This project was started in , electronics and control loops. Because I always need a cool project to learn new things, it was clear that something that can fly had to be built.
The project started as a "tricopter-only" project, but as I wanted to build smaller vehicles with more payload capacity, I decided to make some quadrotor, hexacopter and Y6 hexacopter firmwares too. My main interest is to build very small MAVs that fly as good as larger ones (or even better) and that can be controlled by wireless video link. I also experimented with autonomous flight in GPS-denied areas (video), and with GPS assisted autonomous hover (video).
-- William

Contact: Shrediquette @ g m x . d e --- All content published under CC Attribution-Noncommercial-Share Alike 3.0 Germany

How to measure top speeds of multirotors

How fast can your multirotor fly? The simplest solution for this question would probably be to attach a GPS device (tracker / OSD) to your multirotor and to read out the maximum speed.

But hey, this is pretty imprecise...!

GPS devices suffer from measurement noise (e.g. through signal degradation), which becomes problematic on rapidly accelerating objects like our multirotors. More advanced GPS chipsets (e.g. u-blox LEAx) have the option to choose between different filtering modes (e.g. pedestrian, car, airborne), that make some assumptions on the maximum accelerations and movements of the sensor and filter the results accordingly. That will most likely improve the accuracy of the measurements, but still noise might remain. Here is an interesting article on GPS speed measurements, which states that the classical methods to calculate velocities in GPS receivers have an accuracy in the order of a few meters per second, due to significant noise.

I was recently attaching a pretty good GPS logger to a quadrotor (HUMs with Cobra 2204, 75C 4S 1300 mAh, C-prop 5x3). There seemed to be hardly any wind. The "Top Speed" value reported by the tracker was 113 km/h. But this value includes the influence of measurement noise and wind, and it is therefore not precise.

The only way that I know of how to deal with these problems and how to get the real (air)speed out of a GPS logger is to do the following:

  • Fly full speed at a constant height, on a straight line and against the wind for at least 300 meters. 
  • Then, turn around and fly the same straight line but in the other direction. 
  • Download the data from the logger and calculate the average of each of these two straight line runs (this will remove or at least attenuate noise). 
  • Now, calculate the average of the two runs (this will remove the influence of wind and yield the true airspeed).

The true airspeed of my copter was calculated to be 95 km/h, which is 16 % lower than the maximum speed reported by the logger. And I must add that this was a day where I felt hardly any wind. Still the wind velocity was about 10 km/h according to the GPS measurements. More wind will of course make the difference even bigger.

So if you hear people saying "Hey, my quadrotor is flying 150 km/h!", then be careful until you saw the measurements...

Today, I'll hopefully measure the top speed of my Shrediquette DERBE, let's see what comes out...!

Shrediquette DERBE ... almost there...

Here are some images of my racing copter. Some soldering of the ESCs etc. still has to be done. I am also waiting for some resistors and capacitors for my flight controller to arrive... But I guess the maiden flight will be this weekend :-D.

Aerodynamics in racing multirotors!

Why future racing copters really should look different.

by Dr. William Thielicke aka Willa aka Shrediquette


In this article I try to demonstrate why FPV racing multirotors need to look different. Some small modifications to the frame would (in theory…!) result in 70 % higher top speed! All that needs to be done is to align the arms parallel to the propeller flow, and to tilt the main body of the copter by about 40 degrees. I am presenting a very simple and robust racing copter design that incorporates these ideas. Furthermore, I am calculating the aerodynamic drag of different copter concepts using basic equations. The aim of this article is to make you realize the importance of aerodynamics and to stimulate people to design more innovative racing frames.


Until recently, multirotors were mainly used as a “hovering device” and the top speed of these copters did hardly matter. Now, multirotor racing has become popular and all competitors are seeking for very fast and agile multirotors.
FPV racing in Bexbach, Micha vs. Willa

Apparently, the aerodynamics of multirotors have been neglected in the past (with a very few exceptions), but aerodynamics actually become very important at increasing flight velocities. Just as a quick example with numbers that are realistic for copter racing: A quadrotor that weighs 0.7 kg and flies at a pitch angle (alpha) of 45 degrees has an aerodynamic drag that is also 0.7 kg (about 7 N) – which is clearly very significant. A large portion of the force that is generated by the motors is therefore just “wasted”! If this drag could be somehow reduced, then a racing copter could fly a lot faster (or it could fly longer at the same speed).

Forces acting on a constant-speed multirotor at 45° pitch angle. Drag is important.



The real flow of the air around a full multirotor has not yet been measured to the best of my knowledge. But it is very likely, that the flow around the multirotor can be divided into two distinct parts: The flow around the arms, and the flow around the main body: Directly under the propellers, the flow is pretty fast (between 100 and 200 km/h in racing copters) and it is pretty much perpendicular to the propeller disks. Just as in this picture showing the flow of a large propeller:

Usually, the flow is perpendicular to the propeller disk.
The faster a propeller turns in relation to the flight speed ("advance ratio"),
the more constant is the angle of the flow.

The flow that is hitting the main body or frame is mostly horizontal, as it is hardly influenced by the propellers: It is mainly exposed to the flow velocity that results from forward flight (I however really need to verify these assumptions in a windtunnel or with in-flight measurements, I'll do that asap and report the results here).

Simplified model of the main flow components on a quadrotor.


It would make a lot of sense to reduce the drag of the arms below the propellers, and to reduce the drag of the main body of the racing copter. My first and simple idea to achieve this was to tilt the motors forward, and to design an aerodynamic canopy. These ideas were implemented in the “Shrediquette Gemini” in 2013 (which was later produced by Team-Blacksheep).

Shrediquette GEMINI / TBS Gemini
Some force measurements in a wind tunnel revealed that these two features reduced the drag of the copter by 14 – 40% (depending on the pitch angle).
My next idea was to tilt the body instead of the propellers. The goal was to align the body with the oncoming, horizontal flow, which will significantly reduce the aerodynamic drag. If the arms are aligned parallel to the propeller flow (= perpendicular to the propeller disk), the drag will be further minimized. My HEXO+ design (which was used in a Kickstarter project too) realized the idea of a tilted body.

The HEXO+, a larger and modified variant of the Gemini

The tilted body concept was subsequently ported to my next frame, the “Shrediquette QRC5” FPV racing copter:

The QRC5 FPV racing quadrotor
Recently, I realized that a frame like the QRC5 is not crashproof enough for everyday FPV racing. Therefore I recently finished a “classical” FPV racing copter design that incorporates the idea of a tilted body and arms that are aligned parallel to the propeller flow. The “Shrediquette DERBE” is probably as crashproof and light as conventional racing copters, but it has much less aerodynamic drag.

The main body of the Shrediquette DERBE is tilted by 40 degrees.

The following image nicely illustrates the two main sources of drag as introduced earlier, and how drag is minimized in the Shrediquette QRC5 and the Shrediquette DERBE. The idea is to minimize the frontal area of the copter at typical pitch angles (about 50 degrees in FPV racing), and to minimize the area under the propellers.

Three different concepts: Comparison of frame area (green)
and area of the arms under the propeller (blue).


The aerodynamic effects and the performance of these different frame concepts can be approximated by applying some relatively basic equations. I did the calculations in Matlab (source code here) with a numerical, iterative method. I am sure there’s an analytical way to solve the equations, but rearranging these equations would take me much more time than just letting Matlab do all the work for me… Here is how the aerodynamic drag, and the top speed of three different copters (standard, QRC5, DERBE, each with the same motors, propellers and weight) are calculated. The numbers that I am using a suitable for MN1806 2300kV motors with 4S and 5x4” propellers: The most difficult part of the calculations is to approximate the thrust that is produced by the motors. Thrust decreases linearly with flight speed (if the rotational speed of the motors would be constant) until it goes down to zero at some flight speed. Thrust at different flight speeds is not so easy to measure, therefore I am using an equation that seems to fit experimental data from a wind tunnel reasonably well. I am however planning to do real windtunnel measurements of the propeller performance in forward flight very soon.

Thrust of 5" propellers vs. flight speed.

Furthermore, I am assuming the flow to be perpendicular to the propeller disk below the propellers. The main body of the copter experiences flow that is purely horizontal in this simplified calculation. The copter is set to a pitch angle (alpha) of 1 degree and full power is applied to the motors. The horizontal drag of the main body is calculated from the dorsal and frontal area of the main body at alpha, together with a drag coefficient of 1 (which is a reasonable approximation for a classical racing copter frame) and a horizontal velocity. Additionally, the drag of the arms needs to be calculated and subtracted from the available total thrust: The flow velocity in the jet of the propeller is (based on the equations already mentioned above) approximated to be about 150 km/h and independent of the flight speed. A drag coefficient of 1.5 is assumed. To give a number already: The drag that is created by the arms of the “standard racing copter” at full throttle is 6 Newtons (about 0.6 kg)! This is equal to the static thrust of one motor…! With the remaining thrust, I am calculating the horizontal and the vertical force component at the current pitch angle of 1 degree. Now, the horizontal velocity is increased step by step until the drag equals the horizontal force component. If this is the case, then the pitch angle is increased by one degree and the calculation starts from the beginning. At each increase of the pitch angle, I am checking whether the vertical force component is larger than the weight of the copter (as a side note: I am completely ignoring the vertical velocity of the copter as I am only interested in the final solution where drag equals horizontal force AND weight equals vertical force). At a certain pitch angle, the vertical force is not larger than the weight anymore. At this point, the calculation stops and both the pitch angle and the velocity of the copter are printed.

Here are the results:

  Standard racer Shrediquette QRC5 Shrediquette DERBE
Max. tilt angle [deg] 38 29 34
Top speed [km/h] 107 178 170

The numbers for the “standard racer” seem pretty realistic, although I am quite sure that the top speed is always overestimated. It is very likely that there are additional sources for drag on a copter that were ignored in my simplified assumptions, and that the hypotheses on thrust vs. flight speed could be improved. However, I am doing the same errors in the calculations of all three copters, therefore the above results represent the correct trend: Aerodynamic designs fly much faster and at lower pitch angles. These two factors a highly relevant in FV multirotor racing and in the future, more attention should be paid to designing simple aerodynamic copters like the Shrediquette DERBE. Furthermore, I will be experimenting with optional airbrakes in the first prototype. These will be deployed at very low throttle and when the throttle is quickly reduced (hence reacting to the derivative of throttle). In my imagination, this could really help on tight racetracks, as it enables the pilot to brake without having to change the pitch angle (which would take more time and also results in a camera that points to the sky).

A few renderings of the Shrediquette DERBE (more soon):

I just received all parts and really quickly snapped the parts together. Everything seems to fit really well :-D


Tilted body race quad update 2

All carbon parts are milled, the 3D printed parts are ordered, and motors + ESCs arrived. I can start to assemble soon :-D.

One thing that is new (except for the aerodynamic concept) are the airbrakes. I don't know yet how good they'll work and if they might introduce too much moments around some axis. But I think that they might be advantageous in racing tracks with a lot of tight turns. They'll be extended if the throttle is quickly reduced (hence reacting to the derivative of throttle).

Here are some renderings of the latest iteration:


DERBE airbrakes

Tilted body race quad update 1

Here are some more renderings of my latest project. I'll write an article on multirotor racing aerodynamics soon, then you'll hopefully agree that this design makes sense ;-D

Working on a new tilted body racing frame....

I am working on a new racing frame with good aerodynamic properties. The intention is to make it crash proof and easy to repair too. I think this concept (a mixture between QRC5 and other standard racing frames) might work pretty well. More details soon!

FPV Airrace!

We just returned from the awesome airrace in Bexbach. It was so much fun, we had intense battles, crashes and many broken propellers... I finished second in the "Pro" class, right after "Metaldanny" from the Netherlands. He was extremely fast, and had a very clean flight style.
After my QRC5 was knocked out the sky during an awesome battle for the first place in the qualifications, I was flying the finals with a TBS Gemini proto.

Big thank you to Rolf Venz who organized this great competition!

Rookie Class
1: Matthias Schwarz
2: Jonas Schnell
3: Florian Maussner

Pro Class
1: Metalldanny
2: William Thielicke
3: Udo Michel

Fastest Lap Pro Class

Fastest Lap Rookie Class
Matthias Schwarz

Longest Distance Travelled
Ulrich Wirrwa

Best Crashes
Oliver Deventer

Here are some pictures that I copied from the Airrace Facebook group:

My QRC5 (left) in an absolutely awesome battle against Michael S. It ended in a midair collision.

Around 32 people from Germany, Switzerland and the Netherlands were competing

Winners of the competition

Michael S. is chasing me. We were changing the lead constantly...

FPV Air race Saarland

I am currently attending an airrace event in Bexbach. At the moment we're having some training flights. There are a lot of people that really know how to race! I crashed during every test flight, just because I am completely out of practice... Let's see how the races will be like!

QRC5 first FPV flight

I had the chance to fly a few batteries in FPV mode, and it's really great fun. The top speed and throttle response is very good, the inertia in tight turns is also not too high although the copter weighs around 400g.

Here's a short Dom HD capture:

The coming evenings, I'll try to get as much practice as possible to prepare for the FPV Airrace in Germany.

QRC5 first flights

Today was the first "garden flight" of the QRC5. It flies really great and has tons of power (4s motors + 2300 kV motors drawing 12.5 Amps each at full throttle). Due to some modifications in the code of my flightcontroller, it is more smooth and responsive (sounds like a contradiction, but is isn't!) than ever before...: I am using kiss escs with "oneshot" enabled. That means that these set the speed of the motor according to a 125 - 250 microseconds PWM input pulse. My code automatically determines how long the calculations of the control loop take and then synchronizes itself with the PWM pulse. This results in a minimum lag between reading the sensors and reacting to them. Furthermore, I also calculate the velocity of the throttle stick and add it to the motor output. This gives a very good response to throttle commands (something like feed forward). I will soon also include the accelerometer to further enhance this. Now I need to 3d mill the moulds for the canopy, and then I am ready for the first aerial grand prix in Germany / Saarland!
Next week I will practice for this event, actually I have flown only five battery packs this year... This needs to increase!

QRC Progress - Part I

Tonight, I assembled the first parts... Everything seems to fit as expected :)

QRC5 design

The QRC5 incorporates some ideas that I had when I was designing the GEMiNi and the HEXO+. First of all, it's not a hex- or trirotor anymore, it is one of my first quadrotors... Personally, I think quadrotors are a bit boring. But there are many FPV racing events coming up, and I am sure that most of them will have several different regulations and limit the battery voltage, the amount of motors or the rotor diameter. The most 'standard' racing copters are quadrotors, they have 3S batteries and 5" props. I am sure that they will get their own racing class. I would like to be able to compete with them, therefore I decided to design a copter for this standard class, even when I am not the biggest fan of quadrotors or racing rules and classes.
The Shrediquette QRC5 fits the 250mm quadrotor racing class
Tilted propellers have become popular since the GEMiNi was presented, and in my opinion they really have an advantage. The aerodynamic drag is reduced quite a bit in fast forward flight. And, also important, the drag is increased in slow flight and during braking. This part becomes important on racetracks with sharp corners.
The QRC5 does not really have tilted propellers, but a tilted body (which is very similar). When only the body is tilted, there is little vertical offset between the propeller disks, which seems to be advantageous for the flight control.
The body is tilted 30° backwards, leaving the propellers on a relatively similar plane
The arms that connect the motors with the body are flat plates that are aligned with the dominant flow direction: Below the rotors, the flow will be 95% perpendicular to the rotor disks, no matter how fast we fly. The flat plates have very little drag, which will increase the maximum flight speed. Each of these plates are made from a three-layer sandwich (glass fibre, end grain balsa, glass fibre). This results in extremely rigid and lightweight arms.
Top view: The flat arms are parallel to the main flow direction
I'll again use the 'CCD-Killer' CMOS camera by Fatshark. For me, this is still the best camera, even much more expensive CCD cameras don't give such a great image. The camera is tilted 13° up (with respect to the propellers).
The CCD-Killer sits between the arms