Posted on

GPS Speed Run Analyzer

This spreadsheet was created to meet the required analysis pertaining to Guinness speed records and posted here to allow scrutiny of calculation methods.

Contents:

  1. Assumptions
  2. The Purpose
  3. What It Does
  4. Using the Spreadsheet
  5. We Still Need Humans
  6. How It Works

_

Assumptions

To use this spreadsheet, I’m assuming that you are using a GPS antenna and have blackbox logging enabled. As long as the GPS is connected to the flight controller, blackbox automatically records the GPS data. However, the data can only be viewed using the blackbox decoder tool (link below). GPS data does not show up on the blackbox viewer.

_

The Purpose

To save a lot of tedious work and time. But this depends on how you want to slice your data. Finding the peak speed is simple and takes only a few seconds. However, if you want to go for something official (like a Guinness World Record), analysis is a bit more involved. This spreadsheet makes this analysis a lot easier and is specifically tailored to the speed record requirements for Guinness.

I covered several topics concerning GPS accuracy, found here, the reasons for using GPS logs and how the data is interpreted, both found here.

Although one of the links above describe how to interpret GPS log data, this blog has a link to a spreadsheet that can automatically do all the analysis in just a few copy and paste steps.

Downloads:

_

What It Does

The spreadsheet will analyze the data to find consecutive data points that cover a distance of 100m or more but it will only analyze data that is above a certain speed threshold (which you can adjust).

Since speed is always changing, the number of data points required to cover a 100m distance also changes. This means the data must be analyzed using different data sampling sizes. Sample sizes can also be adjusted if needed but the default values (in whole number increments from 4-14) usually work just fine.

When the spreadsheet identifies the first speed run out of the data, it sets this heading as ‘Direction1’ and all subsequent runs fall into this category. All runs in the opposite heading (within a set range) will fall into the ‘Direction 2’ category.

Once complete, the following is shown:

  • Fastest pass over a distance of 100m or more for each direction
  •  For each direction, the distance covered during the pass
  • Change in altitude during the pass for each direction
  • Angular deviation from horizontal during the pass for each direction
  • Heading of each direction
  • Average speed of the fastest opposite direction passes

  • Average angular deviation from horizontal for the fastest opposite direction passes

  • Angle between the headings of the fastest opposite direction passes

  • Fastest overall 100m or more speed pass

  • Peak recorded speed

_

Using the Spreadsheet

Although this spreadsheet saves a ton of time, there are a few very simple steps involved to use it:

  1. Open the GPS Speed Calculator spreadsheet.
  2. Download blackbox tools here.
  3. Unzip the file from above into a folder.
  4. Place a copy of the blackbox log file into the same folder:GPS Speed Calculator 01
  5. Click and drag the blackbox log file onto the blackbox_decode.exe file. After a few seconds, 4 files will be generated.
  6. Open up the file that has the .gps.csv extension:GPS Speed Calculator 02GPS Speed Calculator 03
  7. Once open, press Ctrl+A to select all the data, then Ctrl+C to copy the data.
  8. In the Speed Calculator, highlight the cell that says ‘time (us)’ as shown below and press Ctrl+V to paste in the data:GPS Speed Calculator 04GPS Speed Calculator 05
  9. Next, highlight the light gray rectangle of cells on the right side of the spreadsheet. Press Ctrl+C to copy them:GPS Speed Calculator 06
  10. Scroll down until you see the first highlighted section of speed run calculations and paste the copied cells below the data as shown:GPS Speed Calculator 08
  11. Continue scrolling down and pasting the copied cells below the calculated data. Opposite direction runs will be highlighted in alternating colors:GPS Speed Calculator 09
  12. Once all the calculated data have the analysis cells pasted below them, scroll back to the top left side of the spreadsheet to see the finished analysis:GPS Speed Calculator 11

_

We Still Need Humans

Depending on what you are trying to accomplish, there are 2 things to verify:

  • Were the speed runs in opposing 180° directions within reason?
  • Was the total deviation from horizontal relatively level? NOTE: Since gravity is a constant, it only matters that the total horizontal deviation of both passes are within reason. If direction 1 pass is -10° and direction 2 is 10°, any kinetic energy gained in direction 1 is offset by the kinetic energy loss in direction 2.

If there are issues for either one, you can simply delete the bad data so the calculations don’t take it into account.

_

How It Works

Here is an overview of the calculations done in the spreadsheet and how it analyzes the numbers.

In order to make the spreadsheet a little cleaner, most of the calculation columns are hidden.In order to see them, highlight the column headers, then right click and select unhide from the drop down menu. I’m no expert in Excel so there may be a more efficient way to have Excel do this analysis, but for now, this gets the job done.

After pasting in the GPS data:

  • Highest recorded speed is found by finding the max value in column F and displayed in under the Summary.
  • Column I finds the time difference between the current data row and the previous row and displays it in seconds.
  • Column J finds the altitude above ground by subtracting the first GPS altitude reading (cell E31) from the current data row reading.
  • Column K finds the distance travelled between the current data row and the previous row by multiplying the time by the speed.
  • Column BD finds the first speed run and displays the average heading for the next 16 data rows. All cells for the rest of this column are filled with the same number to be used as a reference for direction 1. The spreadsheet identifies each speed run in alternating colors between direction 1 and 2.
  • Columns L through O repeat 10 times. Each set of 4 columns carry out calculations using the number at the top of the columns (row 44) as the sample size of the GPS data points. Different sample sizes are used since the number of samples needed to cover 100m+ changes with speed. Also, since speed is independent of distance travelled, a wide range of sample sizes are used and analyzed as long as the distance is 100m+.
  • Column L: the sum of the distance travelled over x number of cells (x = sample size, and cell range is current row-x+1 rows) is calculated. If the sum is equal to or over 100m, the distance travelled is displayed.
  • Column M: if a number is displayed in column L of the same row, column M will calculate the speed over this distance by dividing the distance travelled by the time change over the range of sample points.
  • Column N: if a number is displayed in column O of the same row, then the distance travelled for the sample range is shown.
  • Column O: if a number is displayed in column M of the same row and is the max velocity in x rows above and x rows below in column M, the altitude change is displayed for the sample range.

After pasting the analysis cells below each speed run:

  • In the top row, the max speed for each set of the 4 repeating columns are displayed below their respective columns. For the max speed in this row, the distance travelled and altitude change is displayed in the row below it.
  • On the right side of the analysis cells, the max overall speed over 100m distance is displayed along with the corresponding distance travelled and altitude change. The average heading for the run is also calculated. This will be displayed for either direction 1 or 2 depending on the average heading of the speed run compared to the reference heading.
  • In the upper left corner of the spreadsheet, the max for the corresponding pass directions are displayed by finding the max value in the corresponding analysis cell columns. Average opposite direction speed is calculated along with the angle of the run relative to the horizontal.
Posted on

Prop Unloading

What Is It?

We hear about prop unloading here and there, but not much attention is given to it. It is an important concept to understand when considering how our quadcopters (or any propeller driven device) perform. My goal here is to explain what it is and why it is important.

Contents:

  1. Unloading
  2. Pitch Speed and Slip
  3. Thrust is Dynamic
  4. Why Do We Care?

_

Unloading

Unloading is exactly what it means: taking the load off of the prop. What type of load is this? It is the force created when the prop spins in the air. As we know, the prop spins by the power of our motors, or more specifically, the torque generated by the motors. In other words, when we talk about prop unloading, we are actually talking about motor unloading. Before going further, here is a breakdown of the motor/prop/air interaction:

  • For every action, there is an equal and opposite reaction.
  • At any given throttle level, the motor generates torque which directly spins the propeller.
  • As the propeller spins, pressure differences cause the air to move from the front of the propeller to the back of the propeller.
  • We measure the torque created by a motor my measuring thrust.

_

Pitch Speed

When we say pitch speed, we mean the speed at  which the quad would be moving at a certain RPM and certain prop pitch (explanation of prop pitch here) if there were no prop slip.

What is prop slip? Think of a prop on a test bench. As the prop spins, it is going nowhere: the prop slip is at 100%. Since air is a fluid medium, the prop is able to push and “slip” through the air. Now think of the prop as you would a screw going into wood. As the prop spins, it moves forward and goes into the wood. Prop slip is at 0%. Since wood is (more or less) a solid, the prop can’t push aside the wood; the prop must move as it rotates. As we will see, we will always need some prop slip.

_

Thrust is Dynamic

Now back to the bench prop. We have our motor and prop spinning on the bench at 100% throttle, and therefore, max thrust. Lets come up with a few easy to work with numbers:

  • Pitch speed is 150 mph. **A good way to imagine pitch speed is to think of it as being the speed of the air directly behind the propeller**
  • Max thrust is 1500g

Next, let’s think about what happens if we have our test motor/prop in an air tunnel. With the wind tunnel, we can change the speed of the incoming air to the prop. What will this do? Since thrust is basically created by the difference in pressure of the incoming air ( directly in front of the prop) and the pressure of the outgoing air (directly behind the prop), the thrust will change accordingly:

  • If the air tunnel is at 0 mph, then the difference of the incoming/outgoing air  is 150 mph. Since 150 mph is 100% pitch speed, then the thrust is 1500g and the prop is 0% unloaded.
  • If the air tunnel is at 50 mph, then the difference of the incoming/outgoing air is 100 mph. Since 100 mph is 67% pitch speed, then the thrust is 1000g and the prop is 33% unloaded.
  • If the air tunnel is at 140 mph, then the difference of the incoming/outgoing air is 10 mph. Since 10 mph is 6.7% pitch speed, then the thrust is 100g and the prop is 93.3% unloaded.
  • If the air tunnel is at 150 mph, then the difference of the incoming/outgoing air is 0 mph. Since 0 mph is 0% pitch speed, then the thrust is 0g and the prop is 100% unloaded.
chart1
As our quad gains speed, the props produce less thrust.

_

Why Do We Care?

Efficiency. Although we will never see a 100% unloaded prop in the real world, we definitely want to be unloaded as much as possible and as often as possible. Why? Notice how the more unloaded the prop is, the less thrust is produced. Now remember about the equal and opposite reaction thing? That means less load is put on the motors which equals efficiency. How to do this?

Reduce air drag: There are many ways to do this, but to understand the relationship between unloading, air drag, and efficiency, consider the illustration below:

Prop unload
When thrust = air drag, our quad is at top speed and we no longer accelerate.

The horizontal component of thrust is used to offset air drag. Let’s say you are at top speed and travelling at 100 mph. Using the graph above, at that speed, each motor is producing 500g of thrust. This means that the total air drag is 2000g. What if we reduce the drag down to 1600g? Now each motor only needs to produce 400g of thrust each to offset drag, so if we look at our chart again, we should now see a top speed of around 110 mph.

Kind of makes you think a little… I’m going faster, but I’m producing less thrust!?!? A more in depth look in to drone physics is here.

Reduce weight: This is a very simple one. The lighter our quad is, the quicker it accelerates which unloads the props quicker.

Note that quicker acceleration is also a byproduct of reducing air drag.

Prop pitch: We all want to go fast so we like to have an aggressive prop on our quads. But in those instances where you know you won’t be needing a high top speed, use a milder prop. A milder prop usually means that the pitch speed is slower. A slower pitch speed means that for any given speed, the prop is unloaded more than if you were using a higher pitch prop.

Posted on

Choosing the Right Motor

With so many motors on the market, choosing the right motor these days is becoming tougher and tougher. My aim here is to make the choice a little easier for both beginners and experienced pilots. Although I will be going through the motor selection process for a 5-6″ speed quad, this process can be applied to any size quad. It can also be applied when looking for run time rather than speed (see analysis). It sounds like a straightforward process, but a balance has to be met. We can narrow down the motor choice to a few candidates, however there is still some fuzzy decision making (a lot of buts). We’ll start from the most general characteristics of motors and move on to more specific aspects.

Contents:

    1. The Process
    2. We Want It All
    3. Aren’t We Forgetting Thrust?
    4. Motor Numbers
    5. Does Size Matter?
    6. Early Summary
    7. Crunching the Numbers
    8. Analysis
    9. Conclusion

_

The Process

The only way to make a good comparison between motors is to have test data that comes from the same source and the same set up. Since there is no standardized test for generating motor performance, the numbers given by the motor manufacturers are useless. They can give us a general idea, but the comparison is not ideal.

Two great sources of motor performance numbers are:

  • Miniquad Test Bench – Tons of data, charts, etc.
  • EngineerX – YouTube videos of bench tests with links to spreadsheet data
    • NOTE: RPM data is very important! The spreadsheet links of EngineerX does not have the RPM data, but you can get it off the video.

When comparing motors, be sure to use numbers from one source. Also, the most important thing to remember is to compare data that uses the same propeller between all the motors.

For this post, I used the numbers from MQTB. I didn’t download all the data, just most of the more recent and popular motors (27 motors). Also, I only used the data from 100% throttle and 75% throttle. From there I combined the data in a spreadsheet, made a few calculations, and then sorted the data. From there, I could make a more educated decision. Before I get into the results, there is a little info to understand.

_

We Want It All

We want our motors to be quick (acceleration) and fast (top speed). What is limiting us from getting this? To be quick and fast, we need torque. It may come as a surprise to some, but brushless motors have gobs of torque regardless of the size or kv value. However, we only have gobs of torque if we can feed enough amps to the motors. This brings us to our true limiting factor: the battery. Not the voltage, but the amount of amps our batteries can provide. Too much amp draw kills batteries and kills performance:

  • Amps kill batteries by irreversible chemical reactions – the higher the load, the shorter the life of the battery.
  • Amps kill performance by reducing voltage (voltage drop) – have you ever noticed the voltage drop under hard acceleration? The higher the amp draw, the larger the drop therefore you lose precious RPMs

The battery problem

Having witnessed independent battery testing on batteries of various C ratings (C rating is explained here), it is easy to see that the C ratings from the manufacturer do not reflect reality. I have seen 30C batteries out perform 70C batteries (of the same mah). The best approach to choosing a good battery is to see what is popular or by word of mouth.

The bench test problem

Although the bench tests show voltage drops, they do not reflect reality and fall short of what we actually see when flying a quad. If a motor shows a voltage drop of 1.0 volts, multiply that number by 1.5 to get a more realistic voltage drop.

On the same note, bench test amp draw is much more than we see when flying. A good rule of thumb is to cut the amp draw on a bench test in half. But remember, once you cut it in half, multiply it by 4 to get the total amp draw of all 4 motors.

When it comes to what we want, it’s all about the amps… but we have to strike a balance – we need to be able to turn some high torque RPMs without dropping too much voltage and/or killing the battery.

_

Aren’t We Forgetting Thrust?

Although thrust isn’t what gives us speed, it still gives us a good comparison between motors using the same prop since the only way to produce more thrust with the same prop is by spinning it at a higher RPM. However, as we stated above, what price do we pay for this extra RPM? If it’s significant (more amps), then the extra amp draw will cause a larger voltage drop. This can actually cause your quad to go slower since the volts aren’t there.

_

Motor Numbers

First the basics. Most multirotor motors are defined by 2 sets of numbers such as 2205 2300kv. Here is a quick look at what the numbers mean.

Motor Size

The motor size is defined by the first set of numbers which is usually a 4 digit number in the form XXYY where XX is the diameter of the motor stator and YY is the height of the stator. For example, a 2207 motor has a stator of diameter 22mm and stator height of 7mm.

Stator
The green part with the spokes (the wire is wrapped around the spokes) is the stator.
Stator Side
The side view of the stator. The vertical height of the spokes (greenish gray rectangles) is the stator height.

The size of the motor can give us a general idea of the torque a motor can produce, but this comes at a price: amp draw and weight which we will get to later.

Motor kv

Motor kv value is defined by the 2nd set of numbers (easy to see since “kv” follows these numbers). The kv value of a motor is a guideline to tell us how many RPMs a motor will give us per volt. This number isn’t set in stone; typically these numbers vary +/-100kv.

Now we have an understanding of the numbers. Since I am concentrating on 5-6″ type quads, the motor sizes typically range from 22-24mm stator diameter with a stator height of 5-7.5mm.

_

Does Size Matter?

As we will see in the analysis section, motor size doesn’t follow a perfect trend; the difference between sizes is a little fuzzy. In fact, I was having a hard time finding any true trend in the spreadsheet numbers. However (with the motors I analyzed), there is a very loose trend: the larger the stator diameter, the more efficient the motor. That didn’t quite satisfy me though. If they are more efficient, why don’t we see all motors with larger stator diameters?

Stator Diameter

What was the drawback to a larger diameter? In an effort to find a trend, I went to the Data Explorer page on MQTB and scrolled down to the “Motor Comparison” graph at the bottom of the page. I chose 3 motors with a similar kv, but with a different stator diameter. Since I knew the performance of the motors differed once they reached max RPM, I was looking any difference in the time it took each motor to reach max RPM and there was very little difference. However, I did notice that the amp draw at startup differed greatly: the larger the diameter, the larger the amp spike at startup.

Chart-4

Stator Height

For the most part, this startup amp spike also held true for the stator height: the taller the stator, the larger the spike.

Also of note is that when comparing motors with the same kv value and same stator diameter, the taller stators have more torque which means that the motors RPMs are less affected by the air drag on the props: it won’t slow down as much and is able to handle more aggressive and larger props.

Chart-5

The thing to take note here on motor size is that the larger the motor, the larger the spike in the startup amp spike.  This isn’t as evident in the spreadsheet numbers as is the efficiency and RPM gain.This spike in amps probably holds true for acceleration too – just an assumption.

_

Early Summary

Not so much a summary, but more of a reflection on what we learned so far: not much in terms of choosing a motor. Although we now know that amps are the key, how do we know what is too much or what is worth sacrificing? The answer is still subjective. We can organize and crunch a few numbers (which we will do shortly), but even after that, it’s not clear cut. But it will definitely narrow down the field.

_

Crunching the Numbers

To start, we can go to MQTB and get some numbers from the motor test results page, or get the numbers from another source that has the thrust, amps, and RPMs. I chose a common motor prop and put the data into a spreadsheet. NOTE: I have a lot more numbers in there than needed; all that is required is thrust, amps, and RPM.

Spreadsheet-1

Next, I added 3 columns. The first column calculates the RPM per amp (simply RPM÷amps). Since we are comparing the same prop, it works for thrust per amp also.

The next one calculates the velocity of the prop (see a more detailed explanation here) at the given RPM. This is calculated by:

Prop pitch X RPM X 60 ÷ 12 ÷ 5280

The last column is just a copy of the amps just so I can easily see them next to the other 2 columns.

Spreadsheet-2.jpg

Next, you want to sort the data according to the RPM/amp. Using MS Office, this is done by highlighting all the data as shown:

Spreadsheet-3.jpg

On the data tab, click on the sort button:

Untitled-3

In the sort dialog box, select sort by (whatever column your RPM/amp data is in) and order values from largest to smallest and hit OK:

Untitled-4

Now the data looks like this. NOTE: I highlighted a few things:

Spreadsheet-6

A few things I highlighted:

  • In the last 3 columns, I highlighted the the best in each with green text and the worst with red text.
  • In the theoretical velocity column, I highlighted a few of the motors that stood out (to me at least! See the Analysis below): Light green are runners up and dark green is the winner.

_

Analysis

Although it is the most important attribute, there is more to motor selection than just the RPMs per amp. The next significant aspect is the theoretical velocity. You could have a large amount of RPMs per amp, but what if the motor is only drawing 3 amps? It might be efficient, but also very slow. So what I like to do is go down the velocity column and highlight the ones that have a nice bump in speed. Generally, it’s only worth going halfway down this column since, by the time you are halfway down, the amp draw is getting to be a little too much.

The Emax RS2306 2400kv, Cobra Champion 2207 2450kv, and EFAW 2407R 2500kv are all very, very close. The Emax falls short 4.5mph of the other 2 motors, but they require 10.2% more amps to gain 3.6% more speed. However, both the Cobra and EFAW have over 100g more thrust than the Emax. But still, they need 10.2% more amps to gain 9.2% (for the Cobra, 8.5% for the EFAW) in thrust. Very close.

Why the Cobra?

There is one more thing to consider. In the Drone Physics series, I talked about the role thrust plays. Since thrust only offsets the force of drag, thrust becomes less important the more aerodynamic your quad is. So, in summary, this is how I would choose:

  • For an aerodynamic quad, the Emax. Is it enough of a difference to ditch my Cobra motors? No.
  • For an everyday freestyle quad, the Cobras. And truthfully, the difference between the Cobra and EFAW is negligible so either one would be good.
2017_0731_00242600
The VXR-190 hit an average of 165.80 mph with the Cobra 2207 2450kv motors (verified by GPS)

What If I Don’t Care About Speed?

Then simply choose a motor that is higher on the RPM/amp column and don’t pay attention (or pay very little attention) to the theoretical velocity column. According to the data here, the F80 is the champion. But… in this case, it would also be helpful to compare motor weight since the Sunnysky is so close to the F80 in terms of efficiency. The Sunnysky weighs in at 29.9g while the F80 is 42g. But (#2) the F80 puts out 53g more thrust.

Weight

Weight should have very little influence on the decision. As seen above, some motors make up the difference, but it is never quite that that straight forward.

Physics: I will eventually expand on this so I’m keeping it light right now… Using Newtons second law F = ma, it can be calculated how much more thrust a motor needs to generate to compensate for added weight. That works out to be approximately 10g of thrust for every 1g increase in motor weight. Usually larger motors more than make up this difference.

What about the increased moment of inertia? This is non existent. Remember that the weight increase is in the motor, not in the frame. Why does this make a difference? Because the center of mass of the motor is in line with the center of thrust, no moment arm is created. Even if we take the center of thrust and center of mass of all 4 motors, they are still in alignment and no moment arm is created.

Update January 31, 2018: I am working on the moment of inertia issue. With linear acceleration, moment of inertia is taken care of with Newton’s 2nd law. However, angular acceleration is affected much more by the moment of inertia, more specifically, the weight distribution.

Response Time: Does weight influence response time of the motor? In most cases, no and in a few rare cases, barely. The benefits of the extra thrust and torque from the heavier motor out weighs any benefit of a light motor.

2018-01-01 15.48.15
Although the Bolt Worx is 7.1g lighter than the Cobra of the same size and kv, it has roughly the same response time but 150g less thrust. Thats a whopping 600g less thrust on a quad. Do you want to be 28.4g lighter or have 600g more thrust? Hmmm….

Further Analysis

Although this has nothing to do with choosing a motor, I wanted to find the trends in motor size vs efficiency. I decided to color code the motors according to stator diameter. It is not an exact trend (due to the different stator heights), but it is relatively easy to see the trend of higher efficiency with larger stator diameters.

Chart-6

The same was done for the kv value and the opposite conclusion can be made: the higher efficiency motors have a lower kv value. This is no surprise as this is generally a well known fact with brushless motors.

Chart-7

_

Conclusion

Although the motor choice was easily narrowed down, there is still a good deal of “educated guessing” when it comes to making a final decision. I will never say my choices are always correct since there are still many factors to consider. As it stands, I am actually going to stick with the Cobra’s for now (I don’t get any kick backs from them). There still might be a motor out there that isn’t in the MQTB database that will outperform it so I will just have to keep my eyes open.

Also, I plan on adding more and more motors to the spreadsheet I have made, especially since MQTB lags a little in terms of getting the latest motors tested – I’m not blaming him, it’s not easy doing that stuff, especially with a day job! 2 motors I want to see are the new Emax LS2207s and the new Cobra CPL2207s…

emaxls22072_2

Posted on

Quadcopters Have Hit the Sound Barrier

This isn’t a joke.

Since hitting crazy speeds with the VX1 and VXR-190, I have been doing a bit of thinking and observing. How were those builds able to hit 154.5 and 165.8 mph on a 5s lipo with 2450kv motors while the fastest quad on a 6s with 2750kv motors “only” managed to hit 145mph? Of course the VX1 and VXR-190 are much more aerodynamic, but something told me that there is something else going on to cause such a huge discrepancy. I will admit that hitting this speed was a shock, but careful number crunching and observing has revealed how it was possible.

The Sound Barrier

I am not sure how many in the multirotor community have done this, or even realize this, but it’s true: we are knocking on the sound barrier. Specifically, the helical tip speed of our props are going transonic. Transonic speeds (between mach 0.8 and 1.2) has traditionally been called the sound barrier due to the abrupt differences in the nature of subsonic air flow vs. supersonic air flow. When an object reaches speeds of approximately mach 0.8, localized regions of airflow will actually hit mach 1.0. This is due to the fact that the speed of the air flow around an object varies at different points on the object (example: air moves over the top of an airplane wing faster than the air below the wing).

So what?

They call it the sound barrier for a reason. Once transonic speeds are hit, the pressure differences between the super and subsonic regions start wreaking havoc and don’t play well together. This leads to vibration, distortion, and a skyrocketing drag coefficient which in turn requires a skyrocketing amount of torque. To sum it up: transonic flight not only requires loads of power, it is extremely unstable. This is EXACTLY what happened to WWII fighter planes and why supersonic flight was once thought to be unbreakable. However, all air flow is supersonic once an object is over (approximately) mach 1.2. From here, things begin to stabilize and the drag coefficient drops.

Helical tip speed

Example:

Helical tip speed = prop tip velocity + quadcopter velocity

If we observe during a speed run using a 5s lipo, the voltage drops to 15.3 at 154 mph using 5 inch props:

Prop tip velocity = 15.3v x 2450 RPM/1 minute x 5 inches x Pi x 60 minutes/1 hour x 1 ft/12 inches x 1 mile/5280 ft = 557.6 mph

Helical tip speed = 557.6 mph + 154 mph = 711.6 mph

To figure our mach number: 711.6 mph/761 mph = mach 0.935

As you can see, even with a 5s lipo, helical tip speed is well into the transonic region.

Propellers and transonic speeds

Why don’t we just chuck on another cell or 2, add more kv’s, and blow through the barrier? As noted above, the further out from the center of the prop, the faster its speed (angular velocity). The problem with this is that the propeller will always be in a state of transonic speed (unless, of course, the object it is attached to is going supersonic). The tip of the prop can hit supersonic speeds at relatively low RPMs compared to the root of the prop. So basically, once a propeller hits transonic speeds, we are doing no favors by having them go any faster. As a side note, the first time we encountered problems associated with the sound barrier were on airplane propellers (a propeller driven airplane cannot go supersonic since the prop cannot reach supersonic speeds). Here is a great link about the history of the sound barrier:

https://history.nasa.gov/SP-4219/Chapter3.html

Too Much Power

Of course we want the most amount of RPMs possible for speed. Of course, the easiest way to get more RPMs is to increase voltage (volts x motor kv = RPMs). Now we know that we can have too many RPMs. With a 2750kv motor and a 5 inch prop, we are already hitting mach 1 at 18.65 volts! Of course that is assuming the prop is fully unloaded. In reality we will have some RPM loss. However, we still haven’t taken into account the speed of the quad which adds to the problem.

The Significance

What does it all mean? Basically this means we are limited (for a given prop size) to a maximum RPM. This is extremely significant since this will enable us to mathematically figure out how much motor kv and lipo cells we need to obtain our maximum RPM. Anything more will only serve to decrease efficiency and overload our electronics – this is why it has been observed that adding more cells and motor kv has resulted in very little if any gain in speed.

Conclusion

It appears that we have more than enough power to spin our props to their physical limits which leaves us with 2 factors left that affects speed: prop pitch and drag. However, prop pitch is relatively limited too as most prop manufacturers only produce props within reason (if pitch is too high, the prop will stall at low RPM and you won’t be able to take off). Now we are only left with drag. Once optimum motor kv, prop diameter, and prop pitch are determined, I believe that the quad with the lowest drag coefficient will end up being the fastest. I’m hoping the VXV project will help solidify this conclusion.

Posted on

Smart ESCs? The Wraith32 35A

Update October 16, 2017: I will be re doing this review with the updated blheli32 32.1 firmware. Go straight to the conclusion to see why.

Contents:

  1. Smart ESCs?
  2. The Wraith32
  3. Blheli_32
  4. Comparison to Aikon/Spedix ES 30 HV
  5. Comparison to Super Racerbee 30A
  6. Size and Weight
  7. Blheli_32 Suite
  8. Raising the Bar on the World Record
  9. Flight Test/Video
  10. Conclusion

_

Smart ESCs?

Inherent in the Wraith32’s name is the fact that it is a 32 bit ESC. Gone are the days of the 8 bit EFM8BB21 MCU: we are now in the dawn of a new generation of ESCs that are using the 32 bit STM32 F0 ARM Cortex-M0 MCU. Although both MCUs run at 48Mhz, these ESCs 32 bit processing power will enable our quads to indulge in a new range of tools to enhance performance.

_

The Wraith32 35A ESC:

The Wraith is easily the best looking (in terms of quality) that I have seen. However, after installing a set, one of the ESCs just quit – only change I made was reflashing the fc. After A LOT of troubleshooting, the ESC was at fault.

Wraith32 Features:

  • Blheli_32
  • 35A with 45A burst
  • Dshot 1200
  • 3S-6S
  • 32 bit STM32 F0 ARM Cortex-M0 MCU @ 48Mhz
  • RGB LED
  • Current sensing resistor
  • Telemetry
  • Takes commands from flight controller (via dshot) such as make motors beep, enable bidirectional mode, (list will grow as I learn more)

Wraith32 02a

Wraith32 01a

_

Blheli_32:

Since the Wraith is a 32 bit ESC, you will have to use the Blheli_32 Suite to do any programming.

Blheli_32 Features:

  • Automatic motor timing
  • More precise control if manually setting motor timing
  • Set max motor acceleration (helps prevent desync)
  • Set max current
  • Smoother throttle
  • Better handling of spikes in throttle response

_

Comparison to Aikon/Spedix ES 30 HV:

Wraith32 Spedix 01
Top: Wraith32. Bottom: Spedix
Wraith32 Spedix 02
Top: Wraith32. Bottom: Spedix

_

Comparison to Super Racerbee 30A:

Wraith32 Bee 01
Top: Wraith32. Bottom: Super Racerbee
Wraith32 Bee 02
Top: Wraith32. Bottom: Super Racerbee

_

Size and Weight:

Wraith32 weight
Wraith32 35A
Super Racerbee 30 weight
Super Racerbee 30A
Spedix weight 01
Spedix ES 30 HV without heat sink
Spedix weight 02
Spedix ES 30 HV with heat sink

_

Blheli_32 Suite:

_

Raising the Bar on my personal record:

Since the VX1 is now in the bone yard, the Wraiths will be going into the VXT build. I wasn’t finished seeing what the VX1 can do and I’m hoping the VXT/Wraith/APC 5050s can break the 154.5 mph mark set by the VX1.

Project VX1 01
VX1

_

Flight Test/Video:

_

Conclusion:

I decided not to finish this review due to horrendous quality control… or is it that Blheli_32 has some bugs to work out? Either way, out of EIGHT of the wraiths, I had THREE ESCs that were bad. The last straw was when one of the FETs burned up on my 4th flight – I wasn’t under hard throttle, I didn’t crash, etc. Rather than replace the wraith, I decided to ditch them for something more reliable. I may have to revisit these another time.

Posted on

Mobius Mini: The Best FPV Racing Action Cam?

Is it the best?

In the quest for designing more and more efficient quadcopter frames, I really disliked strapping on a huge, bulky action cam air brake to get video. I gave the Mate 808 a shot and wasn’t too impressed which I expected knowing what the specifications were. However, I decided to take another look at the Mobius Mini and was pleasantly surprised when I found out it was the same size as the Mate 808. So how does it measure up?

Contents:

  1. First Impression/Quality
  2. Features/Specs
  3. Size and Weight
  4. Configuration
  5. Screenshot Comparison to Mate 808
  6. Screenshot Comparison to DBPower Action Cam
  7. Flight Screenshots
  8. Flight Videos
  9. Conclusion

_

1. First Impression/Quality

I was impressed with the packaging and the overall quality of the camera. You can definitely tell from the weight of the camera that they put everything into as small of a case as possible.

Mobius Mini 02

Mobius Mini 03

Mobius Mini 09
Included accessories

Mobius Mini 05

Mobius Mini 04
Shutter and Mode buttons

Mobius Mini 06

Mobius Mini 07
Lower right hole is the reset button “… in case the camera no longer reacts…”
Mobius Mini 08
Mounting holes for the tripod mount

_

2. Features/Specs

A few of the specs that I like about this cam. Check out the configuration section to see all the available settings.

  • CMOS image sensor
  • Up to 1440p @30fps
  • Up to 1080p @60fps
  • Records in H.264 format
  • Includes adapter to add external power
  • Adapter also outputs video (can connect to a vtx)
    • 60ms video latency @1080
    • 40ms video latency @720
  • Mode button to switch between 2 different video recording configurations
  • Compact and low profile

_

3. Size and Weight

I measured the dimensions at 55mm x 29mm x 14mm and it tipped my scale at 26g.

Mobius Mini 10

Mobius Mini 11

Mobius Mini 12

_

4. Configuration

To configure the Mobius Mini, the sofware has to be downloaded from here. The Mini uses the same software as the Mobius 2.

Also, the pdf manual can be found here.

Once the Mobius was plugged in, I opened up the software and it immediately told me there was a firmware update (Highlighted in red at the bottom of the screen).

Mobius Mini Config 01

Mobius Mini Config 05
Tools window. Firmware downloaded and ready to install.
Mobius Mini Config 06
This warning needs to be updated…

Once I reconnected the camera and saw the LEDs blinking, I left the computer and returned about 5 minutes later. I am not sure if the camera turned off during that time, but it was on (LED was green) and the warning box was still open. I checked to see if Windows recognized the cam as a drive. Nope. After about 5 minutes of more nothing, I went ahead and unplugged it. Everything was fine, I didn’t brick it.

Screens of all the configuration tabs:

Mobius Mini Config 09

Mobius Mini Config 10

Mobius Mini Config 11

Mobius Mini Config 12

Mobius Mini Config 13

_

5. Screenshot Comparison to Mate 808

The quality is leaps and bounds beyond the Mate 808.

NOTE: All images on the left are the Mate 808 1080p @30fps. Click the image for full size:

Mobius Mini Mate 808 Compare 2
Mobius Mini 1080p @ 60fps
Mobius Mini Mate 808 Compare 4
Mobius Mini 720p @60fps
Mobius Mini Mate 808 Compare 3
Mobius Mini 720p @120fps

_

6. Screenshot Comparison to DBPower Action Cam

Well… I didn’t realize how beat up my DBPower action camera was until I took a close look at the image capture. The only usable portion of the image (for doing a comparison) is the far right side due to misalignment of the lens. However, exposure and color can still be compared.

The image is a bit darker on the Mobius, but it is actually a lot more accurate in terms of the lighting and color for the late evening.

DBPower on the left 1080p @30fps and Mobius Mini on the right 1080p @60fps. Click the image for full size.

Mobius Mini DBPower Compare

_
7. Flight Screenshots

Screenshots from a sunny video (click for full size):

Mobius Mini Still Sunny03

Mobius Mini Still Sunny02

Mobius Mini Still Sunny04

Screenshots from a cloudy video (click for full size):

Mobius Mini Still Cloudy05

Mobius Mini Still Cloudy04

Mobius Mini Still Cloudy06

_

8. Flight Videos

_

9. Conclusion

So far, I am extremely happy with what I have seen so far. I was under the impression that the video quality would suffer as a compromise for a low profile design, but this has surpassed my expectations.

Another great thing about this camera is that it has proven to be very durable – I had 2 rough crashes (as seen in the video) and it was still ticking. Can’t say that about my frames though 🙁

If you are looking for a lightweight action cam that won’t act like an air brake, has great quality, and is durable, the Mobius Mini is the way to go.

Posted on

EMAX Bullet Series 20A ESC

Preflight:

Emax just recently released their latest line of ESC’s – the Bullet Series. I originally came across these ESC’s on Banggood as a pre-order item. However, opting for faster shipping, I decided to purchase them from Piroflip (they even had them available the same day and price as Banggood).

I’m finding out that many ESC’s are fairly equal lately in terms of size, price, and weight. The Bullet Series has everything you could want going for it:

  • Oneshot125
  • Oneshot42
  • Multishot
  • Dshot
  • Smallest 20A ESC (that I know of)
  • 30A Burst
  • 3.5g (without wires)
  • Heat sink (dual purpose)
  • 2S-4S

Packaging:

The Bullet Series ESC’s come packaged with the power wires (20awg) and signal/ground wires already soldered. It also comes along with three 50mm long 20awg motor wires that are not soldered to the ESC, but already stripped and pre-tinned. Most motor wires are long enough to reach the ESC and time is saved by not having to remove pre-soldered wires off of the ESC. The package also includes shrink tube to cover the ESC once the motor wires are soldered.

emax-bullet-20a-08
Packaging with yellow sticker denoting which Blheli_S version it is flashed with

Genuine Emax:

The packaging also comes with a sticker that has a scratch off area that reveals a code. If you visit the Emax website, there is an anti-counterfeit link near the top of the page. Enter in the code to verify the ESC is a genuine Emax product.

emax-anticounterfeit-02

emax-anticounterfeit-03

Heat sink:

This is the main reason I bought this ESC since my builds typically don’t allow for much air flow. However, I did notice that the heat sink is rather large – it covers all the MOSFET side electronics. The reason, according to Emax, is that it will help to protect the electrical components in the case of a crash. I can think of at least one instance when this would have been helpful where my son crashed his quad into a tree – one of the Littlebee ESC’s was destroyed due to a propeller strike.

emax-bullet-20a-09
The heat sink not only cools, it also protects

Size:

I thought the Spedix ES 20’s were small but the Emax ESC’s are even smaller (although not by much). I measured these 2 ESC’s with calipers along with the Littlebee Pro 20a’s and got the following measurements:

  • Littlebee: 24.6mm x 12.6mm (310mm²)
  • Spedix: 24.4mm x 10.7mm (261mm²)
  • Emax: 19.6mm x 11.9mm (233mm²)

The Spedix is 16% smaller than the Littlebee while the Emax is 25% smaller than the Littlebee.

emax-bullet-20a-littlebee-pro-20a-mosfet-side
MOSFET Side: Emax top, Littlebee bottom
emax-bullet-20a-littlebee-pro-20a-mcu-driver-side
MCU/driver Side: Emax top, Littlebee bottom
emax-bullet-20a-spedix-es-20-mosfet-side
MOSFET Side: Emax top, Spedix bottom
emax-bullet-20a-spedix-es-20-mcu-driver-side
MCU/driver Side: Emax top, Spedix bottom

The project:

The project I’ll be using these on is the C3 Project which is a completely sealed, weatherproof frame. The frame will be like the C2 project except the body tube will be aluminum tubing (0.9mm wall thickness). This will allow the Bullet ESC’s to be mounted inside with their heat sinks in contact with the aluminum body so they dissipate heat in the absence of air flow.

Flight Test:

Unfortunately, I don’t have any video (worthy of posting) of the first few flights with these ESC’s but I will say that they definitely worked as intended for the C3 project and they are extremely smooth. Truthfully though, ever since blheli_s came out, I have not been able to notice any difference between different manufacturers. These are the only ESC’s that I have run with dshot so far. In the little bit of flying I was able to do, I haven’t been able to tell the difference with dshot either. However, even if I can’t tell the difference, I know the FC is getting a cleaner signal which helps everything in terms of accuracy and efficiency.

I Killed Them:

I outline the whole story on the C3 Project page…

2017_0217_20545700

Posted on

Make Your Own “Snowman” Type Antenna

You Already Have One

That is, if you have ever bought a video transmitter. Most (if not all) VTX’s already come with an antenna – the infamous “rubber duck” antenna. I frown upon the idea of decieving people and selling something they already own so I’ll show you how to transform your rubber ducky into a snowman.

What’s the Appeal?

Although a cloverleaf circular polarized antenna works better overall, FPV racers (or anybody that flies around a relatively small and unobstructed course) don’t need a cloverleaf antenna. It doesn’t hurt, but the appeal of a snowman type antenna is that it is small, lightweight, flexible, and nearly indestructible – it won’t get caught on racing gates and it won’t get bent up (race tested and approved by Daniel “thatcharacter” Harvey).

Step 1:

You will see a seam about 1/3 of the way from the base of the antenna. Carefully bend the antenna here to pop the top off.

snowman-antenna-02

snowman-antenna-03
The “naked” antenna

Note: The top white part of the antenna is the part that transmits the 5.8ghz frequency. It is made precisely to a certain length and care must be taken not to cut or damage it. The lower part (the metal shield and braided wire) must not be damaged either since these parts shield the lower part of the antenna.

Step 2:

Carefully cut the pivot pins with an exacto knife.

snowman-antenna-04

snowman-antenna-05

Step 3:

This is the “hard” part. Remove the top plastic piece by using a pair of wire clippers or a dremel. I have used wire clippers with no problems, you just need to take your time and be careful.

snowman-antenna-06

Step 4:

This naked antenna is ready for action but it could use a little protection. Use 3/16 inch shrink tube for this. I also used a little bit of 1/4 inch shrink tube at the base.

snowman-antenna-00
At least I know my antenna will survive…
Posted on

Aikon Clone??? The Spedix ES 30 HV ESC

Contents:

  1. The Spedix ES 30 HV
  2. Speedy Spedix
  3. Comparison to Racerstar V2
  4. Comparison to Aikon
  5. What Does BLHeli Think?
  6. Dshot
  7. Weight
  8. Small Modification
  9. Flight Test/Video
  10. Conclusion

Since my Racerstar 4 in 1 30A was unexpectedly ruined in a crash, I was on a quest to find replacement ESCs for the SK1/SK2 project that could handle up to a 6s battery. This time I opted not to get another 4 in 1 esc…

_

The Spedix ES 30 HV:

I am not sure what had pushed me into it (maybe it was the $12 price), but I ran across some Spedix ESCs at pirofliprc that I decided to take the plunge on. I couldn’t find any reviews or flight performance information on these ESCs other than a hand written note on my shipping invoice from Rich at pirofliprc: “Been flying these on 2700’s. They Rock!!”

Spedix ES 30 HV Features:

  • Blheli_S
  • Dshot ready
  • Can handle 3S-6S
  • EFM8BB21 MCU
  • Large MOSFETs
  • Heat sink
  • Enhanced filter circuit

Update: Unrelated to this purchase and no need for the details, but I want to show my appreciation for pirofliprc‘s customer support/service. They definitely went above and beyond my expectations.

_

Speedy Spedix

Just recently, the Spedix ESCs hit a new high. In combination with the Cobra 2207 2450kv motors, they were able to help the XLR project hit a new world speed record with an average speed of 195.99 mph (315.42 kmh) and a top speed of 202.11 mph (325.26 kmh).

0005r6
The world’s fastest quadcopter uses the Spedix ES 30 HV ESCs

_

Comparison to Racerstar V2:

I decided to do some comparisons to the Racerstar V2 (Sunrise Cicada) 30A ESCs. Although they can’t handle more than a 4s, it was the closest ESC I had.

A side by side comparison shows the larger sized MOSFETs on the Spedix. Also, the Spedix has many more low ESR capacitors which, according to Spedix, is due to the filtering circuit being much more robust. I am also assuming that it is due to the fact that it can handle higher voltages.

I also noticed that the Spedix has a bus bar (long metal bar on the Spedix in the bottom pic) which helps to lower resistance.

spedix-30a-01
MOSFET side: Racerstar V2 top, Spedix bottom (heat sink removed)
spedix-30a-02
MCU/driver side: Racerstar V2 top, Spedix bottom

_

Comparison to Aikon:

These Spedix ESCs are extremely similar to the Aikon SEFM 30A 6S ESCs. Aikon’s claim to fame was that they were the first to introduce an ESC  with hardware generated PWM that made it capable of using Blheli_S.

Other than a few minor differences in the layout, these 2 ESCs are nearly identical. The main differences are:

  • Aikon ESC uses Infineon Technologies IRFH5300 MOSFETs
  • Spedix ESC uses Toshiba TPH1R4 MOSFETs
  • Aikon ESC uses 3 separate dedicated MP1907 gate drivers
  • Spedix ESC uses an FD6288 gate driver which is actually 3 gate drivers combined on one chip
aikon-spedix-1
MOSFET side: Aikon SEFM top, Spedix bottom (heat sink removed)
aikon-spedix-2
MCU/driver side: Aikon SEFM top, Spedix bottom

_

What Does Blheli Think?:

Once I soldered one of the Spedix ESCs to the SK2, I plugged it into the PC and ran the Blheli Suite to find out what it would see. After a warning about the ESC having an outdated 16.3 Blheli version, the Flash Assistant window popped up showing that the Spedix was flashed with the C-H-15 hex which is the same hex used on the Aikon. However, I also know that the C-H-25 hex is used for Aikon V2’s. The -15 is for the V1’s. I am not sure what the difference is between the two versions, so I went ahead and updated with the C-H-15 hex as suggested.

blheli_s-01
Blheli Suite Flash Assistant with C-H-15 hex selected
blheli_s-04
Default settings after flashing to 16.4

_

Dshot:

UPDATE: I just ordered a set of 20a Spedix ESC’s for one of my winter projects (Project C1) and noticed that the pirofliprc page says they are dshot ready. In fact, all of the Spedix ESC’s from piroflip are now dshot ready. I received my shipment today and the ESC’s do have the smoothing cap removed. As a side note, these 20a ESC’s are incredibly small!

With dshot on the horizon, which is a new digital ESC protocol, I found it nice to know that Betaflight testing is being carried out on the Spedix ESCs and that it is confirmed that the dshot600 is working on the Spedix HV 30A. Dshot is exciting because, since it is digital, it will be much more accurate and produce less jitter. It’s also interesting to note that you will no longer have to calibrate your ESCs. More information on this new protocol can be found here.

If you are brave enough and want to try out the Spedix ESCs on the Betaflight 3.1 Alphas, you will need to remove the smoothing capacitor on the signal input to the ESC. Smoothing caps are great for reducing analog noise, but not wanted for a digital signal. A video on how to remove the cap can be found here.

Also, be sure to see if your flight controller/ESC combination has been verified in testing – or be brave once again and be one of the testers. FC/ESC combo list can be found here.

_

Weight:

Since I don’t own an Aikon ESC, I compared the weight to the Racerstar. As expected, the Spedix was a bit heavier than the Racerstar. Not only does the heat sink make it heavier, the Spedix board has many more (and in some cases larger) components. Also, the Spedix is a hair larger at 15mm x 29mm compared to 14mm x 28mm.

Note: Not the most accurate scale is being used; only 1g increments… good enough for milk and water though.

spedix-30a-03

A bit of shrink wrap was added to the Racerstar measurement to account for the shrink wrap that was removed

racerstar-30a

_

Small Modification:

Since these ESCs will be sitting inside of the body on my quad, I decided to cut some of the shrink wrap off the heat sink to expose it directly to the air.

spedix-30a-04
Modified Spedix shrink wrap

_

Flight Test/Video:

Making it short and sweet: No issues with these ESCs, smooth as butter, and work awesome with air mode since the min throttle setting can be set extremely low (lowest out of the ESCs I have owned so far). 

_

Conclusion:

Although I’ve only had about 5 battery packs on these ESC’s, they feel just as smooth (if not smoother) than any other ESC I have flown. For the price and features, I highly recommend these ESC’s!

I now have a set of the Spedix ES 20 Lite ESC’s that will be thrown on the C1/C2 project. Once the weather cooperates (and project issues are addressed), I will put down my thoughts on those as well.