Drone PCB, need advice.

[Willy_Tepes]

I am making this drone of my own design which has a novel control system. It is not quite a helicopter since it has a fixed pitch rotor, but it allows me to reduce the control system to 1 motor and 2 servos. I am a complete novice at designing PCB's and this is my first attempt at anything like this (never flown a drone either). Seeing that help and advice is very hard to get in real life and I knew nothing about electronics, I spent 3-4 years studying this on youtube and am very grateful for people like Dave who have such a great way of explaining stuff. My background is in general mechanical tinkering and I've worked several years as a structural aircraft mechanic in the military.

These are basically carrier boards as I want to market this as a DIY kit that people can assemble themselves. The form factor 30x70mm is chosen to allow the use of a Raspberry Pi zero as the middle board. The drone can be operated without a microcontroller, but an MC allows for the mounting of external sensors like for example gas sensors, distance sensors or microphones for object tracking through I2C. The entire electronics package is modular allowing it to also be used on traditional RC airplanes. I do not have room for any caps on the power supply side of the power distribution PCB, so they will be mounted very close to the battery connector and hidden inside the battery compartment.

I simply want some comments and advice on the PCB layout and how I can improve it.
I've included some images of the drone itself to clarify the physical placement of the boards.

For those interested in drones, this has a 40 amp ESC with 3 amp BEC, a 2316 (980Kv) motor, and runs on a 2s, 3s or 4s battery pack. It weighs about 250-300 grams depending on the battery pack. The propeller is 10.5 inch but this is up to the user to chose. The setup as it is will give me a 3:1 lift to weight ratio. With the right combination of motor, prop and battery you could push this up to a 5:1 lift to weight ratio. The camera has 180 degrees tilt and can fit 19mm and 14 mm FPV cameras. In the standard configuration it has a strobe light on top and red/green navigation lights, but additional lights can be mounted to the battery compartment.

[Willy_Tepes]

[KE5FX]

Interesting project.  You're in for quite a learning curve, to say the least!  How many commercial drones have you taken apart and studied?  That would be my first move. 

The engineering being done at DJI is the most impressive work since the Apollo project, IMO.  I'd start there.  Buy some cheap crashed DJI quadcopters, a set of very small screwdrivers, and a blank engineering notebook. 

Oh, and I think I'd want to get a few flight hours under my belt before tackling this project, too.

[Willy_Tepes]

Experience flying drones is absolutely not required to design one. Aircraft designers very seldom have a pilot's license and remember, I am an aircraft mechanic. I know mechanically what I am doing, I am also a chemical engineer, I can do the math. I am 4 years into this project so I am not "starting". I have taken apart more things than most people have seen. Picking apart electronics is a very important part of learning the field. You see what solutions others have chosen, what components and technology exist, and how they are integrated into the product's mechanical design.

I do not consider the work DJI has done as particularly impressive, they are literally just quadcopters. This is a completely novel control system for any type of aircraft. I considered patenting it, but opted against it for several reasons. A great deal of DJI's development process (software) has been geared towards keeping control of the drones with DJI and not with the customer. If I were to develop this into a regular commercial drone and produce it myself, then it would be wise to look at the solutions they have chosen, like integrating all the electronics. But this is purposely designed to allow people to solder this together themselves using commercially available RC and drone parts. I do not want to produce them myself as I live in a social democratic welfare state and business here is hell. I also wish to avoid the hassle of regulations concerning production, labor, sales and export. As investors inherently mean that you are giving up some of the control over your intellectual property, that is not the way I want to go. That would only end up with them being used as military weapons or by police. I do not like the way technology is being used (RIP Uncle Ted). I will only be selling the design files.

I am also including a perfboard to use instead of the Raspberry Pi Zero or the Tiny 2040, so you can develop your own system in case you have some special sensor you want to develop. That gives you 30x70 mm of 2.54mm spaced double sided perfboard to play with. Or you can just design your own PCB. This drone is powerful enough to carry a 500 gram sensor package externally, though this of course affects the flight characteristics. In it's basic configuration it has enough power to weight ratio to fall into both the Freestyle and Racing drone market. Because this is a new flight control principle, it does have it's own special flight dynamics. It will be fun to explore these. I suspect it can do some neat tricks.

This drone is meant to be spread far and wide for a very cheap price, I am not in this for the money. I want only some bucks per download to compensate me for the time I have spent on development. And I need a few bucks because I am dirt poor and am financing this project on a very small paycheck.

The only thing I need help with is the PCB layout, everything else is under full control.

[KE5FX]

Sounds like you're the man for the job, all right.  Please keep us updated on how it goes.

[Willy_Tepes]

Sure looks pretty though.

[Willy_Tepes]

In case anyone is concerned about the inter-board connectors mounted in a 3d printed frame, the electronics are sandwiched between two carbon fiber boards which act as stiffners and provide a solid mounting point for both the motor, battery pack, and the mounting screws.

[mr ed]

There's a reason drones use thrusters- easy to control. A big prop will need massive anti torque and the control system will have to be  quick to control this light mass. I assume that's all figured out already.

 I would try to get everything on as few pcb as possible, connector weight won't help performance or reliability.

Rember bill Lear? He designed the first Lear jets, and he said he would sell his granny to save a pound of weight.

[Willy_Tepes]

Yeah, the torque of the motor is counteracted by an adjustable slanted tail fin that vectors the downwash opposite of the motor rotation. Since torque increases when you increase rpm, the wings (ailerons) have to dynamically counteract this force. They move independently. This must be programmed into the flight control software. The standard software on such flight control boards contains all the parameters I need to change. This functions basically like a regular RC airplane flying what is known as 3d (hovering), but it lacks a tail plane or any other lifting surfaces and thus is not susceptible to leveling out. It hangs by its propeller. I think the speed of even cheap analog servos is up to the task. They are not moving a great deal of mass nor do I think what is essentially an aileron needs to have a very large surface area. The ailerons are simply vectoring the downwash from the propeller allowing pitch and yaw. Roll is not really needed in this design as this action naturally happens in a turn due to the low center of gravity. In forward flight and most other situations, it will behave like a helicopter because the weight distribution, structural layout and forces are nearly the same.

The size of the wings (ailerons) determines how large of a controlling action they have. The deflection is pluss/minus 45 degrees. I have just chosen the current size out of gut feeling and guestimation. The same goes for the surface area of the tailfin and its angle. Both are easy to change with CAD and 3d printing. It might in fact have some interesting effect on flight dynamics to make the wing/aileron larger than needed for the counter-torque action. But they are not intended as aerodynamic lifting surfaces.

Flight controller, VTX, DVR, 2.45Ghz RC reciever, and microcontroller could all fit on a single board, but that would require the builder to order a fully populated PCB from a PCB manufacturer and upload the software on it, thereby defeating the purpose of the design, which is to allow builders to use existing hardware, and allow a very low barrier to entry. Apart from this, I do not feel qualified to program my own flight control software. I am not that skilled in programming. I will be using flex PCB to connect the camera as it has to span 180 degrees and this would stress any kind of wire connection. This could also be an option for connecting the daughter boards, as it would eliminate any chance of stress because of vibrations, but the main three PCB's are mounted in a very rigid frame and the tolerances are as tight as 3d printing allows. The rigid board to board connectors assist in assembly, so I am not very keen on replacing them. The weight reduction of going from 3 to 2 PCB's in the main assembly and integrating all the electronics is miniscule. It is already a lightweight craft with more than enough power. It can carry a 500 gram payload. If you wish to design such a board, it would simply drop in without any issues and would free up a lot of space for anything else you'd want to add, but integrating all the electronics would not reduce the size of the drone as other components like the camera system require this large of a cross section.
It's not bigger than this.


It is important to not lose sight of the mission while working on a project. My mission is not to make millions. And the intention of the design is to provide a flying platform on which you can develop whatever aerial system you need. It is not a use-specific design. It is meant to be modular. I will be using the entire electronics package unmodified in a motorized glider at a later date.

Motor selection is up to the builder. It will fit any motor from a 2208 to a 2316. That is up to about 1.6-1.7 kgs of thrust on a 250-300 gram craft, so in the extreme upper limit I have a 5:1 lift to weight ratio. I don't know of any quad that can match those numbers. Bigger propellers with more blades and steeper pitch require higher torque from the motors but also produce more trust. Battery voltage and weight can range from a 2 cell 18650 or 2 cell LiPo, to 4 cell LiPo, from 850mAh to 1600mAh capacity. This is because the design allows you to strap on a standard battery pack, or use one of the 3d printed battery housings which slot in with a key way. The more cells you have in the battery, the faster the propeller will spin (Kv-rating). 18650's have a much lower discharge rate and will result in a very docile drone when paired with a motor that has a smaller stator size, like a 2208 and a 7 inch 2-blade or tri-blade propeller with less pitch. The bullet connectors for the motor are there because of assembly considerations.

There is a GPS hidden away in the tail section, and the VTX antenna is located in the tail, far away from the motor. It has an additional 3 digital and 3 analog IO pins, 3.3V, 5V and VCC pads, and a I2C line for sensors. If you use the Raspberry Pi Zero, you have even more IO pins and the ability to do object recognition and tracking. The DVR function can then be preformed by the Pi.

[Willy_Tepes]

This is the flight control principle it uses. Hovering.

[uer166]

What advantages does this have over a quad? This seems like a lot more moving mechanics that are much trickier than 4 rotors.

[Willy_Tepes]

-This is not meant to compete with products from DJI, this is primarily a development/sensor platform. You can make it into whatever you want.
-Works like a standard racing drone. There is no need to mount the microcontroller if you don't want it.
-There is only one motor, cheaper! 1 motor = fewer failure points. A quad does not have motor redundancy like 6 and 8 rotor drones.
-It is simple. One motor, three servos and a camera does not make it complex.
-No complex rotor hub like a helicopter has, no tail rotor. Has fewer servos than a standard RC airplane.
-You can order the 3d print, carbon, and PCB's directly online, or 3d print the drone and etch the PCB's yourself. It is a very simple two layer board. You could even assemble this on a single sided perboard with wires and it would still work. A child could assemble it.
-No regulatory shit. It is up to you if you want to comply with stuff like Remote ID. Totally untraceable if you want it to be. No paper trail.
-No military or law enforcement sales (End user agreement). I'll sue any agency who tries, and I hope they do. I won't say no to free money.
-It is an extremely cheap and simple design. I cannot imagine that you could find anything else at this price point with this much power and features. I only want 20 dollars per download for the design files.
-You can use existing drone parts you have laying around. The motor and servos are cheap, a very common size in the hobby. (No specialized parts).
-A wide range of flight controllers, cameras, DVR and video transmitters can be used, making this user upgradable.

Possible uses:
-Drone racing.
-Freestyle flying.
-Environmental monitoring (Gas/radiation/particle sensors).
-Search and rescue (you can fit a FLIR camera and a LED matrix under the battery).
-Object recognition and tracking, both visual and audio (with microphone array).
-Counting livestock (sheep and shit).
-Flying network repeater.

It's advantages over a consumer drone are many, but it will mainly appeal to people who like to make stuff or people who like to fly really fast. Technically, this is just my attempt at reducing the complexity of a drone. You simply can't make a propeller driven craft simpler than this. 1 motor and 2 servos is the minimum amount of components you can fly with. I challenge you to prove me wrong.

Now, back to the topic.
What I really need is advice on the PCB layout. All the mechanical design stuff is finished. This is the final stage before testing. It would probably work just fine as it is, but I want to do this properly. Even though I could not give a rat's ass about FAA and FCC regulations, I don't want to cause problems for other people. And I don't want it to look like the PCB was designed by an amateur. I'd like a clean video signal and as little noise as possible.

[uer166]

It's advantages over a consumer drone are many, but it will mainly appeal to people who like to make stuff or people who like to fly really fast. Technically, this is just my attempt at reducing the complexity of a drone. You simply can't make a propeller driven craft simpler than this. 1 motor and 2 servos is the minimum amount of components you can fly with. I challenge you to prove me wrong.

2 servos and a motor is more moving parts and complexity by 2-3x (depending on how many gears the servos have), so your design is a net increase in complexity when compared to a quad. It doesn't mean it's not exciting though!


Re: PCB: as long as you're not attempting at making your own ESC, I don't see many issues you'll hit. I recommend reducing the number of PCBs and interconnects by at least 2x from what you have now, this may require to densify the design slightly, but you have so much space on the boards it should be possible. E.g. some of the daughter boards can be integrated onto main board. The high power board is redundant: just use wires. I see you have length matched traces, that is completely unnecessary here. Always use at least 4 layer PCBs with 2 solid ground planes in inner layers to avoid SI issues.

[uer166]

-No military or law enforcement sales (End user agreement). I'll sue any agency who tries, and I hope they do. I won't say no to free money.

If it's open source-ish you cannot enforce this whatsoever. Looks at all the Ukrainian FPV drones destroying tanks 

[Willy_Tepes]

I've explained why I can't make the PCB's denser or reduce their number. That would contradict the purpose of the design.
The daughter boards are simply connection points for the servo leads and for routing the circuit out to the LED's, GPS and camera system. Running wires directly from the control system to the "peripherals" would only complicate assembly as they are on opposite sides of the carbon fiber stiffeners.
We'll have to disagree on the complexity issue since this consideration also includes the failure rate of the different components. The high power board is not redundant as the 40 amp ESC takes up almost the entire board area and I don't want it flopping around in high G turns, it needs to be secured to something and there are no mounting holes on a ESC. I also have to consider cooling as both the ESC and VTX get very hot. The cooling system is already finalized. I am pinching millimeters in this design.

Matched trace lengths are not required on such long UART lines?
Almost all the traces are on one layer. The bottom layer is almost an uninterrupted ground plane. I really can't see the argument for using 4 layers.

[Willy_Tepes]

Just as I cannot stop Chinese companies from copying my design outright, I cannot stop non-legal entities from using this drone. But did I ever say that was my intention? 
I only want to stop military and government agencies from using it. That is something I CAN enforce, because they are bound by the legal system. If I did sell to them then this would automatically fall under ITAR (International Traffic in Arms Regulations). All countries have such regulations.
I do not think DJI had any clause against military use when this conflict started (they obviously have now), but the drones used in Ukraine are not purchased by government agencies, they are acquired by straw purchase using civilians as buyers. In a proxy war most things are done by proxy to avoid accountability. Neither the Ukrainian government or any of the western governments supplying them bought drones directly from DJI.

I am absolutely NOT a pacifist, I just regard the government as criminals and refuse to do business with them.

[uer166]

You asked for PCB advice and were given some, but then ignored all of it for reasons unknown. Feel free to do whatever since it's your project, but what exactly is the point of the post?

The PCBs are very beginner/naïve, and can easily be improved with a little bit of attention.

[Willy_Tepes]

No, I asked for help with the PCB layout, not the physical design of the boards, mechanics, or assembly methods. I have been given advice on everything BUT the PCB layout.
I gave you reasons why your suggestions were not relevant or feasible, so the reasons are not "unknown". I have explained everything in detail.

My specific questions are:
-Are the vias I have placed sufficient? I believe I have provided a return path for everything, there are no loose copper islands, is there anything else?
-Is there anything else that can cause noise issues?

This should be a very simple question, but if you are not qualified to answer, just say so. I came here to ensure that the PCB's don't look amatuerish, but I seem to only be getting answers from amateurs.  If you could explain WHY I'd need a 4-layer board and two internal ground planes it would help a lot because I'm not in the habit of doing something on the sole basis of "some rando said it". 
The connecters are not really a serious issue. Everything is securely mounted, and the mechanical tolerances are down to pluss/minus 0.1-0.2mm.
The connectors are there to aid in assembly, period! I am not gonna run fricking wires all the way to the LED board because I can't solder something already mounted to the 3d print. PLA melts like butter at 60 degrees and printing in Nylon or ABS won't solve that issue. Soldering irons and plastic don't mix.

Now, what about the PCB layout? Specifically the traces, NOTHING ELSE. Are you qualified to answer or not?
Just to clarify: I cannot move any components as their position is constrained and dictated by the mechanical design, as in ANY other product. I will NOT integrate any components onto the board because that defeats the purpose of the product. These are simply carrier boards, nothing else. The user must be able to assemble and disassemble the drone for maintenance or component replacement without soldering (modularity).
The electronics package needs to plug into the rest of the structure as this is the last step of assembly.


Your suggestions are based on the presumption that I am making a consumer drone like a bloody DJI product, or that I can just secure shit with double sided tape and wires. This is a DIY kit build and a sensor development platform as I have explained over several posts.
Now if you wanted to integrate all the electronics, you could design your own PCB for this. That is the entire purpose behind this project. You can integrate any electronics you want. I am just providing a simple and powerful drone for you to tinker with at a very low price point. If you like quads, fine. Have all your electronics exposed. Here they are all protected within a robust plastic structure that would survive a light crash with only superficial, or very limited damage.

[mr ed]

Two more issues to consider.
 1  Ground effect. The downwash will do funny things close to the ground effecting control.
2 stalling of the control wings/flippers. You only have +/-  17 degrees of angle of attack to play with then the torque control has a large discontinuity to manage.

[Willy_Tepes]

Stalling is only an issue if it is an aerodynamic lifting surface. I am literally only deflecting the air stream from the rotor. The air pressure will in any case be larger on one side of the aileron than the other, and thus I have the force I need to counteract motor torque.
Downwash on the ground is an interesting phenomena, but this design is no different from a helicopter in this respect. Only testing will give definite answers. I have designed a test rig for it.
The deflection is pluss/minus 45 degrees. Where did you get 17 degrees from? I probably won't need to use all of that for counter torque, and the slanted tail fin does some of the work (adjusted to counteract torque during hover). The ailerons also control pitch. Mixing control channels is simple in todays flight control software.

[Willy_Tepes]

I caught some small errors myself, so if I don't get any real advice on the layout within a week, I'm going with this.
If you had not been so preoccupied with giving me advice on exactly the things I told you to not to bother about, you would have seen that I forgot pull up resistors on the I2C line UR166

[Daixiwen]

Matched trace lengths are not required on such long UART lines?

No you don't need matched trace lengths here for multiple reasons. UART signals are asynchronous, and what you have is a tx and an rx signal, which are completely unrelated (timing wise). You only need to think about matching lengths when you go higher in frequency and have a strong timing relationship between the signals (either a differential pair, a data signal with setup/hold requirements relative to a clock, or a bunch of data signals with a timing skew requirement).

I really can't see the argument for using 4 layers.

I don't really see it for such simple PCBs either.

[Daixiwen]

If you had not been so preoccupied with giving me advice on exactly the things I told you to not to bother about, you would have seen that I forgot pull up resistors on the I2C line UR166

Those kind of mistakes would be easier to spot on a schematic than on a PCB layout

[Willy_Tepes]

Thank you for advice with arguments that make sense
I did not make a schematic for this because it was so simple. But still, one should try to answer the question being asked, not everything else.

[liaifat85]

You have already got many valuable advice. You can also take a look at this drone PCB: https://www.pcbway.com/project/shareproject/WIFI_controlled_drone.html
It's a full Arduino based drone in single PCB and controlled via mobile IOT.