Multicopters - fundamental design questions

[SlowSteve]

Hello all.

I'm sorry if this is in the wrong place - I couldn't see a place for it to go.

Some friends and I are planning to design and build a multicopter from the ground up. We're heading down the route of full computer control, rather than R/C control - it's the nerdy software development side that interests us the most.

However, because I've built model planes and heli's before, I have been volenteered to build the quads themselves. I've been looking at the basics of the design - aerodynamics, different layouts, pro's/cons of different layouts etc and been reading a number of papers.

I have some questions that I'm hoping to bounce around if I can and get some responses from people who have already gone through the pain if possible?

1) Nacelles
A lot of the power of a jet comes from the Nacelle design - it proves about 40% additional air capacity. I haven't seen any aero nacelles around blades in most quads - although it would seem it would protect from damage a lot and also improve performance. Has someone looked at this and decided it's not worth it? Are there any papers about this?

<<edit - there are a number of quads with a nacelle design, but these seem to be used for either protection or just looks, as opposed to allowing the props to access a lot more air - i.e for performance reasons.>>

2) Ducted fans - has anyone ever looked at building multirotors with ducted fans, as opposed to regular props? If so, how did it turn out? Pros? Con's?

3) I have read a number of papers about all the benefits of building multi's with a single large prop right in the middle - like a helicoptor main prop - which takes around 90% of the AUW, which then allows the outlying motors to be for direction control. Benefits seem to be much improved aero, 15% better battery life, more agile, much easier to programme against. Downsides are dealing with torque.
Seeing as there seem to be so many benefits regarding this design it's the pattern I'm giving most thought to. However, I can't find anyone who has built this pattern in the real world..... But... I'm not sure what this pattern would be commonly known as either so not sure I am searching for the right keywords. Has anyone looked at these designs? built them? any feedback?

4) Purely from the point of view of programmability, has anyone looked at moving the arms of a multi, rather than changing motor speeds? I.e on a quad, where the arms are spread 90 degrees to each other, has anyone experimented with moving the arms themselves to induce pitch/yaw etc, rather than motor speeds? If so, how did it turn out? Was the extra weight and complexity worth the benefits?

5) Has anyone written any good papers or posted anything to this form about leg design for landing? I've not been able to find out much about them and, at least initially it seems easier to build flexible legs rather than write code to make a safe landing.

6) What are the reasons that motors point upwards on multi's, with puller props? Obviously needed if the multi doesn't have landing gear, but if they do, then having the motors point downwards would seem to make the dynamics of the multi a lot more easy to work with, and there should be a slight area advantage as the air is fully "clean" and not impinging on the motor spar.

7) When you have your multi pretty much built, and it comes time to balance it - i.e balancing the multi itself, rather than the props, which is done right up front, what is the most common method? Do you move the motors, use ballast or make small changes to the props? My guess is either motors or ballast is prefered, as the prop thrust is a dynamic, rather than a static, but not sure of what the usual approach is?

Many thanks indeed - and apologies in advance for asking a lot of really basic questions.

Many thanks

Steve

 

[sledge57]

I can answer a couple things based on my limited knowledge and what I've seen videos of.

Ducted fans have been tried but ducted fans are not as efficient as props are.

Look up info on Tri-Copters, they have a moveable motor mount on the rear motor to provide yaw control.

 

[Carapau]

Here are my thoughts:

1/ Nacelles have indeed been used on some multis but I think they then become an issue for things like size of the multi (makes it take up much greater volume for transportation). I wonder how much efficiency is gained on these small props v the added weight. Cost might also be another issues as the tolerances with the props are going to have to be tight in order for there to be any benefit of the nacelles.
2/ See above.
3/ if you are going to go to the trouble of what you mention in 3, then why not just stick with the single rotor design? It deals with the torque issue, would be a far more simple design and is very well proven.

4/ This I think would just be too complex and you would add a lot of weight to the design with things like servos and bearings for the arms.

5/ Not sure what you are getting at here really but I don't see what is wrong with the retractable design as per the S800 or the legs fixed to the gimbal as per the Cinestar/Movi. A safe landing has to be a safe landing and the type of legs will not make this change a great deal. Certainly some legs would allow a quicker descent rate but you have to think of all the other equipment taking a large jolt on landing.

6/ One good reason for motors pointing upwards is that it gets them higher up ie out of the lens of a camera if you are into AP. The other is that the prop is seeing a 100% clean feed of air without a boom above it causing turbulance.

7/ This is normally achieved through shifting your batteries about a bit and or the gimbal. Ballast is not a good idea as you are adding dead weight to the aircraft- best off moving parts of the system that are already there.

 

[SlowSteve]

Hi,

Many thanks for the replies so far. Really appreciate the input.

Carapau - seperating ducted fans from nacelles - a well designed nacelle can be brought in pretty light - for 8 inch props to do all four I would guess around 150g - 200g. You can have a fair bit of seperation between the blade tips and the nacalle walls - a few mm is fine - and it's fairly easy to provide around 10% - 15% additional workable air to the props - it has the same effect of using a larger prop, but with the additional benefit of higher speed air near the tips where the most work is do-able. General there are minimal effects to battery life as well - however, they are a pain to make if your making them in carbon. I think unless I struggle for thrust, I'll leave it be.

2) - Thanks to Capapau and Sledge57 on the DF question - at least that saves a lot of money!

3) Single blades - i.e a heli, have a lot of disadvantages - they're not an especuially good mechanical design and are a pain to write code against because the dynamics are really hard to work with.... actually... I'll rephrase that as " I find it a pain to write code against them, because the dynamics is right on the limit of my maths ability". On the flip side, multis seem to be a far more elegant design, and are also far easier to write simpler, more obvious algorithms for. Also - there doesn't seem a huge amount of "new" in the Heli world, where as forums like this show that multi's are still a fast developing field, and seem a lot more interesting.
I hadn't thought about the Foriegn Object Damage issue - that a really good point. I wonder how to balance that risk against the stability improvements from downward facing motors?

5) Regarding legs - I was thinking about outdoor use - the multi might be facing landing in anything from a grass field with divots in it to a carpark on a slope. We're not intending to have radio control, so there is no "Man in the Loop" to guide the multi away from a divot etc. I was thinking that is the multi had movale landing struts, like an olio system on a plane - something nice and simple that can accomodate slopes or slightly rough ground - then it would save a lot of work and/or a lot of crashes..... Definately not looking for a system that can survive a big drop, just a less than perfect landing site.

7) Thanks for the info about the batteries. Are their adjustable battery harnesses available? On the Heli's I've had previous it was just a piece of velcro strap and you could move forwards and backwards, although they LiPos would slip occasionally. With the need to adjust in two dimensions, is there a way of setting them up differently?

 

[kloner]

for something with 8" props to carry another 150-200 grams is too much,,,,, like a second hero onboard a phantom

 

[Carapau]

7) Thanks for the info about the batteries. Are their adjustable battery harnesses available? On the Heli's I've had previous it was just a piece of velcro strap and you could move forwards and backwards, although they LiPos would slip occasionally. With the need to adjust in two dimensions, is there a way of setting them up differently?

That's probably because you didn't also have velcro stuck to the battery and the airframe as well as the straps. This ensures there is no movement. Ever wonder how the 3D heli pilots manage their ridiculous moves without ejecting batteries? Here's what I mean:

 

[Cheshirecat]

Are you planning a completely autonomous ship or something that operates on pre set way points? If the later isn,t that an A2 with an iPAD. Also with nacelles or ducts you would probably end up with increased surface areas subject to cross winds and would have spend some of you efficiency gains correcting for it. Not that there is anything aerodynamic about a multi rotor apart from the props.

 

[SlowSteve]

CheshireCat:
Right now, I'm not too sure - I don't think either are out of the question. We do know we have no plans for AP, FPV, stunt flying etc. We're really interested in the type of stuff that Rafello d'Andrea showed on the TEDtalk - it's what made us think about multi's initially -

https://www.google.com/url?sa=t&rct...v-txZrtIDa76GS_NOrzvH3A&bvm=bv.59378465,d.eW0

The multi is really just a platform to try different techniques and idea's - we've been playing with hexapods for a while and multis are a fair step up in terms of complexity. We're pretty nerdy, and quads and multis looks really interesting from the point of view of algorithms and coding. The idea is to have them fully autonomous, i.e give them a task and have them work out the actions and then execute - without any input from us- for example, fly around the edge of my garden and come back to land on a piece of white card - with no direct input from a human with R/C, control from a computer etc. So the multi will have to take off, adjust it's flight for the environment - wind, tempurature etc, work out where it is, see whats around it, plot a course, identify obstacles ( not too easy ) avoid obstacles ( easy ), avoid obstacles safely - i.e don't crash into a fence while avoiding a tree ( very hard ) and then come back to land and land independently.

If anyone who doesn't know anything about this stuff saw that it would look pretty dull to see, but as grown ups it's frowned upon to play with kids toys so this is a good excuse - it's not designed to achieve a lot - just give us a reason to play with soldering irons, write code

Initial work is going to be on the basics - take off, landing, manovouring, hovering etc. We'll be using the paparazzi framework (http://paparazzi.enac.fr/) as a baseline but it has stuff that we want that isn't there so we'll add it and contibute back into the project. It will be a 2 board system with a homebuilt main control board based around 2x Arm Cortex A7 chips - this will be where the main flight control processing is done - IMU, standard GPS, tempurature etc. We're also planning on a different type of ESC which used both Hall effect as well as back e.m.f to cross check speeds to give us more granularity - these are already built and are on a 2cm x 1cm board but we need to finish the code - target is to be able to get to an accuracy of around 10rpm for the prop speed - we have about 25rpm accuracy now and our "maths guy" is asking for more granularity.

There will be a second board - mostly likely a Raspberry Pi clone - removing the networking ports and changing some things around to consume less amps, but maybe a BeagleBone Black ( as it has a lower power requirement ) where more "hardcore" processing is done - for example using OpenCV to handle computer vision, decison tree processing etc, and it will also act as an offload to the control board if that starts to bog down - for example if the Kalman nests start to get screwy. This will be in Python unless we hit something that is really time sensitive - which I hope we don't as I'm terrible with C. Still thinking about the bus between the two boards - for the hexapods we used I2C followed by CANBUS, but in a multicopter, this may have too much latency to be practical - with a hexapod it's fine just to say "i'm overloaded - don't move for a moment..." - and we may have to go to something else instead. The last purpose for this board is to offload signals from the GPS chips to do Distributed GPS processing - this lets us have an accuracy of about 20cm - 25cm so it's practical to use GPS indoors, as long as the signal quality is decent i.e no thick walls in the way.

A good chunk of this is not new and we're bringing it over from previous geekery with hexapods - the board controls, IMU, buses and ESC's will be new, and a good chunk of the software, but a lot of it is being reused.

After we've flown around the house, or some other challenge, we've no plans on whats next so, as the guy who's putting the hardware together, I'm trying to build something thats adaptable and generic.

The biggest problem I'm personally having at the moment is that our groups "maths guy" - Lester - is a ex NASA trajectory specialist - although he works in IT now - and for him this stuff is easy, but he seems to talk in a different language - everything is expressed as Inertia Tensors and Screws and there seems be a lot of multi-dimensional symmetry involved - hence my questions about how to get very precise static balances and needing to get really granular measurements of the prop speeds. It's a learning experience, but it's a bit like being back in class again. The other issue is that as a group we have a budget of about £400 - $600 USD and so some of what he wants - for example really precise RPM measurements - we just can't buy within the budget so we have to built it ourselves or finagle a way around it. From his NASA days, it was fine to spend hundreds of thousands of dollars on this sort of stuff.

 

[skquad]

That's probably because you didn't also have velcro stuck to the battery and the airframe as well as the straps. This ensures there is no movement. Ever wonder how the 3D heli pilots manage their ridiculous moves without ejecting batteries? Here's what I mean:

That is insane!

 

[PeterLester]

IMHO
I was under the impression that a nacelle is simply a cowling to enable more efficient airflow around a motor.
The MR motors can benefit from a bit of extra airflow for cooling as would ESCs mounted in the prop airflow but that also creates more drag.
Ducted fans have their efficiency benefits but more so for fans aimed horizontally as they do add extra weight.
In the case of VTOL the Basic and Attitude gains will be adversely affected as you add more weight at the end of the motor arms.
The optimum landing gear design for best strength to weight ratio would be a 3 point landing gear similar to what is used on the Cinestar 6 or 8.
The Sikorsky X2 is likely one of the most efficient and stable VTOL designs possible but mechanics are more complex than the Multirotor.
https://images.search.yahoo.com/sea...x?_adv_prop=image&fr=yfp-t-901&va=sikorsky+x2

Hello all.

I'm sorry if this is in the wrong place - I couldn't see a place for it to go.

Some friends and I are planning to design and build a multicopter from the ground up. We're heading down the route of full computer control, rather than R/C control - it's the nerdy software development side that interests us the most.

However, because I've built model planes and heli's before, I have been volenteered to build the quads themselves. I've been looking at the basics of the design - aerodynamics, different layouts, pro's/cons of different layouts etc and been reading a number of papers.

I have some questions that I'm hoping to bounce around if I can and get some responses from people who have already gone through the pain if possible?

1) Nacelles
A lot of the power of a jet comes from the Nacelle design - it proves about 40% additional air capacity. I haven't seen any aero nacelles around blades in most quads - although it would seem it would protect from damage a lot and also improve performance. Has someone looked at this and decided it's not worth it? Are there any papers about this?

<<edit - there are a number of quads with a nacelle design, but these seem to be used for either protection or just looks, as opposed to allowing the props to access a lot more air - i.e for performance reasons.>>

2) Ducted fans - has anyone ever looked at building multirotors with ducted fans, as opposed to regular props? If so, how did it turn out? Pros? Con's?

3) I have read a number of papers about all the benefits of building multi's with a single large prop right in the middle - like a helicoptor main prop - which takes around 90% of the AUW, which then allows the outlying motors to be for direction control. Benefits seem to be much improved aero, 15% better battery life, more agile, much easier to programme against. Downsides are dealing with torque.
Seeing as there seem to be so many benefits regarding this design it's the pattern I'm giving most thought to. However, I can't find anyone who has built this pattern in the real world..... But... I'm not sure what this pattern would be commonly known as either so not sure I am searching for the right keywords. Has anyone looked at these designs? built them? any feedback?

4) Purely from the point of view of programmability, has anyone looked at moving the arms of a multi, rather than changing motor speeds? I.e on a quad, where the arms are spread 90 degrees to each other, has anyone experimented with moving the arms themselves to induce pitch/yaw etc, rather than motor speeds? If so, how did it turn out? Was the extra weight and complexity worth the benefits?

5) Has anyone written any good papers or posted anything to this form about leg design for landing? I've not been able to find out much about them and, at least initially it seems easier to build flexible legs rather than write code to make a safe landing.

6) What are the reasons that motors point upwards on multi's, with puller props? Obviously needed if the multi doesn't have landing gear, but if they do, then having the motors point downwards would seem to make the dynamics of the multi a lot more easy to work with, and there should be a slight area advantage as the air is fully "clean" and not impinging on the motor spar.

7) When you have your multi pretty much built, and it comes time to balance it - i.e balancing the multi itself, rather than the props, which is done right up front, what is the most common method? Do you move the motors, use ballast or make small changes to the props? My guess is either motors or ballast is prefered, as the prop thrust is a dynamic, rather than a static, but not sure of what the usual approach is?

Many thanks indeed - and apologies in advance for asking a lot of really basic questions.

Many thanks

Steve

 

[SlowSteve]

Hi Peter,

Nacelles can take two forms really. If you have a plain tube, you get protection and cooling benefits. However, if you look at the nacelle around a jet engine, you'll see that the front "lip" is an aerofoil shape - it has the effect of pulling in a lot more air for the same, in our case, prop size. With jet engines, the value of this can be up to 40% of total air flow, but with props, and a small model, 15% would be a target and 20% a stretch.

Thanks for the other comments - most useful.

 

[Motopreserve]

That video gave me an anxiety attack!!!!

Steve,

Really intresting stuff here. Always nice to see people thinking outside the box and pushing the boundaries for peaceful purposes.

Admittedly, the only part I even come close to understanding is the potential benefits from increased airflow to the motor. I know with motorcycles, for years manufacturers have flirted with RAM air intake on race/performance bikes to the motor to increase efficiency. More recently I saw a bike put onto a dyno, as well as rigged with sensors to measure performance on a real world track, to measure scientifically the mechanical benefits of this increase in air flow. Turned out that only at a very small section of the RPM range (very close to top/peak performance) was there a noticeable benefit. But it was there.

Your challenge it seems, in regards to this specifically, would be performance gain vs weight gain.