for that data point. Do you have a low-buck solution for that?
The solution I used on the Efflux Mk. I was a 5kg rated beam load cell, a HX711 wheatstone bridge interface board, and an arduino nano for data logging. From a quick look there's load cell / interface board combos for under $10 on amazon. The beam form factor was great for an externally mounted EDF, which helped get it closer into the freestream. You may find the "S-beam" or "pancake" type load cells to be better for mounting a motor and prop on the front of an airplane.

1673284584909.png 20210527_212406_HDR.jpg 20210609_190412_HDR.jpg

load cell highlighted in blue in CAD, installed on aircraft. The load cell is the only thing attaching the fan to the aircraft, i.e. all the thrust force between the fan and the airframe has to go through the load cell.

Although instrumenting the plane is something I'd want to do, I don't see it as a path to a control solution. I see it mainly as a verification tool. I don't see mapping as a viable solution of control. You've basically described the method of how they map various sensors in car engines to control fuel injection / crack timing to eek out the maximum efficiency / performance, EPA the exhaust and keeping the ICE motor from detonating. That would make a solution very one-dimensional. You make any change and you'd have to do the whole thing over. Put the power unit (CPU/ESC/Motor/VariPitch Prop) on a different plane, add weight to the plane, use a different motor, use a different propeller blade and you'd have to do the whole mapping over. I think I would tackle the problem with a more AI type mentality. It might be as simple as using a PID algorithm and give it goals and let the routine make adjustments based on conditions and input settings. That way any changes in plane, motor, prop are figured out by the controller.
I agree that it's impractical to do this sort of testing, especially if the design ever changes, requiring a full retest.
A PID system would require some sort of feedback, which I presume to be force in this case - though now that I think of it, current/power based feedback could work if you monitored RPM and made sure the blades didn't stall by calculating their angle of attack. Then the controller would drive the blade deflection in the direction of increased thrust / power and find its maximum, within limits. That's a more intelligent and scalable solution if it works - just different tuning parameters for different configs.
 

Inq

Elite member
Yeah, the Clark Y airfoil is way too thick for a prop

If I'm going to go to all this trouble and mechanical/computer controlled complexity, I might as well use a modern airfoil. As mentioned above, CFD and aerodynamics is not in my tool bag. So I'll stand on the shoulders of giants on this one. I'll use Martin Hepperle's foils for an optimized racing propeller for F3D Models. It specifies a series of his own custom airfoils along the length of the blade optimized for the scale racing airplanes. More details can be found at https://www.mh-aerotools.de/airfoils/index.htm. Unfortunately, he uses a web development technique where a link to the specific page is not available.

To find a series of three articles on design, on the above link, press Pylon Racing in the left menu, then press, Propellers for F3D pylon racing models.

Anyway... I start with the plan form of the Ta152C propeller blade and his seven airfoils...

MH.png


The journey begins...