Two points: 1. You will have to look into gyroscopic precession for the control system on this project. As the craft has any angular velocity the torques produced by the control vanes will act 90 degrees away from the intended direction. A common solution for this is an angular velocity feed forward controller. I will reply to this comment with a good paper if I can find one quickly. 2. The passive stability does not have much of an effect in this case. A single rotating blade will naturally be unstable. Even for rockets they require significant airspeed before they become stable with a CG forward of CP design. You can see this in sounding rockets that wobble right off the launch rail but then seem to lock into an orientation. This is a very difficult project to complete but it looks like you are well on your way on the hardware side. Maybe setup a gimbal to hold the craft safely while you test modifications to the control system and keep testing! Good work!
ruclips.net/video/RMeEh5OUaDs/видео.htmlsi=7BzNk5PqsWQG3UgI projekter.aau.dk/projekter/files/421577367/Master_Thesis_Emil_Jacobsen_v5.pdf The video is about exactly what you are doing. The paper describes a contra-rotating setup but could be helpful in making a 6dof model of your system.
Big thanks. Yeah it was pretty dumb to do this without a gimbal test thing. I've somehow never heard about gyroscopic precession before so this will be a fun rabbit hole
cool video, ive seen some other people attempt similar projects and it seems pretty difficult. what i would say is instead of using one big battery at the top you could use two batteries that are half the size but the same amount of cells (one on each side) to move the center of mass down a bit. maybe you could also build some kind of test stand so that you can get the tuning down before you actually fly it. would love to see a follow up video!
Thank you!! The two battery set up is smart. I wanted some wiggle room in terms of longitudinal stability so that's why I ultimately went with this. But yeah I REALLY need a test gimbal for this
W&B is key…(heavy weight behind the thruster, not in front), test with a gimbal to tune your values this even help with ruffing) and try using a launching scaffold instead of a pad. it allows the vehicle to avoid ground effect issues, giving the thrust column more room to have an effect on the vector/direction
@@mrday60 I disagree about the weight behind the edf. Look at any rocket, the centre of gravity is infront of where the thrust is produced. This is called the rocket paradox, there's a great Tom Stanton video explaining why this is. I do agree about the gimbal.
Great video! I’ve been working on a similar project and noticed something interesting. When tilting the drone from its normal position to a certain angle, the motor speed doesn't seem to increase as needed to balance the drone properly. Since only the vertical component of the motor thrust contributes to lift, this could be why it's not stabilizing as expected. Mathematically, the vertical lift component can be represented as F_vertical = F.cos(θ), and the horizontal component as F_horizontal = F.sin(θ). F_vertical should be equal to the thrust when the drone is in its normal orientation without tilting. Keep up the great work!
Sick! Would love a video at some point :) but yeah thanks for pointing that out, forgot to mention in the video. Right now I’m manually controlling throttle from the TX, so I figured motor speed compensation wasn’t entirely necessary at this stage. Also, I’m only using an MPU and would obviously need an altimeter or some equivalent to maintain constant altitude. That’s why I neglected to add this part of balance in this first iteration. But now that you mention it I will definitely try to incorporate this into the next one, didn’t think of this as a problem until now!
Great video! I did pretty much the same project a few years ago with basically the same components just a little bit bigger (90mm edf) here are my 50 cents: - as gsisk911 pointed out: you have to account for gyroscopic precession: I measured the moment of inertia of the rotating assembly and actively measured the current speed of the EDF. With that, I was able to take into account the effect of gyroscopic precession in order to move to my target angle - a gimbal is your friend (especially for PID tuning) - I first used the same servos as you but then switched to faster digital ones for a faster response -I used 4 smaller Batteries around the circumference of the EDF for better center-of-mass tuning but your idea of putting the battery in front of the EDF also looks enticing - another way to improve the lift of an EDF is the Coanda effect - those EDFs are stupidly fast so mind your fingers (speaking with experience ;)) - eventually, I replaced the Arduino with a teensy 4.0 for a faster loop frequency - and tbh my EDF drone did fly but the control authority was low and it always wobbled a bit in the air. Really not an easy project but you are on the right way
This is huge! Love every bullet, thank you. Working on 2nd iteration with a teensy, new servos, hopefully better aero, would love to get where you got.
This design will probably be impossible to fly using an Arduino with cheap servos. The problem is that it's so short it has very little rotational inertia. You need to put the battery on a tall stick so that the centre of mass is high, giving you more stability. Watch Tom Stanton videos about it.
100% upgrading my servos. I'm having a little trouble understanding stability in this application but I think I understand what you're saying, a larger static margin would also increase my air pull I think.
@@mach10point4 basically imagine trying to balance a long pole on your hand, it's easy, but a pencil, almost impossible without insanely fast and accurate reflexes
That's cool. I reckon the reason why the control system didn't work is because your control response is reversed. You can see in the slow motion at 1:12 that the vanes rotate in a direction that would increase the angular velocity, rather than reduce it (this is bad). This is also seen at 4:44. Reversing the direction of the servos would definitely help to improve the control. EDFs are quite complicated. Their static thrust and static pressure are very low because of the high angle of attack of each blade. This means that when it is sitting on the ground, even at full throttle, you'll get nowhere near 1.6kg of thrust and it'll be really inefficient. This is why EDFs are commonly used on rc jets, because they go fast and have a high intake velocity. Definitely a cool project and it would be great to see it work. With a better software implementation, a well tune PID and probably some stronger servos with less jitter, it would definitely work!
Thanks for the heads up. Yeah I've seen this done with regular props which makes more sense thrust wise, the more I work with EDFs the more I see how limited they are. Appreciate this man!
In the picture of the EDF, it looks like the motor sits beneath the fan. The motor is going to be a significant weight, so I wonder if you can flip that around to be on top of the fan instead to help shift CG up Either way, very cool project. Looking forward to seeing your progress
Huh. Never crossed my mind. With this particular EDF I can reverse the direction but with the lip and stators it wouldn’t be practical. But designing a custom EDF in the way you’re saying would actually really help out a lot
Great job. Keep it up.
good job man
💯
Thats great work!!!
My guy!
Keep it up!
Two points:
1. You will have to look into gyroscopic precession for the control system on this project. As the craft has any angular velocity the torques produced by the control vanes will act 90 degrees away from the intended direction. A common solution for this is an angular velocity feed forward controller. I will reply to this comment with a good paper if I can find one quickly.
2. The passive stability does not have much of an effect in this case. A single rotating blade will naturally be unstable. Even for rockets they require significant airspeed before they become stable with a CG forward of CP design. You can see this in sounding rockets that wobble right off the launch rail but then seem to lock into an orientation.
This is a very difficult project to complete but it looks like you are well on your way on the hardware side. Maybe setup a gimbal to hold the craft safely while you test modifications to the control system and keep testing! Good work!
ruclips.net/video/RMeEh5OUaDs/видео.htmlsi=7BzNk5PqsWQG3UgI
projekter.aau.dk/projekter/files/421577367/Master_Thesis_Emil_Jacobsen_v5.pdf
The video is about exactly what you are doing. The paper describes a contra-rotating setup but could be helpful in making a 6dof model of your system.
Big thanks. Yeah it was pretty dumb to do this without a gimbal test thing. I've somehow never heard about gyroscopic precession before so this will be a fun rabbit hole
cool video, ive seen some other people attempt similar projects and it seems pretty difficult. what i would say is instead of using one big battery at the top you could use two batteries that are half the size but the same amount of cells (one on each side) to move the center of mass down a bit. maybe you could also build some kind of test stand so that you can get the tuning down before you actually fly it. would love to see a follow up video!
Thank you!! The two battery set up is smart. I wanted some wiggle room in terms of longitudinal stability so that's why I ultimately went with this. But yeah I REALLY need a test gimbal for this
keep grinding dude
Appreciate it homie
W&B is key…(heavy weight behind the thruster, not in front), test with a gimbal to tune your values this even help with ruffing) and try using a launching scaffold instead of a pad. it allows the vehicle to avoid ground effect issues, giving the thrust column more room to have an effect on the vector/direction
Very good tips, thank you!
@@mrday60 I disagree about the weight behind the edf. Look at any rocket, the centre of gravity is infront of where the thrust is produced. This is called the rocket paradox, there's a great Tom Stanton video explaining why this is.
I do agree about the gimbal.
Great video! I’ve been working on a similar project and noticed something interesting. When tilting the drone from its normal position to a certain angle, the motor speed doesn't seem to increase as needed to balance the drone properly. Since only the vertical component of the motor thrust contributes to lift, this could be why it's not stabilizing as expected. Mathematically, the vertical lift component can be represented as F_vertical = F.cos(θ), and the horizontal component as F_horizontal = F.sin(θ).
F_vertical should be equal to the thrust when the drone is in its normal orientation without tilting. Keep up the great work!
Sick! Would love a video at some point :) but yeah thanks for pointing that out, forgot to mention in the video. Right now I’m manually controlling throttle from the TX, so I figured motor speed compensation wasn’t entirely necessary at this stage. Also, I’m only using an MPU and would obviously need an altimeter or some equivalent to maintain constant altitude. That’s why I neglected to add this part of balance in this first iteration. But now that you mention it I will definitely try to incorporate this into the next one, didn’t think of this as a problem until now!
Great video! I did pretty much the same project a few years ago with basically the same components just a little bit bigger (90mm edf) here are my 50 cents:
- as gsisk911 pointed out: you have to account for gyroscopic precession: I measured the moment of inertia of the rotating assembly and actively measured the current speed of the EDF. With that, I was able to take into account the effect of gyroscopic precession in order to move to my target angle
- a gimbal is your friend (especially for PID tuning)
- I first used the same servos as you but then switched to faster digital ones for a faster response
-I used 4 smaller Batteries around the circumference of the EDF for better center-of-mass tuning but your idea of putting the battery in front of the EDF also looks enticing
- another way to improve the lift of an EDF is the Coanda effect
- those EDFs are stupidly fast so mind your fingers (speaking with experience ;))
- eventually, I replaced the Arduino with a teensy 4.0 for a faster loop frequency
- and tbh my EDF drone did fly but the control authority was low and it always wobbled a bit in the air.
Really not an easy project but you are on the right way
This is huge! Love every bullet, thank you. Working on 2nd iteration with a teensy, new servos, hopefully better aero, would love to get where you got.
Great Video! The thrust vectoring doesn't seem gross at all. In fact, it seems quite clean.
Haha appreciate that. I do really hate all of these wires everywhere lol
This design will probably be impossible to fly using an Arduino with cheap servos. The problem is that it's so short it has very little rotational inertia. You need to put the battery on a tall stick so that the centre of mass is high, giving you more stability. Watch Tom Stanton videos about it.
100% upgrading my servos.
I'm having a little trouble understanding stability in this application but I think I understand what you're saying, a larger static margin would also increase my air pull I think.
@@mach10point4 basically imagine trying to balance a long pole on your hand, it's easy, but a pencil, almost impossible without insanely fast and accurate reflexes
That's cool. I reckon the reason why the control system didn't work is because your control response is reversed. You can see in the slow motion at 1:12 that the vanes rotate in a direction that would increase the angular velocity, rather than reduce it (this is bad). This is also seen at 4:44. Reversing the direction of the servos would definitely help to improve the control.
EDFs are quite complicated. Their static thrust and static pressure are very low because of the high angle of attack of each blade. This means that when it is sitting on the ground, even at full throttle, you'll get nowhere near 1.6kg of thrust and it'll be really inefficient. This is why EDFs are commonly used on rc jets, because they go fast and have a high intake velocity.
Definitely a cool project and it would be great to see it work. With a better software implementation, a well tune PID and probably some stronger servos with less jitter, it would definitely work!
Thanks for the heads up. Yeah I've seen this done with regular props which makes more sense thrust wise, the more I work with EDFs the more I see how limited they are. Appreciate this man!
In the picture of the EDF, it looks like the motor sits beneath the fan. The motor is going to be a significant weight, so I wonder if you can flip that around to be on top of the fan instead to help shift CG up
Either way, very cool project. Looking forward to seeing your progress
Huh. Never crossed my mind. With this particular EDF I can reverse the direction but with the lip and stators it wouldn’t be practical. But designing a custom EDF in the way you’re saying would actually really help out a lot
use 4 cels lion batterry attaches to the side
How about a counter rotational fan to offset the torque of just one fan like the Russian helicopters use
Counter rotating fans are sick. I read a few papers on them recently and it sounds like the complexity outweighs the benefit
Aerodynamic center wont matter here cause the copter aint going too fast.
Your servos are goind the wrong way (see 4:46)!