Elementor #3689



Making It Up As I Go:

I expected this project to run longer than anticipated, but not quite this long. The reason for this is the my build process. As the build progresses, I see improvements that can be made. Since I am in no rush to go fly in this weather, I have no problem taking a bit longer to make the changes. With the amount of time it has taken, I wish I could say that the changes are obvious, but they have all been very minor (mostly construction methods).

It’s not very obvious, but many things have changed since the first mock up frame

The Fleet:

          The grey frame will be the first to take flight… or crash

Differences between XLR V1 & V2

I have to admit, the V1 definitely had more to give but I didn’t want to chance destroying it. I think it might have done a bit better on an R-line battery. Also, due to the vtx antenna issue, the antenna had to be placed on the outside and a very crude, last second “aerodynamic” covering was made for it using electrical tape (not shown). This definitely slowed it down.

Overall Layout:

It came as a surprise that the V1 was pretty close to an optimal layout. The only major visual difference are the motor direction, motor cones, and prop spinners. The not so obvious difference is the location of the CG which was the biggest issue with the layout of the V1. This was why the 5S versions of the V1 (about 45mm shorter) were extremely unstable. The shorter body moved the CG behind the Cp (center of pressure) and caused instability at higher speeds. Luckily the 6S version moved the CG forward enough – it still wasn’t in front of the Cp, but close enough for the flight controller to keep things stable.


Although the V1 was derived from a NACA profile, it wasn’t optimized at all. Contrary to popular belief, the best shape is hardly ever a perfectly scaled rendition of an airfoil profile – it will most likely be a derivative. Finding the best shape for the V2 involved several steps to find the optimal solution (least drag) using multiple factors that are dependent on each other. These factors were NACA 00XX number (nose and tail cone), wetted surface, Reynolds number (at target airspeed), component placement, component shape, minimum diameter, and CG in relation to Cp.


The motors are in a much more aerodynamically favorable “pusher” configuration with a nose cone and large spinners for the propellers. Although both styles were designed, I opted to try a traditional puller first (only by a small margin), but this flight by Annika B tipped the scale back to a pusher. Small cooling inlets were added to the nose cone tips to allow cooling for the motors and ESCs. NACA inlets were scrapped since the size and shape of the nose cone didn’t allow for favorable air flow. Having the inlets in the tip of the nose allows a smoother transition (through the use of some internal ducting) of airflow over the ESC MOSFETS and then through the motor.


These are a hybrid construction of an aluminum spinner and a 3D print. These have yet to be flight tested but have been bench tested (without props) at roughly 50,000 RPM.

Cross section:

The V2’s arm width is slightly wider (0.1mm) than the V1 since I opted to go with 3D printed sleeves. The sleeves are much more consistent and cleaner looking than the previous finishing method. Theoretically, this should make up for the added width.

The fuselage diameter is 46mm which is 4mm smaller than the V1 and equates to a reduction of 15.4% cross section of the fuselage. Overall, the cross section of the whole frame is 4.8% lower than the V1.


The V1 was simple since it basically had an exoskeleton (carbon fiber arms and tube with 3D printed nose and tail) and the weight was roughly 95g. This worked well to protect the battery and that was about it. I also didn’t like the fact that the battery was not able to “break away” during a crash which could lead to shorting out components and possibly starting a fire. The V2, on the other hand, uses a carbon fiber endoskeleton frame surrounded by a  3D printed body. The V2 frame is not expected to survive a crash but some measures have been taken to make the electronics salvageable. The FC, rx, and vtx are well protected by the frame and the battery is held in a way that will allow it to break away in a crash (not a guarantee though). The fpv cam and gps is in the front of the 3D printed fuselage and the wires will be mounted in a way so they will be able to break away instead of ripping out the solder pads. The frame is 29g while the 3D printed shell is 49g for a total of 77g.


  • Motors: The same old Cobras 2207 2450kv motors will be used
  • ESCs: The Airbot Furling Mini 45A ESCs are just the right size
  • Flight controller: Matek F722 Mini and Airbot F4 Micro R6
  • Vtx: RDQ Mach 3, AKK FX3
  • Battery: Custom made 1300mah 4S R-line (for testing) and 1300mah 6S R-line (for speed runs)

Overboard With Detail:

Because I tried to optimize and refine as much as possible, there isn’t much room to work with on the inside. I think this was a mistake. The only area that is not so cramped is the nose right where the camera and GPS is. Other than that, wiring is a painstaking process – some of it is much like running pipes for hydraulics.

The Ultimate Limiting Factor:

Prop tip speed. There is a way to circumvent this and experiments are have yielded some very promising results. If successful, this alone could increase speed by an additional 15-30%.


I may change/add to the above information, but this is where I will post all updates regarding test flights, etc. This also includes all posts that began from the start of the project. Right now, it is in reverse order…

January 15, 2019:

Another Quick Update

I plan to reorganize this page in the very near future since I dislike the running update format. Instead, I will use an index.

The first frame is almost ready to go. Most of the body work is finished but a lot of wiring remains to be done.

January 3, 2019:

A Quick Update:

As expected, a lot has been reconfigured for the underlying frame so while waiting for the frame parts, I did some body work. The arms are untouched since they have already been redesigned.


December 26, 2018:

3D Printing

Christmas was good. I finally have a 3D printer of my own. I printed out the nose and tail cones for the body and also the cones for the motors. Everything fits perfectly (including all the internal components) but the motor cones need a little work since the NACA ducts interfere just a wee bit with the motor mounts.
Push It

A last minute decision was made to make the layout into a pusher. I had been debating this for a while until I saw Annika B’s flight with her 113mph 3s 6″ ARX-R build.
It’s Still Early

Although it’s nice to see this mock build up of the frame, there is still a lot of work to do on the inside and outside.

XLR V2 0.16

XLR V2 0.19

XLR V2 0.20

December 15, 2018:

Full speed ahead

After a failed attempt at a production frame, logic says it’s best to stick with prototypes so attention is back on this project… at least for a little while. This time around, CAD is being used and the details are about 90% done. The original XLR frame was all done with hand measurements, hand tools, and a Dremel (except for the nose and tail cone).
The Secret Design

There is nothing extraordinary about the frame, just lots of engineering time. It’s a disappointment that it isn’t something radically different – all the math shows that the basic layout should stay the same. Although a lot of research was done, much of aerodynamics is a sort of art form which is learned through experience. This is where Forrest Frantz and James Frantz come in – they have helped throughout this experience since theory can only do so much.

That’s what it’s all about this time. No radical changes to the layout means that:

  • Drag must be minimized
  • Maintain or improve motor efficiency
  • Keep the center of pressure (CP) below/behind the CG

The hardest factor to deal with when optimizing is the center of pressure.
Under Pressure

That’s where the CG was on the XLR V1 which isn’t a good thing. Much like a rocket, the center of pressure should be below/behind the CG (relative to the direction of motion). Luckily, the CG wasn’t too far off and it still flew with decent stability. What is the significance of the CP? This is the point where the force air resistance is acting on. Unlike CG, the location of CP can change depending on the direction of the incoming air (this gets complicated very quickly). How can CP become a problem? As a quick reminder: free objects (like a quad) always rotate around the CG. In general:

  • CP in front of CG – if the quad deviates from the flight path, the force acting on CP causes a rotation that makes the quad deviate even more from the desired flight path
  • CG in front of CP – if the quad deviates from the flight path, the force acting on CP causes a rotation that makes the quad return to the desired flight path

An easy example: If you’ve ever played a game of darts, you’ll notice the front end is usually metal while the back is light with large fins. The metal in front keeps the CG in front and the fins increase the surface area – this gives a larger surface area for the air to act on which means the center of pressure remains behind the CG. This gives a dart directional stability. This is why throwing a dart backwards works poorly. All aerodynamic objects have a certain amount of directional stability.
Another example of directional stability: When you stick your fist out the window of a car going down the freeway, you can rotate your fist without it being forced (by the wind) in any particular direction. When you try the same thing with a flattened hand or any other flat object, the wind will force your hand to move in a different direction as you rotate it.
Doesn’t the Flight Controller Take Care of Stability?

The CP only becomes an issue for aerodynamic quads at high speeds at WOT:

  • At WOT, the flight controller loses some ability to control the quad
    • With all motors at 100%, the FC can only reduce RPM to maintain stability
  • This isn’t a big problem for typical quads since, due to the distribution of drag and that most of the air is turbulent (like your fist out of a car window), the quad stays responsive and maintains stability
    • In other words, a typical quad has poor aerodynamics and therefore poor directional stability; the quad isn’t fighting the flight controller to maintain a certain heading
    • For an aerodynamic quad, the design favors stability through the air in a particular direction.
    • At high speed, aerodynamic stability becomes an overwhelming force. Any deviation from the direction of stability is enough to out muscle the FC – or at least cause a struggle between the 2 stabilizing mechanisms

October 4, 2018:

First flight was somewhat successful – there was an RFI issue on landing so a few arms need to be and have been replaced. PIDs were not good (much different from the original XLR due to a new Betaflight version) so it had a hard time tracking straight. Therefore, peak speed was never reached and it “only” hit 185mph.
Crunching the Data

Although the flight had mixed results, tons of data was gathered and crunched (and still crunching). With the help of a few friends I have made, we can now make some highly customized components. Flights from here on out will mostly be an added bonus to try out some other intangibles.

September 9, 2018:

Just a brief status for the moment until I have a little more time. Right now, I have an “intermediate” data frame that is 98% built – I just have to solder together 6 more wires:

  • The data frame will be used to gather some crucial flight data. This data will be used to determine the dynamic state of the propeller (since static tests tell us nothing) at top speed
    • The full explanation is beyond the scope of this web page, but the key here is to determine the degree of the propeller blade’s AOA (angle of attack).
    • From here, it can be determined (along with other data) how to adjust the physical parameters of the frame and/or electronics to increase speed.
    • Propeller AOA is dependent on the physical parameters of the frame and RPM at top speed. This is why static tests tell us nothing. Wind tunnels also fall short due to several factors that can only be determined by field tests.
  • The data frame is smaller in cross section than the first XLR project and has a very high potential to go faster – especially since the VTX antenna can safely go inside the tail this time
  • The data frame’s nose and tail have been heavily reinforced to help it survive any crashes

I hope to fly it this upcoming week, but as always, delays are expected