This may be the most amazing thing we've ever posted here. DIY Drones member Brad Hughey built an electric multicopter capable of carrying a person (him), and then actually tried to fly it in his driveway (without a helmet!). Let's just say it didn't end well. But he's figured out what went wrong and he's going to give it another go.
In an email to me, he explains:
History was indeed made on August 10th, 2011 when the Revelation PoC prototype crashed unceremoniously in my driveway. It did briefly leave contact with the Earth, and one could argue that you have to fly in order to crash, but I do not have the audacity to declare a success out of this debacle. A root cause analysis has determined that multiple Magically Obliterating Smoke and Fire Emitting Transistor (MOSFET) failures are to blame. If you listen real closely, you can hear the power rail line inductance ringing (a bit of electronics levity). I wasn't laughing at the time, but an important lesson is finally learned; MOSFETs fail shorted (full throttle). One failure in the back started the pitch forward, then three in the front failed, catapulting me down the drive perilously close to a parked car, missing a rotor strike by mere inches.
The resolution isn't great due to the use of USB instead of FireWire to copy it off of the camcorder. That said, I'd rather this didn't go "viral", as it is a bit embarrassing. Such is the nature of invention. I proffer it mainly as a veracity enhancer; this effort is real and very close to success.
It is interesting to note that half the array out of ground effect managed to push the whole craft with me in it dragging against the asphalt for almost 20 feet before I managed to shut everything off. The power is certainly there. It's all a matter of control now, and the first thing to do next is make the power MOSFET stage for each thrust unit "bullet-proof".
The damage isn't as bad as it looks. The real work involves a total redesign of the power stage including FUSES for each thrust unit. There are much better MOSFETs around now, considering this iteration is seven years old.
New changes frantically being applied include:
- Higher current and more modern MOSFET devices
- A resistor-capacitor snubber network across every MOSFET to help mitigate ringing overvoltages
- Transient voltage suppressors (zener diode-based technology) across every MOSFET
- A complete rewiring to minimize power rail inductance
- FUSES on each motor as a fail-safe
- Larger decoupling capacitors on the outrigger thrust units
We're a couple weeks away from another run at it.
Yours in Daring Invention Progress,
Brad
Comments
That old saying that helicopters don't fly, the earth simply rejects them sadly does not not seem to work well when it needs to. I have witnessed people cut in half by rotor blades. Safety cannot be taken too seriously. I don't see what was wrong with putting sandbags in the pilots seat and testing it with an RC system.
The Darwin Awards??
In my practical experience I can say for sure that prop. noise goes with inefficiency. Running blades close to stall angles is simply a non starter in that race. That is unless you are planning to have the vertical airspeed of an Apache gunship.
Ducted fans are of no use in this regard unless you are building a compressor. However a bellmouth shroud can augment flow so long as the prop is significantly smaller in dia. This design is about 30% more efficient than an open plan. Not shown here are the forward swept blades which would not work outside of those shrouds and would be far too dangerous. Lower pressure above the hull and higher pressure below contributes to the efficiency
@Jan Detlefsen: Yes, they overlap, and no, quite the contrary - it affords a larger virtual disk area in a more compact space. Absolutely-the production unit will have stagger and overlap. Ducted fans are a aerodynamic fashion statement which barely offers enough tip loss reduction (when done VERY precisely) to make them worth the effort. Even then, simply making the rotor slightly larger achieves better efficiency gains for the extra weight. The "Chinook Effect", named for the CH-46&47 series dual-rotor helicopters is well-known (although obviously never before tried on quite this level), and in particular, this paper inspired me to go with the overlapping design for maximum efficiency:
http://terpconnect.umd.edu/~leishman/Aero/AHS2002_Griffiths.pdf
It's worth reading if you're interested in the aerodynamic justification for the overlapping design. (I guess it's no secret that I am a huge fan of J. Gordon Leishman's work, although I have also studied Stepniewski and Keys, Hayden, and many of the early NACA propeller thrust papers back to the 1920's.)
P.S. The rotor diameter is 1.1m, and the actual inflow velocity at the inner thrust unit disks is, of course, not negative, but the pressure differential effect on the virtual angle of attack is, shall we say, counter-intuitive. :-)
Let's assume that the total hover thrust required is 272Kg, that the distance between each "virtual quad" is 3.66m, and the center of mass is approximately 1m below the virtual disk plain, if that helps anyone with knowledge of how the algorithms in the code work.
@Denny Rowland: In light of the above, the ideal induced power at each disk of the 36 thrust units using basic momentum theory is 442.2 watts. The FM is about .60, but that's OGE. IGE, strange things start to happen. With all due respect, at an average blade element Reynolds number of 250K, low KV and fine pitch would make for a wonderful noise-maker, even OGE. Just for the record, the RPM is approximately 1500, and each blade is about 200 grams. Rotational inertia can thereby be approximated.
The aerodynamic problems here are far beyond what anyone has documented before, and not even Messrs. Cheeseman & Bennett (the pioneers of empirical helicopter ground effect measurements in 1941) explored pressure differentials of this magnitude (for obvious reasons). It is NOT merely thrust vector addition, as I can attest from 7 years of development. What you see in the video somewhat is the unique nature of In this extreme ground effect, where every rotor is struggling with a virtually negative induced flow velocity (due to the rotor next to it gasping for air and a place to put it). Without giving away too much hard-earned intellectual property, operating the rotors near the airfoil stall angle is essential if this is to be accomplished with fixed-pitch blades. That's part of the problem with control I'm experiencing; the minute the rotors near the edge of the virtual disk break free of the induced high pressure "bubble" underneath the array, they surge. In short, the motors are working properly, although I'm intrigued by your "balance beam" assertion. Take this offline?
I forgot to mention. Low KV and Very fine pitch is the way to go!
I think the first step would be to start making some motor prop. tests to see where you are with efficiency. Basically you need to absorb your available power into the largest swept area. Motor KV will tell you what RPM you should be aiming for. PID settings come well after that has been finalised and could be done with a simple balanced beam and a motor at each end. Once you have two motors working properly then move on.
Yes, about that...
Anyone have any suggestions about how to scale up the configuration settings for a much larger than normal application?
-
1
-
2
-
3
-
4
-
5
of 7 Next