How would you design a cyclocopter?

[OCPatch]

Hey Nick, Jude, or anyone else:
What do the Cl vs. alpha and Cd vs. alpha curves look like during the dynamic stall phenomenon? I get that it is a transient condition at best, but it must predictably happen with every revolution of a cyclorotor blade if you guys are getting best thrust at 35 or even 45 degrees incidence wrt the blade path.

 

[Gabik09]

Hey Nick, Jude, or anyone else:
What do the Cl vs. alpha and Cd vs. alpha curves look like during the dynamic stall phenomenon? I get that it is a transient condition at best, but it must predictably happen with every revolution of a cyclorotor blade if you guys are getting best thrust at 35 or even 45 degrees incidence wrt the blade path.

Hey! During dynamic stall, the Cl vs. alpha curve typically shows a sharp drop after a peak lift coefficient as the flow separates and the stall begins. It often leads to a sharp decrease in lift and a significant increase in drag. The Cd vs. alpha curve will show a sharp rise in drag at high angles of attack, reflecting the increased flow separation and turbulence.

For cyclorotors, if the blade is reaching such high incidences (35-45 degrees), you'd expect these dynamics to occur every revolution, with periodic stall and recovery. This transient behavior is what makes achieving high thrust at these angles a bit tricky, but also why dynamic stall is crucial to the performance at certain RPMs.

 

[Caeden]

Thanks a lot for your feedback
At the Momment I am Testing the influence of Ground Effect on Cyclocopters at high pitching Angles of about 45°. As a Reference I also tested a 5inch Propeller does anyone know ho

w to interpret these Numbers. With the Propeller there is a very clear distinction that is almost not noticeable with the Cyclocopter

 

[Jude Schauer]

@Caeden That's really cool, I don't think I've ever seen anyone look at the ground effect for cyclorotors.

Here's some CFD simulation of a large rotor: https://core.ac.uk/reader/286449981 I would guess that the decreased ground effect might be caused by a rotors airflow spreading out more? Or from taking in air from the sides vs the top?

 

[Caeden]

Yeah that does seam quite interesting, might be worth investigating. For the time being I've been testing out different LW PLA 3D printed wings with fibre glass, which habe been working quite well.

 

[OCPatch]

Hi All,
I'm glad I checked back in here - more interesting developments!
It is a impossible to be certain without the particular parameters of the tests, but I'll lay out my assumptions that I'm making that influence my comments.

With the relative distances and (I assume comparable size of cyclorotor to propeller) the two cases each for propeller and cyclorotor are essentially IN ground effect and OUTSIDE of ground effect.

I think the sharp drop in efficiency for the propellers in likely transonic tip effects. If your 125 mm propeller is really producing ~58 Newtons of thrust (5900 gr-force) and the in-ground-effect case is producing ~78 newtons of thrust (7900 gr-force) then it is almost certainly at very high RPMs. Propellers don't work well when the tips approach supersonic speeds.

The cyclorotor is likely turning at considerably slower RPMs and rotor airspeed to generate comparable thrust per watt so transonic effects are likely not a factor.

As far as the marked difference, particularly for the propeller in- versus out-of-ground-effect, here is one possible explanation:
Thrust is mass flow rate multiplied by the velocity of outflow. Power is mass flow rate multiplied by the square of the velocity of outflow. Higher outflow velocity implies less power-efficient thrust generation.

The fluid outflow behind an unducted propeller running in compressible fluid tends to condense or contract in to a smaller, faster jet or plume. When in ground effect, the result is the opposite because the static pressure of the stopped fluid in the center of the outlet jet causes the flow to spread instead of contract. If we assume the same mass flow into the propeller disc in both cases then the flow through the smaller area at the exit must be faster to have the same mass outflow. Higher velocity of the outflow means less power-efficiency.

And if we look at the data and conclude that the cyclorotor is less sensitive to ground effects than the propeller, then two points might explain this:
The outflow side of a cyclorotor is producing vortex pairs from the trailing edge of each passing blade. The lower surface vortex that contacts the ground surface tends to 'roll' itself along that ground surface with lower velocity relative to the surface (lower relative velocity equals lower drag and losses) and the vortex shed from the upper surface effectively 'rolls' itself like a meshing gear on the lower surface vortex. It is possible that this behavior has the effect of causing relatively efficient flow out to the 'spinward', 'windward' side of the outflow instead of allowing for the stagnation point that occurs below the large, planar area below the in-ground-effect propeller. Note also that the entire rotor disc area below the propeller is in close proximity to the ground while the cyclorotor only has a small area of its 'drum' shape that is close to the ground.

Thanks for sharing your data. Best of luck on your research.

 

[Caeden]

Hi All,
I'm glad I checked back in here - more interesting developments!
It is a impossible to be certain without the particular parameters of the tests, but I'll lay out my assumptions that I'm making that influence my comments.

With the relative distances and (I assume comparable size of cyclorotor to propeller) the two cases each for propeller and cyclorotor are essentially IN ground effect and OUTSIDE of ground effect.

I think the sharp drop in efficiency for the propellers in likely transonic tip effects. If your 125 mm propeller is really producing ~58 Newtons of thrust (5900 gr-force) and the in-ground-effect case is producing ~78 newtons of thrust (7900 gr-force) then it is almost certainly at very high RPMs. Propellers don't work well when the tips approach supersonic speeds.

The cyclorotor is likely turning at considerably slower RPMs and rotor airspeed to generate comparable thrust per watt so transonic effects are likely not a factor.

As far as the marked difference, particularly for the propeller in- versus out-of-ground-effect, here is one possible explanation:
Thrust is mass flow rate multiplied by the velocity of outflow. Power is mass flow rate multiplied by the square of the velocity of outflow. Higher outflow velocity implies less power-efficient thrust generation. View attachment 249057
The fluid outflow behind an unducted propeller running in compressible fluid tends to condense or contract in to a smaller, faster jet or plume. When in ground effect, the result is the opposite because the static pressure of the stopped fluid in the center of the outlet jet causes the flow to spread instead of contract. If we assume the same mass flow into the propeller disc in both cases then the flow through the smaller area at the exit must be faster to have the same mass outflow. Higher velocity of the outflow means less power-efficiency.

And if we look at the data and conclude that the cyclorotor is less sensitive to ground effects than the propeller, then two points might explain this:
The outflow side of a cyclorotor is producing vortex pairs from the trailing edge of each passing blade. The lower surface vortex that contacts the ground surface tends to 'roll' itself along that ground surface with lower velocity relative to the surface (lower relative velocity equals lower drag and losses) and the vortex shed from the upper surface effectively 'rolls' itself like a meshing gear on the lower surface vortex. It is possible that this behavior has the effect of causing relatively efficient flow out to the 'spinward', 'windward' side of the outflow instead of allowing for the stagnation point that occurs below the large, planar area below the in-ground-effect propeller. Note also that the entire rotor disc area below the propeller is in close proximity to the ground while the cyclorotor only has a small area of its 'drum' shape that is close to the ground.

View attachment 249059

Thanks for sharing your data. Best of luck on your research.

Wow, thank you so much for your time and insight! It's clear that your knowledge of aerodynamics is far greater than mine.

Just to clarify, I made a significant conversion error with my load cell. The Newton values should be a factor of 10 smaller, meaning the maximum thrust is about 7.8N.

For the propeller, I used a 5-inch Ethix S3 Watermelon prop with three blades and a pitch of 3.1. The motor I used was a T-Motor F90 1300KV (2806.5) running at approximately 16V. Based on rough calculations, the propeller tips should be well below the speed of sound.

For the cyclorotor, I used a wingspan of 30 cm, a wing chord of 10 cm, and a rotor diameter of 20 cm. The motor and power source remained the same, with a 1:7 ratio between the motor and rotor. The wing shape is based on the NACA 0012 airfoil. The maximum blade pitch was close to 50°, so they may have been in a stalled condition.

Regarding the propeller’s behavior, one widely accepted explanation is that a propeller operating in ground effect (IGE) experiences higher pressure beneath the wing compared to a propeller out of ground effect (OGE). As a result, at the same rotational speed, it generates more thrust.

Unfortunately, I didn't record the RPM during the propeller test. However, if we use PWM as an approximation for RPM, we can see that the propeller produces more thrust at the same PWM but also consumes slightly more power. This suggests that, in ground effect, the propeller encounters more parasitic drag. However, the additional power consumption is nearly within the margin of error, so I’m unsure if we can confidently draw that conclusion—although it does align with what I've seen online.

As for the cyclorotor, I currently don’t have the means to visualize the vortices to confirm or disprove any hypotheses. If you have any suggestions for cost-effective ways to analyze airflow over the wing, I’d love to hear them.

Once again, thank you all for your insights!

Also, if you don’t mind me asking, what do you do for work, and what did you study?

 

[OCPatch]

Hi Caeden,
Last questions first: I have worked at corporate and private aerospace jobs - mostly scientific programming - and in medical device development where the bulk of the work was designing electronics and firmware. I attended university in the 1980's and I received degrees in aerospace engineering and computer systems engineering.

The thrust that your propeller produces seems much more reasonable now that you have explained the scaling error on the graph of efficiency versus Newtons of thrust. And it also gives me one more idea about why the propeller efficiency appears to hit a brick wall at a particular point. I agree with your assessment that the issue is not likely transonic tip speed. I think the issue is one of electrons, not air molecules.

As I look at your PWM graph I think that maybe those sharp drops in prop efficiency occur at approximately the same PWM setting. There is a phenomenon that occurs with DC motors called "Back ElectroMotive Force" or Back EMF. Basically a DC motor will produce a voltage that acts against the supply (battery) voltage that is nearly proportional to the motor RPMs. A DC motor will produce its greatest torque at nearly zero RPM and then successively less power as RPMs increase. There is less and less net voltage left over to actually drive a propeller as the Back EMF voltage increases. Maybe this causes the sharp loss of efficiency, but Back EMF is is usually seen as a gradual flattening of the thrust versus power applied curve. (Or the down-sloping efficiency curve above)
The more likely reason may just be your Electronic Speed Controller (ESC), or if you are using batteries to power the tests, then a Battery Management System (BMS). These electronic devices may be limiting the current drawn from the batteries at the highest PWM settings. You already know that the ESC does not turn on until it sees nearly 1200 microseconds of pulse width on its signal input. If the motor is turning excessively fast then the ESC may not be able to switch phases efficiently, correctly sense the voltage rise on the other phases, etc.
I have an annoying "smart" ESC on a solar-powered plane project which tries to determine how many lithium cells are in its (assumed) battery pack when it first powers up and then won't drive the motor if the input voltage is below 2.x volts per cell . Well, I don't have a battery pack - I have solar cells that don't have to be protected from over-current or under-voltage. But the ESC faithfully denies my PWM requests for more throttle if the panel voltage drops to some lower voltage. After that, even if the panel voltage increases again to what it thinks would be 4.0 volts per cell, it won't allow motor power above the equivalent of about half throttle because it thinks it has batteries that it needs to protect. So I have to 'trick' the ESC by never letting my controller give a throttle command high enough to cause the ESC input voltage to go too low. The ESCs that I have can be programmed for many different things, but if the ESC is designed to handle different battery voltages (2S, 3S, 4S, ...) then this battery protection 'feature' seems like it cannot be disabled. And I don't have enough free time on my hands that I want to go off and design my own ESC.

I contend that if you used a high enough gear ratio on your cyclorotor (let the motor spin faster) you would eventually see similar behavior to the thrust "brick wall".
The problem at the low end of the RPM range is that the motor can draw too much current from each phase as it is driven. Rotor locked and full pulse width, your 16 volts, F90 1300KV TMotor could (until it melted!) pull over 200 Amps of current. If you run the motor too slowly then too much power is lost as heat (current squared multiplied by resistance) and you just have a slow spinning, very hot motor. :-/
I also contend that if you use a larger propeller (6x4 or 7x4 two bladed prop) on the motor then you would both see higher baseline efficiency numbers (see my post above on thrust versus power efficiency where flow velocity is considered), and those sharp drops in power efficiency would occur at higher values of thrust (or maybe not at all).
And as one of my old professors used to say "Go figure out the optimum solution, or at least a sufficient one. That is why engineers get paid the big bucks!" ;-)

Visualizing flow and vortices produced by the cyclorotor would be useful if it could be done. I'm not sure what the state of the art currently is, but here is an older YouTube video on Particle Image Velocimetry (

). Presumably for the cyclorotor you would have the laser sheet running parallel with the flow direction and you would be interested in a series of images just as a rotor passes the bottom position. Here a two-bladed cyclorotor would allow best visualization. The stationary camera would look from the side, perpendicular to the laser sheet, and you would take a series of images as the blade passes through the field of view and then animate them into a short video. You might be able to get really clever about releasing successive vertical lines of smoke puffs or particles upwind of the test area to form a grid as the blade passes. Schlieren imaging probably wouldn't illustrate the vortices very well.

I just now realize that I assumed, but didn't say it explicitly except in the diagram, that the direction of rotation of the cyclorotor and the airflow are such that the airflow and the rotor blades at the bottom are going the same direction for the ground effect test. I think that is the more interesting case. The cyclorotor blades are only in proximity to the ground for a short time and then they are up in almost unaffected air the rest of the time.

 

[OCPatch]

This is a follow-on to the message above:
If my calculations are correct then the 'brick wall' drop in efficiency seen in both the IGE and OGE propeller cases occur at approximately 156 watts. It seems like there is a power limit or current limit in your power supply.
Oddly enough, if my calculations are correct, the cyclorotor cases seem to be able to pull over 250 watts at their maximums.

 

[Caeden]

that could be. for more information the tests with the propeller where all conductet to 2000 pwm but i think that my esc only increases the rpm of the propeller to about 1900pwm to last of the ramping of the test the motor doesnt spin faster so it could be that the esc motor combo is less efficient at higher pwm values or that the battery sag reduces efficiency witch would make sense

 

[Jude Schauer]

I put together and tested a new transmission design recently - I moved away from the planetary gear to pursue this because I was getting too much vibration with plastic on plastic gear contact. I put the idler pulley peg in a "T" slot to constrain movement and threw in a spring. It keeps tension soooo much better than the original pulley design. The belt doesn't slip until I'm cranking on it enough to significantly deform all the plastic, and I'm now working with a 6.8:1 reduction ratio. I actually think I dropped weight overall because I redesigned the motor mount to be more structurally efficient, and the spring weighs basically nothing

The transmission was the final thing that I wanted to redesign before my next steps. I'm rolling back the airfoils and the vectoring system to the design that I successfully hit a 5:2 T/W ratio with. I realized with the last rotor redesign that I tried to incorporate too many ideas so I didn't know which changes were causing issues. This attempt should work fine because the only big change is the transmission idler. For the next steps I'm looking to put together a test rotor to verify tolerances and my manufacturing methods, and then I'll throw together 8 and get to testing controls. *I'll be adding landing gear, cross bracing, and maybe a rotor tip cage to the design pictured below*

 

[Jude Schauer]

New update - solved some clearance issues with the transmission so I finally let it go full throttle on the scale. I have some badly formatted data below (ie data not recorded at constant intervals) and its' respective analysis. Sorry if the data output or this explanation is confusing, I was somewhat jumping around and trying different things during testing. I also threw in a current pic of the rotor.

Conclusion: There are some funky things going on when I run the rotor on 24v. I'm going to try to see if I can figure out why it is happening, but I will also be moving on and constructing 8 rotors now that I have the rotor performance validated. It also broke 400g thrust!


This rotor is not performing quite as well as the previous one I made (Fig 1 vs fig 3). This is probably because I had to switch the types of foam for the blades so I think they are bending more. Next, the thrust capped out at 308g for a 3s battery (13v power supply). I moved on to try 24v. This setup is able to run up to a higher voltage/power output than the last time I ran these tests. This is probably because the side plates (main supports for the rotors) are torsionally stiffer and therefore there is less slop in the overall design and it is able to hold shape better at higher rpms (before, thrust was capped at 370g and adding more power wouldn't change the thrust because it was too badly deformed). With a rotor weight of 70g, the new 400g of max thrust gives this rotor a thrust:weight of 5.7:1. Once I throw these together into a cyclocopter, this should result in a rough thrust:weight of 3:1 for the whole cyclocopter, which would be quite sporty. I noticed that at low throttle settings, running the rotor at 24v is more efficient than at 12v (makes sense because less current should mean less thermal waste). However, at high throttle settings running it at 12v is more efficient? Literally no clue why this is happening, I might redo some of my makeshift wiring and try again. From the graphs, it looks like the 24v results drop off much faster than the 12v results, supporting the idea that something is wrong electronically. The 12v results are pretty close to a flat line, which is kinda cool.

Fig 1. Charts of new data

Fig 2. Graphs of new data

Fig 3. Old data

Fig 4. Rotor

 

[Piotrsko]

Higher power may mean more drag overall or less system efficiencies. Overlay the 12 v chart on the 24v chart and you see the charts dont seem to quite match. Could be data error, measurement error, I dunno.

 

[Jude Schauer]

Very successful testing today. I threw together another rotor this week to verify my manufacturing process and test out the new order of motors that came in for the actual cyclocopter build. I tested the rotor until failure (see pics at the end lol). With a rotor weight of 72.1g, the rotor achieved recorded bursts of 540.2g thrust. After testing, I verified the thrust stand measurements with a 450g load and discovered that there is a 4.44% error in the measurement. This error impacts the results not insignificantly and after correction, the true value of thrust is can be calculated at 517.2g. This is over a pound of force! This is a thrust to weight of 7.17:1, absolutely decimating any other rotor out there.

I did much better at taking data in regular intervals this time lol. I recorded amperage and thrust measurements in 25g intervals starting at 50g. Note the performance drop off starting at 400g thrust (the blades begin to significantly bend. There also might be heat buildup in the motor that affects efficiency). I recorded data in the 25g intervals up to 450g, then decided to push the rotor to the max speed and see if it would fail. The limit was reached first by the power supply, as it only goes up to 10A. After a couple of bursts at this max speed (just long enough to collect data), I left it on this power setting while we recorded rotation with a high speed (960fps) camera. After recording some footage, three of the blades snapped in half and the rotor disintegrated. I ran some numbers afterwards and this is somewhat understandable given that the blades were under a calculated 812 Gs of acceleration.

(With the data above, keep in mind that performance should be 4.44% lower than displayed. I'll recalibrate the thrust stand before the next time I start testing things)


If the 7.17:1 thrust/weight ratio doesn't impress you, take a look at the graph. With a 72g rotor weight, I'm expecting an aircraft with 8 rotors + a medium-sized battery to weigh around ~1kg. This would result in a whopping 4:1 thrust to weight ratio. A quadcopter with this thrust to weight ratio would be on the top side of being a sporty quad, and borders on being in the performance racing drone class. I'm sure I could do an ultralight build and get it even higher. With the above assumptions on vehicle weight and a 2200mAh 6s battery (275g), the math works out to a 16.50 minute flight time estimate for hovering. A drone with a 4:1 thrust to weight ratio and a 16 minute hover is fairly competitive with drone builds out there now. Not bad for a novel 3D printed technology. We shall see when I actually throw this together but the stability and translational mobility should also be much higher than a conventional quadcopter drone. Next steps: I am very satisfied by the ease of manufacturing and the performance of this design, and I'll be building 8 rotors without making any more changes. The failure of the rotor at 3100rpm is at a much higher throttle setting than it will actually reach when flying, so I'm ok with it. Before attempting flight, I might do a test of the thrust at something like 300g to validate that it is able to withstand that rotational speed for 10-20 minutes without the blades fatiguing.

I'm really hyped by the testing today so y'all let me know if there's any thoughts or ideas.


Below are an assortment of pics from testing today. The rotor bulge when its generating 520g of thrust is insane. There's almost none at the lower levels of operation, but it really starts to pick up between 300g and 400g of thrust. I do want to note that the spec sheet for the motor says that it can go up to 18.8A - nearly double the 10A that my power supply is limited at. If I can figure out a way to stiffen the blades without adding much extra weight then I'll surely be able to push the thrust even higher.

Here are the stats that I ran to calculate the loading on the blades. The rpm was estimated at 520g from my 960fps camera by frame counting.

 

[Jude Schauer]

@NickRehm question for ya

Are you aware of any reasons not to pursue a design like below? From my knowledge and past trials, the gyroscopic stability offered by this arrangement seems to be a huge advantage. Symmetrical roll/pitch will make PID tuning much easier, and fast translation in every axis will be possible. The "+" rotor arrangement has no gyroscopic stability, and the "H" arrangement only has it on one axis. Has this "O" orientation not been tried yet only because it involves manufacturing a large number of rotors?

 

[NickRehm]

I don't see any reason why that config won't be workable! Keeping the two parallel sets counter-rotating at the same RPMs in flight would drastically simplify your flight control mixing because you wouldn't be coupling the induced torque from the rotors into your control strategy (cancels out) -- control moments could be just from the offset-force moment like a regular multirotor which is much more intuitive to setup.

Yaw would be simple and quite effective if you vector them all to creative a clockwise/counterclockwise moment, and then you can play with vectoring the left/right top/bottom pairs together for translational motion in any direction once stabilization is ironed out

 

[Jude Schauer]

I got it fully built and wired up sometime last week. I also got a dRehm Flight code all modified up. However, I just cannot get the MPU6050 (gyroscope and accelerometer) to connect to the teensy! Every time the MPU6050 initialization is unsuccessful. A little strange given that I've connected the exact same teensy and the exact same mpu6050 before.

Some troubleshooting I've done:
- Three different teensies
- Three different MPU6050s
- 5v and 3.3v wiring to mpu6050 - either way the green light on the mpu6050 turns on. Verified voltage with multimeter
- Fresh unmodified install of dRehmflight - did not update libraries or anything, everything is straight from the github
- Several random I2C scanners - zero I2C devices spotted
- verified SDA and SCL pins are correctly connected to pins 18 and 19 many many times
- I only have the teensy, mpu6050, and 5v power connected together. Took everything else off the circuit.
- As mentioned, 5v and 3.3v outputs have been verified and everything is in the same circuit with a common ground

Not too sure what's going on but I also tried looking through some of the other dRehmflight forums on RcGroups. I've used the full extent of my internet searching as well as AI to see if there was anything obvious. I'm sure it'll be something easy that I'm missing. @NickRehm any ideas for further troubleshooting?

 

[NickRehm]

hmmm at that point in the troubleshooting process I normally toss the IMU and try a new one...sometimes they are just duds

 

[Jude Schauer]

Yeah that's the only thing I could think of proceeding with.

The issue is that I've tried several IMUs and teensy 4.1 boards and none of them have worked. I did order a teensy 4.0 and another pack of IMUs - maybe a teensy keeps messing up the IMUs or something like that. Not sure what else to try.