Thanks for watching this video! We now have a reliable launcher, and we've got the flight computer working. We've got more flight time planned in February to do some more tuning flights and get some landing practice. Once we've proven we can do launch, stabilised cruise and landing reliably, we will then attach a stronger bungee to the launcher and maiden the Kingfisher. We're sorry that the timescales have been longer than previously promised - we want to make sure we have everything ironed out at the sub-scale to minimise risk to the main flight. This is from both a safety perspective and aircraft perspective - repairing the 3DP is orders of magnitude cheaper and easier than repairing the main Kingfisher. We're constrained to flying on weekends with good weather, since we work full-time engineering jobs in the week. Some of you may also be disappointed that we didn't drop the aircraft out of a helicopter, but we think a catapult is a lot easier to repeat for the full-scale. We also cannot afford a helicopter! Overall, this sub-scale flight campaign is about replicating the real flight as closely as possible, that's why we've matched the wing cube loading to Kingfisher, made the 3DP way heavier than it needs to be, and used a slightly overkill bungee system! Should also add, Chris was unavailable for this flight so we had a different pilot.
Are you gonna be able to get it to the recovery the drone also. I’ve been talking about this happening on a commercial level to save weight and complexity of electric aircraft.
RC flight is not my world, but I do fly small airplanes. One thought that came to me was about the use of an AOA (angle of attack) sensor to feed into the autopilot to provide stall warnings. I suspect that this would require watchmaking skills to create one small enough to be practical. It is fun to think about it. Kudos on a great project, I look forward to future installments!
That Kingfisher design is a beauty! Really liked the animation showing the crash sequence. Great video, looking forward to more flight testing! (My favorite part of any aviation project!)
I wonder if you're going about the use of the flight control system backwards. I would recommend starting your flight out without flight control to get your trims, throws, and expo sorted out so that you can fly it "easily" up to 100mph. Only then would I turn on the flight control to help keep things stable for max speed runs, where a human couldn't possibly react fast enough to maintain proper flight envelope. Unless your idea of breaking the speed record is taking it up really high, then accelerating straight down and your top speed is what's recorded a millisecond before it hits the ground.
I think you're absolutely right that we need to get the trim, throws and expo (the basics) sorted first. In order to sort these out we've needed flight time and our major blocker has been the launch method. This video represents our process of sorting this out so now we should be able to get repeatable flights to sort these control issues out over the coming flight tests. The flight launched in manual and we were checking the different modes to work out what needs to change. The rest of the flights would've been adjusting the trim, c.g. and then tuning just like you are suggesting. Thank you for the feedback we really appreciate it! -Josh
If they fly high enough and accelerate fast enough, I think they could reach top speed in time to pull out of the dive but it probably wouldn't count anyway.
@TheMachInitiative Personally I think your project management is pretty LEGIT lads!!!!! Clearly your college did its job properly. Eager to see what your next phase brings. Genuinely sending as many positive vibes as possible from up here in the Norf (west yorks).
Excellent work! As far as those program adjustments, I can't help. I can say the autopilot compensated for the cg being too far back. Switching to acro mode would definitely cause a crash if one was not prepared. However, stabilized mode will work far better if you tune the aircraft to fly in acro mode. Considering the flight speed and flying without stabilization, exponential will likely be necessary for easy handling in the air.
Like other said (maybe), why don't you use a launch rail? Like a mix of a catapult launcher as found on aircraft carriers and what student rocket teams use? I believe you you see greatest improvements by angling the entire runway instead of having a ramp at the end, because that sudden change in direction puts unnecessary stress on your aeroframe and may even cause flow separation right when you need the flow over the wings the most. I did a launch system for my glider using a 1:2 pulley mechanism hat worked great. Look into how sailplanes are being towed and unhook automatically. It would help y'all a lot
i totally agree, a launch rail with weights seems much more consistant and controllable than your current setup, the pros seem very many: adjustability (if you use pulleys and weights), scalability, less stress on the airframe, less setup and the math is basic kinematics. plus you were kind of halfway there already by guiding the cart with a taut line
We agree a launch rail would be ideal, but there is a logistics challenge with design complexity, setup and transport. We’d need to break it into sections to fit in the car. Having 10 plywood boards and a guide wire is very easy to setup at the field. If we made an >10m long inclined rail, we’d need to have locking mechanisms to ensure all parts of the rail are perfectly aligned, we’d need to have a support structure if inclined along the full length, etc Please can you send photos of your pulley mechanism - really interested to see it!
@ Awesome to hear from you! Sorry, I did the pulley thing 12 years ago when I was 12 years old just messing around. I was literally a cord anchored to the ground with a long stick (like in your videos) and a handle with a roller in the middle of the cord. at the end, the glider was mounted with a hook and a lop in the cord. For every X you pulled on the handle, the plane moved 2X forward. That was enough for my needs. If I were you, I would just put a BLDC on a winch with a cord, do the launch rail thing and pull the sled like so. You could make a cheap launch rail out of 20x20 alu profiles, that are easily stowed in a small car. Setting up may take a little longer, but the advantages far outweigh the increased logistical complexity. I'm rooting for you!
@TheMachInitiative You could go for a rocket assisted take-off on a launch rail. However, a 1m rail would require your airframe to be able to handle 24g and you'd need a model rocket motor with that amount of TWR, so probably a 4m long rail with two 2m sections could barely work. But given the small loads, aluminium extrusion profiles would be enough, with a 1m tall bipod at the end. With lockable hinges you could even make it foldable and get the set-up time down to a minute.
While I'm sure you lads have excellent vision (by way of corrective lenses in one case lol), tracking something moving at 761mph will not be easy. Especially something the size of a carry-on bag. Would you be inclined or open minded to including a simple FPV setup? You've got everything you need for on-screen telemetry with the Matek FC and GPS, all you need is a camera and VTx. I'd be happy to sponsor a setup for you, and this epic project.
For launching you could piggyback on a larger plane, that way you don't need a ramp and would have the advantage of height. I used to do it with my old zagi on a friends virtually indestructible plane using Velcro, to separate I would pull up and he would dive, it worked great 95% of the time.
You could stick a motor and pusher prop out the back for these early prototypes. Launching would be much easier and much longer flight times for tuning, then swap in a EDF when everything is tuned
Excellent progress guys! Maybe ACRO mode was a bit ambitious, but you don't know until you try. Looks like you've got the roll damped nicely though, and a sweet airframe!
I second the drop launch idea. Surely rhe airframe and control surfaces want to be optimised for speeds much faster than youd be able to take off and (probably more importantly) land at on grass strips. A drop launch from a bigger, slower RC platform would surely allow you to hit takeoff speeds suitable for a sleeker wing and smaller control surfaces. Maybe a 'stall and chute' type system for a controlled recovery?
Living hinges might be doable with a printed design as small as this. I'd reduce the elevon depth by a couple millimeters and move them farther out - remove the control surface at the wing root and instead extend it to the wing tip. Basically like a Mirage 2000. Couldn't find anything about this model on your website - so no idea what airfoil you're using - but with such a small jet at low Re numbers, I'd make sure the wing has a good amount of washout. Slow it down enough and even a cropped delta can tip stall, and I imagine this thing's gonna need high aoa for a slow landing.
You are overthinking the launch system. You just need two 10’ poles, bury them 3’ deep so you have 7’ above ground. Tie a long bungee cord to the tips and put a hook below the aircraft slightly forward of the cg. Tie a cord with a metal ring to the bungee and hook the ring to the plane. Walk back 2x the bungee length and you’ll get ample velocity when you release the plane from ground level. The natural angle from the ground to pole tips will give you the right initial trajectory. I launched many aircraft this way
Maybe an FPV system would help, it makes it much more comfortable and easier to fly, especially at high speeds. You also have the telemetry data directly on the OSD. FPV takes a bit of getting used to at first, but over time it gets pretty good.
Nice design. I would consider using rails to guide the launch cart. Maybe standard pvc waste water pipes, cut in half? For the launch "kicker" ramp you could look at designs for skateboarding, trick skiing, etc, to get the transition just right. Also, I think you should think about integrating an emergency parachute, to safe the vehicle.
You did great, I really loved your approach and I'm learning a lot from your videos. I'm still messing around with FC with my fpv scale model. What would have been your routine for the test flight? Was like, test in manual mode, test in stabilized mode slow Speed, then high speed, go thei the log and refine repeat analize again? Hope I'm clear enough.
Shouldnt the air intakes be on the sides or the belly for optimal airflow? Anyway the kingfisher seems like a great plane good luck at beating the record
Ideally yes, but our angle of attack will ~0.5 degrees at top speed so it won’t make a difference - we lose a bit of thrust during takeoff and landing, but it keeps the inlet clear or any debris during takeoff and landing that could otherwise destroy the engine
@@keatonhare7691 This is possible but we think it is likely the tune at this stage. The default ardupilot tune is very conservative with less control authority than the manual mode. According to the logs the top speed in level flight was 116 mph whilst during the dive it reached 110 mph and the pid outputs were never large enough to get close to the target value suggesting it is a tune issue. Aero forces are definitely going to be a bigger issue as we increase speeds further though, that's why we have 20 kg-cm servos on the full-size kingfisher! -Josh
@TheMachInitiative It’s probably too late for the main launch method - but I have found success testing high speed/large scale models by mounting them 2-5ft above a cleanish car(Tesla or other electric vehicle ideal) and accelerating up to flight speed - you can get a fairly good idea of basic systems and test servos/control surfaces/motor etc. This was the basic methodology for testing a lot of sub scale designs at NASA Armstrong for a long time
As an outsider looking in would it make sense to start out with a lighter airframe then slowly ramp up the wing cube loading? Anyways love the project and am excited to see how it goes and progresses
Yeah, in the end, a jet engine will be the most heavy part. So start with a EDF . A sturdy airframe is always a good idea. At the beginning for the crash, at the end for the forces mid air.
We actually flew some very light foam versions of the Kingfisher before building the 3DP - you can see one of them in this video: th-cam.com/video/KvIqV9kuFTc/w-d-xo.htmlsi=wol6HZ4WIw3eenIo
While this was just the test version and not at full scale, it looked like you were losing quite a bit of speed during turns which is somewhat expected of a delta wing aircraft as they can turn very hard, consequently bleeding speed. For one, you should practice how to deal with a stall with your small scale version so that if anything does go wrong with the final product you can recover easily. (At decent altitude, stall recovery is usually just forcing the nose down until you have enough speed to get level again or to glide to the ground if your engine is gone) Secondly by practicing these stall recoveries you can identify the minimum safe altitude that you should be flying at in which you can recover the aircraft from stall without catastrophic damage. You can also set limits to maximum extension of control surfaces at certain speeds to avoid kicking into a stall Another thing that you can do for takeoff however is by using a dolly that doesnt use a catapult, this means that you takeoff only when you have reached the sufficient speeds and the aircraft doesnt wobble after takeoff like it does with the catapault. The downside of this however is needing a longer runway, and roads dont tend to be the best runways for small scale aircraft due to the uneven surfaces and gravel and dirt. There are some rc clubs that have runways for their aircraft, which might be worth looking for. If such a runway isnt available, a more reliable method of catapault launching would be to use a rail guided platform, allowing for a straight line takeoff with minimal wobbling.
Hi I also live in the uk and just wondering couldn't you make an electromagnetic rail system to launch the kingfisher or some rocket assisted takeoff to get it up to high speeds?
high pitch angle may disturb top mounted intake? cavitate the fan.. as long as you don't go there then it's not a problem. I guess the forward fuselage is lifting surface? I love your launcher, 'specially the guide rope. as soon as i saw the slingshot, i knew that wouldn't work. from experience, lol. Incline the entire thing and you wouldn't need the kicker.. would be more stabilized launch, rather than chucked into the air, and then fly it out.... I've seen compressed air used, but bungee low tech is perfect... (think V1;).. but deff. incline. what about a shuttle on a rail, or a c-cross section around a pipe, bungee or air driven?
What if you made the wing loading even higher, but gave it a higher than 1 thrust/weight ratio and vectored thrust? You could have it take off vertically. The vertical part of the flight would have to be computer stabilized. Would that still qualify?
Have you considered using a powered sled to launch either rocket powered or maybe just a rc car and using a active decoupling at launch using a launch arm that increases the aoa at launch speed allowing for a smoother launch
Rocket will be too unpredictable and an RC car would need far too much runway. Using bands and string has been a proven method for years. They just need to perfect their sled a bit more. I personally use a rail system with a garage door spring connected to a Kevlar string and a Crossbow mechanism for string release. It gets my belly landing highspeed Turbines up to 60mph in a flash. Sometimes crash in a flash when I F up.
All data is good. You should ask Tridge to confirm your findings. Bring it to the Ardupilot dev conference at Richmond, Yorkshire first week of Sept to have all the Dev team available to help!
INAV is not a high-performance gyro for fixed wing. It is not used in Turbines for that reason. I would go with a Bavarian Cortex Pro that is proven at the speeds you want to achieve. All of my highspeed Turbines have Bavarian Cortex Pro. INAV is fine for FPV or long-range flying but as a performance gyro, I think that is where you are going to suffer most of your failures. Bavarian or Synapse AS3X+ works right out of the box perfectly. Plus with the heading hold and launch assist it will help on the sled takeoffs. I use a rail system with a garage door spring connected to a Kevlar string for launching my high-speed turbines. I would fear using INAV or ardrupilot on my expensive turbines.
Is there a reason you can't just use detachable wings to launch? Basically just shed them after you get stable? Could probably have it detach at a speed with a pretty lightweight coupling.
Planes which break the sound barrier coincidentally also accelerate vertically into the sky. So, please rather use a launch tower. Okay, Concorde, but that plane is big. And the Concorde actually breaks the sound barrier in a climb. It is the justification for the afterburner.
@ but a see vixen was first generation? For a small jet at low altitude to break the sound barrier we need a lot of power more similar to star fighter and English electric lightning. Also: why not fly at altitude like real jets or rockets? Parabola flight. Drop external tank at Mach 0.8 . Ah, I just checked: They gave up on supersonic. So it is Mach 0.8 . I think that some jetliners reach this speed. No need for a delta here, just swept wings. Swept wings are mandatory to keep control at sonic speeds, but very thin straight wings may work at Mach 0.8 . Uh, but it is not elegant. 45° swept wings look so beautiful. Clean without any pods! Actually, if we get to fly at a speed where the final shock sits close to the trailing edge of the wing, forward swept wings are the way to go. If velocity drops in the turns, this may not make that much sense. B-1 was designed for this strange requirement: Fast turns and high low level speed. I am not sure how to achieve this without variable sweep wings.
Just yesterday I stood in front of your poster and was wondering how it's going! That integrator response is painful :( The aircraft looked like it had enough speed to get out of the situation, but if the control surfaces are not reacting, there was just no way :( At 8:20 - the integrator part is increasing the whole time up to the switch to Acro. Does that signify that it tries to correct the nose-up attitude? I am not familiar with flying this style aircraft, but it looked a bit like CoG and CoL were mismatched? Maybe because you had to increase the wingloading?
Oh yeah, let's clarify the basics of a delta plane. For symmetry reason I would love to have an X-wing tail. For take-off and landing the plane is flat (belly), but airborne the tail planes rotate into an x? Anyway, the thing the with the main wing (not an X) is that an angle of attack not on one of the main axes leads to roll. Negative lift inverts this roll. I think that dihedral does not invert there. So the angle of attack needs to be measured. And the elevons need to be actuated to translate one behavior to the other.
Great to hear from you Jens - we’re still committed to the project! The CoG remained infront of the CoL at all times, but the CoL moves forward at high angles of attack which causes the stability to change, and we’re limited with how far the CoG can go forward due to the lack of an elevator. The integrator part was stopping the aircraft from deflecting the elevons up to push out of the dive - standard practice to recover from this hereon will be switching to manual and full stick back
@TheMachInitiative Interesting. When I once tried out paper planes I compared different sweep for the leading edge, but could only see a small advantage without sweep. Or is the long chord compared to the fuselage the problem? So that's why we don't see pure deltas anymore? B-1 has a tail, Eurofighter typhoon has canards. Boom X1 has a tail (looks so much like F-5 ). F16 seems to have only minimal sweep: Only enough to easy transition through Mach 1, but not enough to keep the leading edge subsonic at top-speed.
You should use the flight controller to fly for you, that way it can keep a straight and level trajectory for the speedrun and then come back and land on its own. You can even use it to maintain a set speed while you lineup to land manually.
how did you do the animation in blender? I am building modellrockets with a few friends, and we plan on putting a logging device on there with GPS and stuff, but i don't realy know how to smooth the data and eliminate the tolerances by the GPS...
Mothership. Why build a steerable cart to take off from when what you really want is a big low-wing-loading airframe that takes off slow and gets altitude and speed before letting your orange plane go and do its thing...
Alright dudes, I setup and fly a lot of model aircraft with custom tuned gyro systems. I also know who the test pilot was (not naming names) I’m quite suprised some of your issues here weren’t picked up in the pre flight?
You seem to be reinventing the wheel about the launch system. We (in the EDF realm) pretty much perfected that system 15ish years ago. Are you making the mistake of trying to do advanced flying skills before mastering basic control? Are you aware of the speeds being attained by dynamic soaring?
The challenge is scalability - this launcher also needs to work on a 10kg jet-powered model aircraft. Additionally our stall speed is a lot higher than a typical EDF jet, so we need to launch to >25m/s. Admittedly flying in acro was a mistake - we will only be testing in manual and stabilised until we have the gains tuned from now on. We are well aware of the dynamic soaring record, we’re trying to break the jet-powered model aircraft record (465 mph) if we break the dynamic soaring record (564 mph) that’ll be a bonus!
Perhaps there is some way, most likely in ArduPilot, that can alter control authority depending on airspeed. This likely won't work above a certain speed with how it's currently configured (since GPS prevents usage above 300mph I believe) however a pressure sensor like on actual airplanes could be used. It's fairly simple to math out as it's linear up until mach 1, so as soon as you see strange results at speed, you'd know you've hit the sound barrier. It'd probably be inverse square for how the control authority would be mapped, where you've got 100% initially, then the firmware would limit it exponentially as the speed increases. Especially helpful for preventing roll, but you'd probably be better off with a different set of math for the pitch authority, and even a potential 'emergency' mode toggle which would disable the limits and allow full authority to the user. The way I'd see it, this would likely feel very natural to learn, as as the speed increases, the amount the aircraft reacts would be the same, providing it is tuned correctly. The only difference would be distance travelled to rotate X degrees, instead of rate of rotation increasing with speed. It'd be helpful to the pilot as well as it'd feel "slow" to control, while being fully predictable, instead of having everything happen before any human could possibly react.
I have a question my friend is working on his masters thesis and the topic is around PEMFC where we are looking for the flow happens inside the leaves. So if you guys have an idea which simulation software could be better for this? we are currently working with Genesis-World AI.
No way you can make a simple RC car launch platform that needs to go 40+ mph for grass. You would need to use a 1/5 scale RC car with a clamping and release mechanism to keep the plane stabile on take-off roll. This is a huge project itself. Stick to bungee launches.
Several reasons: (1) Time to design/ build (2) The need to break it into sections to fit in the car - we’d have to break it into 10 sections and have some kind of locking mechanism to ensure each part of the guide rail is aligned. A rope can be folded up so doesn’t have this issue. (3) Assembly time at the field
Not true - we both have different areas of expertise and we’ve both put in an equal amount of time in the evenings and weekends alongside our full-time engineering jobs. Its been a team effort!
So much wrong with this approach I dont even know where to get started from an engineering perspective having worked on the real thing. Still interesting to see your learning curve, I hope that eventually youll get there. Keep developing guys 🤞
Share the flight logs please and let us help you. Without having them, you were out of balance. Please please please for gods sake, set a 1/4 mile dia circle and just let it fly to get data. Pilot input and manoeuvring obscure what the plane is doing.
Thank you for deleting my comment, guys. I was trying to help you with a bit of advice from the real world, but you´re obviously focused on reinventing things in your own way. Good luck with that.
We haven’t deleted any comments - we’re keen to receive criticism, it’s crucial to reaching a watertight design. Please post your comment and we’d be keen to listen to your feedback - Dan
@TheMachInitiative Strange, my comment was there last night... sorry about the delete assumption. Well I was trying to tell you guys that I love your project and your progress this far. But I believe that your current launch setup is way too complicated and unreliable. Especially when you´re going to launch your heavy turbine model. I suggest that you take a look at my inexpensive catapult/bungee setup which is incredibly reliable and will fit in a standard car: th-cam.com/video/ASX42mwIAyA/w-d-xo.html I use the same basic setup to launch my 2m long 5kg 3D printed Saab Draken: th-cam.com/video/Jv5X8N2ZG7c/w-d-xo.html
No worries - thanks for sending these videos and for your support of the project. I remember watching the first video a while back. Looks like a really simple and portable design, my biggest concern is the launch speed. Our 10kg aircraft needs to get to >25 m/s (50 mph) on the launcher which requires a fairly large launch distance. Your saab draken will have a much lower stall speed than Kingfisher and only just clears the ground. That being said with a slightly longer rail and stiffer bungee we could probs get a much higher launch speed. Thanks for the suggestions!
@TheMachInitiative Well, the launch speed is determined by the bungee setup and the weight of the model. Don´t judge too much on the Draken video. It was the first time we launched such a big model and the first time flying a turbine powered model. I was scared stiff standing behind the really loud turbine exhaust and only dared to add about 60% throttle. And the 4 bungees where not fully extended. Since then I´ve had a few more flights with more throttle and more bungee pull. Those launches looked more like the F-84f launch in the catapult video. For a model like your Kingfisher I would arrange the bungees in such a way (more bungees in parallel) that the model would launch with a short powerful pull in order to quickly get it up to flying speed. This way you don´t need a long ramp. You just need to point the model in the right direction. It´s a bit like those Zero-length-launch system from the 50s: th-cam.com/video/EZ_w5K82lu4/w-d-xo.html I would suggest that you test a setup similar to mine. Just two rails and a launch pedal. You don´t need the foam padding for a test like this. But pay attention to the bit about the gasoline rubber tubing. It shouldn´t take you guys more than an hour to build one. If you decide to proceed with catapult launching, let me know, and I´ll help you with the more powerful setup. Good luck 🙂
Thanks for watching this video! We now have a reliable launcher, and we've got the flight computer working. We've got more flight time planned in February to do some more tuning flights and get some landing practice. Once we've proven we can do launch, stabilised cruise and landing reliably, we will then attach a stronger bungee to the launcher and maiden the Kingfisher.
We're sorry that the timescales have been longer than previously promised - we want to make sure we have everything ironed out at the sub-scale to minimise risk to the main flight. This is from both a safety perspective and aircraft perspective - repairing the 3DP is orders of magnitude cheaper and easier than repairing the main Kingfisher. We're constrained to flying on weekends with good weather, since we work full-time engineering jobs in the week.
Some of you may also be disappointed that we didn't drop the aircraft out of a helicopter, but we think a catapult is a lot easier to repeat for the full-scale. We also cannot afford a helicopter! Overall, this sub-scale flight campaign is about replicating the real flight as closely as possible, that's why we've matched the wing cube loading to Kingfisher, made the 3DP way heavier than it needs to be, and used a slightly overkill bungee system!
Should also add, Chris was unavailable for this flight so we had a different pilot.
Are you gonna be able to get it to the recovery the drone also. I’ve been talking about this happening on a commercial level to save weight and complexity of electric aircraft.
New sub! You guys are an inspiration to what I want to do with RC Boats!
the length of the video and the title are perfectly fine together
Love the enthusiasm!! I hope humanity overall can recapture your sense of wonder and awe in trying to solve this problem! Thanks for the vid!
Wow! So cool seeing that thing fly!
Thanks!! Would be great to catchup soon
@ deffo!
Great video and editing guys. Keep it up, looking forward to more
RC flight is not my world, but I do fly small airplanes. One thought that came to me was about the use of an AOA (angle of attack) sensor to feed into the autopilot to provide stall warnings. I suspect that this would require watchmaking skills to create one small enough to be practical. It is fun to think about it. Kudos on a great project, I look forward to future installments!
That Kingfisher design is a beauty! Really liked the animation showing the crash sequence. Great video, looking forward to more flight testing! (My favorite part of any aviation project!)
This is a fantastic project! I wish you all a lot of luck with your next design. Looking forward to seeing the Mk 2 fly.
Nicely presented video and an interesting concept. Keep going! ;-)
I wonder if you're going about the use of the flight control system backwards. I would recommend starting your flight out without flight control to get your trims, throws, and expo sorted out so that you can fly it "easily" up to 100mph. Only then would I turn on the flight control to help keep things stable for max speed runs, where a human couldn't possibly react fast enough to maintain proper flight envelope. Unless your idea of breaking the speed record is taking it up really high, then accelerating straight down and your top speed is what's recorded a millisecond before it hits the ground.
I think you're absolutely right that we need to get the trim, throws and expo (the basics) sorted first. In order to sort these out we've needed flight time and our major blocker has been the launch method.
This video represents our process of sorting this out so now we should be able to get repeatable flights to sort these control issues out over the coming flight tests.
The flight launched in manual and we were checking the different modes to work out what needs to change. The rest of the flights would've been adjusting the trim, c.g. and then tuning just like you are suggesting.
Thank you for the feedback we really appreciate it!
-Josh
If they fly high enough and accelerate fast enough, I think they could reach top speed in time to pull out of the dive but it probably wouldn't count anyway.
@TheMachInitiative
Personally I think your project management is pretty LEGIT lads!!!!!
Clearly your college did its job properly.
Eager to see what your next phase brings.
Genuinely sending as many positive vibes as possible from up here in the Norf (west yorks).
There are people flying rc glider using dynamic soaring, already going faster than there projected speed.
Gread vid; enjoying watching the progress! Strong work team!
Excellent work! As far as those program adjustments, I can't help.
I can say the autopilot compensated for the cg being too far back. Switching to acro mode would definitely cause a crash if one was not prepared.
However, stabilized mode will work far better if you tune the aircraft to fly in acro mode. Considering the flight speed and flying without stabilization, exponential will likely be necessary for easy handling in the air.
Like other said (maybe), why don't you use a launch rail? Like a mix of a catapult launcher as found on aircraft carriers and what student rocket teams use?
I believe you you see greatest improvements by angling the entire runway instead of having a ramp at the end, because that sudden change in direction puts unnecessary stress on your aeroframe and may even cause flow separation right when you need the flow over the wings the most.
I did a launch system for my glider using a 1:2 pulley mechanism hat worked great. Look into how sailplanes are being towed and unhook automatically. It would help y'all a lot
i totally agree, a launch rail with weights seems much more consistant and controllable than your current setup, the pros seem very many: adjustability (if you use pulleys and weights), scalability, less stress on the airframe, less setup and the math is basic kinematics. plus you were kind of halfway there already by guiding the cart with a taut line
We agree a launch rail would be ideal, but there is a logistics challenge with design complexity, setup and transport. We’d need to break it into sections to fit in the car. Having 10 plywood boards and a guide wire is very easy to setup at the field. If we made an >10m long inclined rail, we’d need to have locking mechanisms to ensure all parts of the rail are perfectly aligned, we’d need to have a support structure if inclined along the full length, etc
Please can you send photos of your pulley mechanism - really interested to see it!
@ Awesome to hear from you! Sorry, I did the pulley thing 12 years ago when I was 12 years old just messing around. I was literally a cord anchored to the ground with a long stick (like in your videos) and a handle with a roller in the middle of the cord. at the end, the glider was mounted with a hook and a lop in the cord. For every X you pulled on the handle, the plane moved 2X forward. That was enough for my needs. If I were you, I would just put a BLDC on a winch with a cord, do the launch rail thing and pull the sled like so. You could make a cheap launch rail out of 20x20 alu profiles, that are easily stowed in a small car. Setting up may take a little longer, but the advantages far outweigh the increased logistical complexity. I'm rooting for you!
@TheMachInitiative You could go for a rocket assisted take-off on a launch rail. However, a 1m rail would require your airframe to be able to handle 24g and you'd need a model rocket motor with that amount of TWR, so probably a 4m long rail with two 2m sections could barely work. But given the small loads, aluminium extrusion profiles would be enough, with a 1m tall bipod at the end. With lockable hinges you could even make it foldable and get the set-up time down to a minute.
06:44 The front fell off 💀
Haha!!
Loving it ❤
While I'm sure you lads have excellent vision (by way of corrective lenses in one case lol), tracking something moving at 761mph will not be easy. Especially something the size of a carry-on bag. Would you be inclined or open minded to including a simple FPV setup? You've got everything you need for on-screen telemetry with the Matek FC and GPS, all you need is a camera and VTx. I'd be happy to sponsor a setup for you, and this epic project.
Really enjoyed this video. Awesome effort!
really cool video, enjoying all the updates
This popped up on my feed randomly. Well, now I have to see how this shakes out. *clicks subscribe*
For launching you could piggyback on a larger plane, that way you don't need a ramp and would have the advantage of height.
I used to do it with my old zagi on a friends virtually indestructible plane using Velcro, to separate I would pull up and he would dive, it worked great 95% of the time.
Guys i nice job, great idea for you inspirations!
great progress. congrats !
Amazing work really impressive!!
You could stick a motor and pusher prop out the back for these early prototypes. Launching would be much easier and much longer flight times for tuning, then swap in a EDF when everything is tuned
Looking forward to the landing video.
Great work guys!
Rapid prototyping all the way!. Get crazy with it!
Excellent progress guys! Maybe ACRO mode was a bit ambitious, but you don't know until you try. Looks like you've got the roll damped nicely though, and a sweet airframe!
Thank you! Going forward, we’ll stick to stabilised until we have the tuning locked in
@TheMachInitiativewhy Acro at all? I mean I understand it’s fun 😊 but it won’t help with speed will it?
I second the drop launch idea. Surely rhe airframe and control surfaces want to be optimised for speeds much faster than youd be able to take off and (probably more importantly) land at on grass strips. A drop launch from a bigger, slower RC platform would surely allow you to hit takeoff speeds suitable for a sleeker wing and smaller control surfaces. Maybe a 'stall and chute' type system for a controlled recovery?
Living hinges might be doable with a printed design as small as this. I'd reduce the elevon depth by a couple millimeters and move them farther out - remove the control surface at the wing root and instead extend it to the wing tip. Basically like a Mirage 2000.
Couldn't find anything about this model on your website - so no idea what airfoil you're using - but with such a small jet at low Re numbers, I'd make sure the wing has a good amount of washout.
Slow it down enough and even a cropped delta can tip stall, and I imagine this thing's gonna need high aoa for a slow landing.
Maybe try a ballistic parachute recovery system.
You are overthinking the launch system. You just need two 10’ poles, bury them 3’ deep so you have 7’ above ground. Tie a long bungee cord to the tips and put a hook below the aircraft slightly forward of the cg. Tie a cord with a metal ring to the bungee and hook the ring to the plane. Walk back 2x the bungee length and you’ll get ample velocity when you release the plane from ground level. The natural angle from the ground to pole tips will give you the right initial trajectory. I launched many aircraft this way
Keep up the good work, guys
You should mount all your sensors on dampening foam to help clean up the signal and prevent high speed oscillation.
Agreed - we have some rubber spacers between the threads and matek - for the full size Kingfisher the Pixhawk is mounted on foam pads
Maybe an FPV system would help, it makes it much more comfortable and easier to fly, especially at high speeds. You also have the telemetry data directly on the OSD. FPV takes a bit of getting used to at first, but over time it gets pretty good.
It’s so cool even if it crashed!
This is what life is all about.
Suggestion - parachute recovery system. Adds weight but improves survivability and rapid reuse
Great update
Nice design.
I would consider using rails to guide the launch cart. Maybe standard pvc waste water pipes, cut in half?
For the launch "kicker" ramp you could look at designs for skateboarding, trick skiing, etc, to get the transition just right.
Also, I think you should think about integrating an emergency parachute, to safe the vehicle.
This thing is so sick.
What was your top speed?
You gotta make this thing fpv somehow, it would be that fun to fly POV.
You did great, I really loved your approach and I'm learning a lot from your videos. I'm still messing around with FC with my fpv scale model. What would have been your routine for the test flight? Was like, test in manual mode, test in stabilized mode slow Speed, then high speed, go thei the log and refine repeat analize again?
Hope I'm clear enough.
Shouldnt the air intakes be on the sides or the belly for optimal airflow? Anyway the kingfisher seems like a great plane good luck at beating the record
Ideally yes, but our angle of attack will ~0.5 degrees at top speed so it won’t make a difference - we lose a bit of thrust during takeoff and landing, but it keeps the inlet clear or any debris during takeoff and landing that could otherwise destroy the engine
Is it possible that the speed of the final dive overpowered the servos? aero forces on those big control surfaces may have limited the throw.
@@keatonhare7691 This is possible but we think it is likely the tune at this stage. The default ardupilot tune is very conservative with less control authority than the manual mode. According to the logs the top speed in level flight was 116 mph whilst during the dive it reached 110 mph and the pid outputs were never large enough to get close to the target value suggesting it is a tune issue. Aero forces are definitely going to be a bigger issue as we increase speeds further though, that's why we have 20 kg-cm servos on the full-size kingfisher!
-Josh
@TheMachInitiative It’s probably too late for the main launch method - but I have found success testing high speed/large scale models by mounting them 2-5ft above a cleanish car(Tesla or other electric vehicle ideal) and accelerating up to flight speed - you can get a fairly good idea of basic systems and test servos/control surfaces/motor etc. This was the basic methodology for testing a lot of sub scale designs at NASA Armstrong for a long time
As an outsider looking in would it make sense to start out with a lighter airframe then slowly ramp up the wing cube loading?
Anyways love the project and am excited to see how it goes and progresses
Yeah, in the end, a jet engine will be the most heavy part. So start with a EDF . A sturdy airframe is always a good idea. At the beginning for the crash, at the end for the forces mid air.
We actually flew some very light foam versions of the Kingfisher before building the 3DP - you can see one of them in this video: th-cam.com/video/KvIqV9kuFTc/w-d-xo.htmlsi=wol6HZ4WIw3eenIo
While this was just the test version and not at full scale, it looked like you were losing quite a bit of speed during turns which is somewhat expected of a delta wing aircraft as they can turn very hard, consequently bleeding speed.
For one, you should practice how to deal with a stall with your small scale version so that if anything does go wrong with the final product you can recover easily. (At decent altitude, stall recovery is usually just forcing the nose down until you have enough speed to get level again or to glide to the ground if your engine is gone)
Secondly by practicing these stall recoveries you can identify the minimum safe altitude that you should be flying at in which you can recover the aircraft from stall without catastrophic damage.
You can also set limits to maximum extension of control surfaces at certain speeds to avoid kicking into a stall
Another thing that you can do for takeoff however is by using a dolly that doesnt use a catapult, this means that you takeoff only when you have reached the sufficient speeds and the aircraft doesnt wobble after takeoff like it does with the catapault. The downside of this however is needing a longer runway, and roads dont tend to be the best runways for small scale aircraft due to the uneven surfaces and gravel and dirt. There are some rc clubs that have runways for their aircraft, which might be worth looking for.
If such a runway isnt available, a more reliable method of catapault launching would be to use a rail guided platform, allowing for a straight line takeoff with minimal wobbling.
In french we call this : Bricollage 🛫😎
Have you thought of a launch ramp? Like the V-1 ramp or the Komet-163 type landing with a skid.
You should name you company Sykes-Martin, a subsidiary of Lockheed-Martin ! LOL ! Great job guys ! The future is all yours !
Guys which cad programs do you use to build your project
Is windows good for your field or mac
We use Fusion 360 - its compatible with both but windows offers better support for cad packages in general
Hi I also live in the uk and just wondering couldn't you make an electromagnetic rail system to launch the kingfisher or some rocket assisted takeoff to get it up to high speeds?
high pitch angle may disturb top mounted intake? cavitate the fan..
as long as you don't go there then it's not a problem.
I guess the forward fuselage is lifting surface?
I love your launcher, 'specially the guide rope. as soon as i saw the slingshot, i knew that wouldn't work. from experience, lol. Incline the entire thing and you wouldn't need the kicker.. would be more stabilized launch, rather than chucked into the air, and then fly it out.... I've seen compressed air used, but bungee low tech is perfect... (think V1;).. but deff. incline. what about a shuttle on a rail, or a c-cross section around a pipe, bungee or air driven?
What if you made the wing loading even higher, but gave it a higher than 1 thrust/weight ratio and vectored thrust? You could have it take off vertically. The vertical part of the flight would have to be computer stabilized. Would that still qualify?
The main aircraft has a thrust to weight ratio of 3!
Well done.hit geek and nerd with one vid.🤓
Have you considered using a powered sled to launch either rocket powered or maybe just a rc car and using a active decoupling at launch using a launch arm that increases the aoa at launch speed allowing for a smoother launch
Rocket will be too unpredictable and an RC car would need far too much runway. Using bands and string has been a proven method for years. They just need to perfect their sled a bit more. I personally use a rail system with a garage door spring connected to a Kevlar string and a Crossbow mechanism for string release. It gets my belly landing highspeed Turbines up to 60mph in a flash. Sometimes crash in a flash when I F up.
All data is good. You should ask Tridge to confirm your findings. Bring it to the Ardupilot dev conference at Richmond, Yorkshire first week of Sept to have all the Dev team available to help!
INAV is not a high-performance gyro for fixed wing. It is not used in Turbines for that reason. I would go with a Bavarian Cortex Pro that is proven at the speeds you want to achieve. All of my highspeed Turbines have Bavarian Cortex Pro. INAV is fine for FPV or long-range flying but as a performance gyro, I think that is where you are going to suffer most of your failures. Bavarian or Synapse AS3X+ works right out of the box perfectly. Plus with the heading hold and launch assist it will help on the sled takeoffs. I use a rail system with a garage door spring connected to a Kevlar string for launching my high-speed turbines. I would fear using INAV or ardrupilot on my expensive turbines.
They're not using iNav, they're using Ardupilot which is perfectly up for the job.
Tarps are used as RC runways everywhere. You just need to stretch it 😂
Is there a reason you can't just use detachable wings to launch? Basically just shed them after you get stable? Could probably have it detach at a speed with a pretty lightweight coupling.
Planes which break the sound barrier coincidentally also accelerate vertically into the sky. So, please rather use a launch tower. Okay, Concorde, but that plane is big. And the Concorde actually breaks the sound barrier in a climb. It is the justification for the afterburner.
@ArneChristianRosenfeldt sorry what do you mean? A sea vixen would actually only break the sound barrier in a dive IIRC
@ but a see vixen was first generation? For a small jet at low altitude to break the sound barrier we need a lot of power more similar to star fighter and English electric lightning.
Also: why not fly at altitude like real jets or rockets? Parabola flight. Drop external tank at Mach 0.8 .
Ah, I just checked: They gave up on supersonic. So it is Mach 0.8 . I think that some jetliners reach this speed. No need for a delta here, just swept wings. Swept wings are mandatory to keep control at sonic speeds, but very thin straight wings may work at Mach 0.8 . Uh, but it is not elegant. 45° swept wings look so beautiful. Clean without any pods! Actually, if we get to fly at a speed where the final shock sits close to the trailing edge of the wing, forward swept wings are the way to go. If velocity drops in the turns, this may not make that much sense. B-1 was designed for this strange requirement: Fast turns and high low level speed. I am not sure how to achieve this without variable sweep wings.
I challenge you to do a circuit version of kingfisher! it is fast but very agile
Just yesterday I stood in front of your poster and was wondering how it's going! That integrator response is painful :( The aircraft looked like it had enough speed to get out of the situation, but if the control surfaces are not reacting, there was just no way :( At 8:20 - the integrator part is increasing the whole time up to the switch to Acro. Does that signify that it tries to correct the nose-up attitude? I am not familiar with flying this style aircraft, but it looked a bit like CoG and CoL were mismatched? Maybe because you had to increase the wingloading?
Also, crash or not, great progress! That flight data has already proven essential in the analysis.
Oh yeah, let's clarify the basics of a delta plane. For symmetry reason I would love to have an X-wing tail. For take-off and landing the plane is flat (belly), but airborne the tail planes rotate into an x? Anyway, the thing the with the main wing (not an X) is that an angle of attack not on one of the main axes leads to roll. Negative lift inverts this roll. I think that dihedral does not invert there. So the angle of attack needs to be measured. And the elevons need to be actuated to translate one behavior to the other.
Great to hear from you Jens - we’re still committed to the project! The CoG remained infront of the CoL at all times, but the CoL moves forward at high angles of attack which causes the stability to change, and we’re limited with how far the CoG can go forward due to the lack of an elevator. The integrator part was stopping the aircraft from deflecting the elevons up to push out of the dive - standard practice to recover from this hereon will be switching to manual and full stick back
@TheMachInitiative Interesting. When I once tried out paper planes I compared different sweep for the leading edge, but could only see a small advantage without sweep. Or is the long chord compared to the fuselage the problem? So that's why we don't see pure deltas anymore? B-1 has a tail, Eurofighter typhoon has canards. Boom X1 has a tail (looks so much like F-5 ). F16 seems to have only minimal sweep: Only enough to easy transition through Mach 1, but not enough to keep the leading edge subsonic at top-speed.
@TheMachInitiative ah yes, that makes sense! Thanks for explaining and good luck for the next steps!
You should use the flight controller to fly for you, that way it can keep a straight and level trajectory for the speedrun and then come back and land on its own. You can even use it to maintain a set speed while you lineup to land manually.
_"We clearly need a bit more practise with the landing"_
--- comment by officials during the post-crash analysis
how did you do the animation in blender? I am building modellrockets with a few friends, and we plan on putting a logging device on there with GPS and stuff, but i don't realy know how to smooth the data and eliminate the tolerances by the GPS...
Have you guys ever thought about using an oblique wing to achieve higher speeds in the air and slower speeds while taking off/ landing?
Mothership. Why build a steerable cart to take off from when what you really want is a big low-wing-loading airframe that takes off slow and gets altitude and speed before letting your orange plane go and do its thing...
Gud vid 💯💥
Alright dudes, I setup and fly a lot of model aircraft with custom tuned gyro systems. I also know who the test pilot was (not naming names) I’m quite suprised some of your issues here weren’t picked up in the pre flight?
You seem to be reinventing the wheel about the launch system. We (in the EDF realm) pretty much perfected that system 15ish years ago. Are you making the mistake of trying to do advanced flying skills before mastering basic control? Are you aware of the speeds being attained by dynamic soaring?
The challenge is scalability - this launcher also needs to work on a 10kg jet-powered model aircraft. Additionally our stall speed is a lot higher than a typical EDF jet, so we need to launch to >25m/s.
Admittedly flying in acro was a mistake - we will only be testing in manual and stabilised until we have the gains tuned from now on. We are well aware of the dynamic soaring record, we’re trying to break the jet-powered model aircraft record (465 mph) if we break the dynamic soaring record (564 mph) that’ll be a bonus!
La sonrisa de Dan Sykes da mala impresión. Aparenta ser un asesino en serie. 😂😅😢
Why you aren't using a rlaunch ramp like all other model pilots?
Perhaps there is some way, most likely in ArduPilot, that can alter control authority depending on airspeed. This likely won't work above a certain speed with how it's currently configured (since GPS prevents usage above 300mph I believe) however a pressure sensor like on actual airplanes could be used. It's fairly simple to math out as it's linear up until mach 1, so as soon as you see strange results at speed, you'd know you've hit the sound barrier.
It'd probably be inverse square for how the control authority would be mapped, where you've got 100% initially, then the firmware would limit it exponentially as the speed increases. Especially helpful for preventing roll, but you'd probably be better off with a different set of math for the pitch authority, and even a potential 'emergency' mode toggle which would disable the limits and allow full authority to the user.
The way I'd see it, this would likely feel very natural to learn, as as the speed increases, the amount the aircraft reacts would be the same, providing it is tuned correctly. The only difference would be distance travelled to rotate X degrees, instead of rate of rotation increasing with speed. It'd be helpful to the pilot as well as it'd feel "slow" to control, while being fully predictable, instead of having everything happen before any human could possibly react.
That should be possible via the LUA scripting functionality.
Engine seemed to be way underpowered for the small wings. Whole flight the aircraft had quite a high attack angle.
This is just the electric prototype - the main aircraft has 15 times more thrust, in a planform that is only 60% bigger
@TheMachInitiative I ment thats maybe part of the reason for the crash as well
I have a question my friend is working on his masters thesis and the topic is around PEMFC where we are looking for the flow happens inside the leaves. So if you guys have an idea which simulation software could be better for this? we are currently working with Genesis-World AI.
Can we have the STL files?
No way you can make a simple RC car launch platform that needs to go 40+ mph for grass. You would need to use a 1/5 scale RC car with a clamping and release mechanism to keep the plane stabile on take-off roll. This is a huge project itself. Stick to bungee launches.
Have you considered dropping it from another plane/weather balloon/drone? Gravity should help accelerate it!
Hy don't yo use a launch rail? A wire steems less reliable
Several reasons:
(1) Time to design/ build
(2) The need to break it into sections to fit in the car - we’d have to break it into 10 sections and have some kind of locking mechanism to ensure each part of the guide rail is aligned. A rope can be folded up so doesn’t have this issue.
(3) Assembly time at the field
That puny hammer is for carpenters.
Looks like Josh is a puppet and Dan is the ventriloquist.🤔🤔🤔
Not true - we both have different areas of expertise and we’ve both put in an equal amount of time in the evenings and weekends alongside our full-time engineering jobs. Its been a team effort!
Is speed a.key. it might be a rd that drags foolish enrgunaway from gliding a real treat
Hello. Is this vehicle 70mm edf compatible ? If so, i have a few questions.
It uses a 70mm EDF
NGL thought that was a cardboard cutout at the start of the video
Why does the short haired guy's lips move with what the long haired guys is saying? 🤔
They are both looking at the same script; one person reads out loud, the other's brain is reading it with him, causing involuntary lip movement.
@@DerekWalsh-l4i 😂 ...
So much wrong with this approach I dont even know where to get started from an engineering perspective having worked on the real thing. Still interesting to see your learning curve, I hope that eventually youll get there. Keep developing guys 🤞
What would you do differently?
Why not airlaunch from another plane? In that way you can optimize for high speed instead of takeoff or landing
Measure the engine speed a wind tunnel and measure the air over the wings if these are installed way non argument it should fly
Emergency parachute
maybe add a rocket motor for when its up to speed =)
Rockets suck compared to jet engines. Yeah, need a little money for the latter . But do you know a successful rocket plane?
570mph 😂😂😂
Dreamz are free I suppose 😊
Looked tail heavy
the most fastest... he he
Share the flight logs please and let us help you.
Without having them, you were out of balance.
Please please please for gods sake, set a 1/4 mile dia circle and just let it fly to get data.
Pilot input and manoeuvring obscure what the plane is doing.
Thank you for deleting my comment, guys. I was trying to help you with a bit of advice from the real world, but you´re obviously focused on reinventing things in your own way. Good luck with that.
We haven’t deleted any comments - we’re keen to receive criticism, it’s crucial to reaching a watertight design. Please post your comment and we’d be keen to listen to your feedback
- Dan
@TheMachInitiative Strange, my comment was there last night... sorry about the delete assumption.
Well I was trying to tell you guys that I love your project and your progress this far. But I believe that your current launch setup is way too complicated and unreliable. Especially when you´re going to launch your heavy turbine model.
I suggest that you take a look at my inexpensive catapult/bungee setup which is incredibly reliable and will fit in a standard car: th-cam.com/video/ASX42mwIAyA/w-d-xo.html
I use the same basic setup to launch my 2m long 5kg 3D printed Saab Draken: th-cam.com/video/Jv5X8N2ZG7c/w-d-xo.html
No worries - thanks for sending these videos and for your support of the project. I remember watching the first video a while back. Looks like a really simple and portable design, my biggest concern is the launch speed. Our 10kg aircraft needs to get to >25 m/s (50 mph) on the launcher which requires a fairly large launch distance. Your saab draken will have a much lower stall speed than Kingfisher and only just clears the ground. That being said with a slightly longer rail and stiffer bungee we could probs get a much higher launch speed. Thanks for the suggestions!
@TheMachInitiative Well, the launch speed is determined by the bungee setup and the weight of the model. Don´t judge too much on the Draken video. It was the first time we launched such a big model and the first time flying a turbine powered model. I was scared stiff standing behind the really loud turbine exhaust and only dared to add about 60% throttle. And the 4 bungees where not fully extended. Since then I´ve had a few more flights with more throttle and more bungee pull. Those launches looked more like the F-84f launch in the catapult video.
For a model like your Kingfisher I would arrange the bungees in such a way (more bungees in parallel) that the model would launch with a short powerful pull in order to quickly get it up to flying speed. This way you don´t need a long ramp. You just need to point the model in the right direction. It´s a bit like those Zero-length-launch system from the 50s: th-cam.com/video/EZ_w5K82lu4/w-d-xo.html
I would suggest that you test a setup similar to mine. Just two rails and a launch pedal. You don´t need the foam padding for a test like this. But pay attention to the bit about the gasoline rubber tubing. It shouldn´t take you guys more than an hour to build one.
If you decide to proceed with catapult launching, let me know, and I´ll help you with the more powerful setup. Good luck 🙂
maybe small
skates not wheels
Please skip the music
It crashed coz it reached the ground level too fast.
People who don't know how to make videos, always go for the music.
Nice video but the "music" was brutal
wtf is a kingfisher ?
its a bird, a fish hunting bird. Look up on wikipedia.