Propeller and Motor Balancing

One of the biggest hindrances to the success of any multi-rotor system is unwanted vibration.  Vibration interferes with the delicate sensors on your controller board and can lead to unpredictable and possibly dangerous behavior from your multi-rotor.  In addition it makes for bad video and photos; this is especially the case when using cameras such as the gopro or any other camera which uses a CMOS sensor.  CMOS sensors write image data one pixel at a time starting from the top left of the sensor and continuing across in rows until the entire image is captured.  This is done on average about 30 times per second and hence is not visible to the naked eye.  Under standard conditions CMOS sensors can achieve beautiful and fluid video and photos.  But when vibration is introduced to the cameras it leads to a “Jello” effect on the video that makes the video unusable.

Below is an example of “Jello” video.

Vibrations also cause unnecessary wear and tear on your equipment and can loosen components.  So it is very important to minimize the causes and effects of vibration on your multi-rotor.

There are a couple factors that lead to excessive vibration and they are mostly motor and propeller related.  The first of these factors we will be looking into is motor vibration.  Most beginners in multi-rotor assume that the motors that they buy from their local retailer come pre-balanced.  I have in my experience found that more often than not that is not the case.  This has even been true with motors that are advertised as balanced and were more expensive than similar models.  So our first step in removing vibration from our multi-rotor will be balancing our motors.

The most efficient way I have found to do this is with the motors already mounted on the multi-rotor frame with the ESCs connected either to a controller board or to a servo tester that you can use to control the motors speed.  You will also need some way to detect and visualize the vibrations.  This is where your smart phone comes in handy.  Because most modern cell phones have three access accelerometers built in to them there are free software solutions that will help you quantify the vibrations much like a seismograph.  I have an android base phone and I use Seismos which is available in the Google Play store.  If you have an Iphone there are many solutions available to you as well.

Step 1 The Rig

_WDG9771

As you can see, we have assembled all of the aforementioned components into a rig that makes it easy to run the motor at varying speeds and determine how much vibration is being produced.

The first step is to use the servo controller to slowly run the motor up to full speed while running Seismos on your smartphone.  You will notice that there is not a direct linear relationship between the speed of the motor and the vibration produced.  In fact it varies by RPM and every motor will be different.  What is important is running the motor up and down several times until you have a good idea, on average, of how much vibration are being produced by watching Seismos and keeping an idea in your head.

Step 2

Look at the motor from the top down and create a cross section in your mind.  I usually split the motor into quadrants.

MotorQuadrant

 

The next steps are trial and error.  You need to cut a small square of electrical tape about the height of the bell of the motor.  Then choose one of your quadrants and apply the tape to that side.  In the photo above that would be where one of the lines intersects the motor bell.  Also as shown below.

_WDG9780

 

Once you have your tape applied, run the motor up to speed again and see if the vibration has decreased.  If it has not, move the tape to another quadrant and test it again.  Continue in this manner until you have found a location for the tape that decreases the vibrations the most.  When you find the best spot, you have basically found a place on the motor bell that is lighter than the rest.  You might also want to try adding another piece of tape on top of the one you already have applied just to see if the added weight decreases vibration even further.

Once you have done this for all of your motors you are ready to move on to the next task which is balancing your propellers.  There are different sorts of do-it-yourself prop balancing methods out there but to get the best results you need a proper propeller balancing tool such as the Top Flight Propeller Balancer.

_WDG9782

This type of propeller balancer gives you the best results.  This is because the shaft that you mount the propeller to is suspended between two magnets.  This allows the propeller to move back and forth with almost no friction.

_WDG9785

 

Once you have placed the propeller in the balancer, wait until the propeller has settled and stopped moving.  If the propeller is vertical that means that the blade that is at the bottom is heavier than the blade at the top.  In order to fix this we need to either remove material from the bottom blade or add material to the top.  I use the subtractive method and will remove the propeller from the balancer and lightly sand the back side of the bottom blade.

_WDG9787

I will then place the propeller back in the balancer and I will continue this process until the propeller is balanced.  You will know your propeller is balanced when no matter what position you place the propeller in the balancer the propeller will stay in that position and not move.

Very often it is possible to get the blades balanced but the propeller will still not be.  This is because the hub of the prop may also be unbalanced.  If your propeller will rest in the horizontal position parallel to the table surface then your blades are most likely balanced.  To determine which side of the hub is heavier you can place the propeller vertically in the balancer and then let it go.  Generally which ever side of the hub comes to rest at the bottom is the hub that needs some material removed.

This is a process which can be quite tedious and requires patience and practice to get good at it.  It is, however, of paramount importance as you build and fly your multi-rotor – this cannot be overstated.  Spend a lot of time balancing your motors and props.

Additional Field Work and Processing

After the multi-rotor was repaired, which was a small matter, I began to process the images.  Capturing the photos is just the beginning of this process and the tip of the iceberg at least as far as time is concerned.  Processing the images in Photoscan involves three steps each of which has several sub steps.  The first step matches the photos and creates a point cloud.  The second step generates a polygon mesh and the final step creates and applies a skin to the mesh.  Each step has different quality settings and they can be either very processor intensive or very memory intensive in the case of the mesh generation.  You can see an example of the work flow below.

I initially tried to process the images all in one chunk.  There were about 1800 images to process at the time and after several days of processing I had a big disappointment.  The point cloud ended up being just a cloud of random points.  In order to get all of the photos to align correctly with the processing power available to me would require me to break the photos up into smaller chunks.  The chunks would later be merged.  This process worked much better and after many days of trial and error I ended up with a point cloud that ended up looking very encouraging.

Point_Cloud

While the point cloud is being calculated by the software it also extrapolates the location of the camera for every photo processed.  Being as half of the photos were shot from the ground and half were shot from the multi-rotor, the camera locations would give me a good indication of how steady the multi-rotor had been during the flight and whether or not there was adequate coverage.  Here is one view of the flight.  The blue boxes represent the locations at which each photo was taken.

Point_Cloud_Cam

As you can see, the top two rows of blue boxes represent the photos taken by the multi-rotor.  There appears to be a very consistent distance and height between each photo.  There also appears to be very good coverage and overlap.  Here is another view.

Point_Cloud_Cam_Alternate

 

The black lines through the boxes represent the angle the camera was at when the shot was taken.  Here is the model after the mesh has been created and the model has had the skin applied.

Textured

So with these encouraging results I have continued to process and merge chunks with the goal of modeling the two main parts of the structure and then joining them.

More to come..

Field Test

On the weekend of May 18th I set about conducting my first test on the trolley building.  That Saturday I, with the assistance of my brother, photographed what we could of the building from the ground.  The following day at around 7:30 am we began capturing photos with the multi-rotor.

_MMU2908

 

Much of this process is still new to me and the proper methodology for taking the photos from both the ground and the air so they might work optimally in Photoscan is still very much a work in progress.

_MMU2860

 

There are many things to consider.  Such as the best distance from the target object and the best angle for the camera to take the photos from.

A compact digital camera is used to take the photos from the multi-rotor.  Rather than using a servo to actuate the shutter button I opted for a software solution.  The problem with using the servo actuator arm is that for every image that needed to be taken the operator of the multi-rotor would have to flip a switch on the transmitter.  Having to take the photos in this way makes for added distraction while piloting the multi-rotor.  As I am using a Canon Powershot Camera I was able to find a better solution.  There is a custom firmware, CHDK which is available here.  It allows one to non permanently replace the factory firmware with a much more powerful version.  Along with many other features it allows one to shoot in RAW and to run custom scripts.  I have used the custom script feature to make the camera automatically take images in about 2 second increments.  So when this is employed on the multi-rotor one of the things that is very important to get right is the pacing at which the multi-rotor is moved.  As it is very important to have as much overlap as possible in the photos.

_MMU2880

All in all, the day was pretty successful.  But as the morning progressed, winds began to gust between 15 to 30 mph.  These conditions are not ideal for the multi-rotor and may cause erratic flight behavior.  I decided to cut the day short so that I might prevent a crash.  But not before I tried out the on board video broadcast system with my video goggles i.e. first person view flight.

_MMU2914

 

The image above is a light weight video ground station which receives the video signal from the multi-rotor and rebroadcasts it to my video goggles.  This system has been used before with other aircraft successively at ranges of up to 2 km.  Unfortunately this was not the case for this test.  For some reason after climbing to about 10 meters, the video signal cut out completely which resulted in a bit of a hard landing.

_MMU2918

 

Luckily only one arm was damaged and it can be easily replaced.

Next is processing the images we took.  The next post will cover this.

Images in this post courtesy of Mariano Ulibarri.  Many thanks to he and Tyler Grassie for all of their help.

Initial Testing

Once I had the new gear installed I was able to install and configure the camera gimbal.

ChopperCopter

 

With all systems tested and functioning correctly the multi-rotor was ready for field tests.  I decided to use the old trolley building on the campus of New Mexico Highlands University as my first subject.  This building is significant in that it housed horse drawn trolleys which were used in Las Vegas around the turn of the 20th century and the second being that the shell of this building will serve as the foundation for the future home of the Media Arts department.

This allows for an opportunity to record with photogrammetry the renewal of this building into a modern instructional facility.  I hope to model the building in its current state and after its renovation.

Parts of this structure are over two stories tall which makes for a perfect opportunity to test my multi-rotor as it has been designed for just such a subject.  It will also be a good test for determining how small of an area the multi-rotor will be capable of functioning in.  There are power lines, trees, and guide wires relatively close to the structure.

I will post videos, photos, and hopefully preliminary models of the tests soon.

Engineering Issues

In the coming days I will post some time lapse video of the actual multi-rotor frame build.  The build went very well and I didn’t have any problems.  But when I added the gimbal to the bottom of the multi-rotor I soon noticed a pretty big problem.  The landing gear which I had envisioned using, and had already purchased, was very obviously not going to work.  It was purchased for its price and I had used it in a previous build with some success.  You can find it here.  I found that when the gimbal was mounted to the frame of the multi-rotor there was just no viable place to attach the landing skid mounts.  So I had to scrap them and figure out another way to accomplish the same goal.

While researching a solution I came across a YouTube channel made by a gentleman who used a similar multi-rotor frame to the one I was using and he had come up with his own solution.  He used Sketchup and a Printrbot to make his own leg extensions.  His designs, though promising, proved to be too flimsy and short for my application.  This gentleman does, however, have his own forum and after searching around it I came across another gentleman who had made a much more robust design that snapped and bolted together.  Not only that, but he had made it available on Thingiverse.com.  I downloaded his models and soon realized that the design was for legs that were about 13.5 CM in length.  I needed something closer to 20 CM.  Not having the knowledge to modify his design I found his YouTube page and sent him a private message asking for his help.

His name is Gavin Bendtsen and he was kind enough to email me back.  Not only that, but he went out of his way to modify his plans to fit my application and had them back to me by the next day.  A soon as I got the plans I went to work.  With the help of Mariano Ulibarri I got the legs printed and they came out very nice.

_WDG8625

I went about mounting them to the multi-rotor frame and found that I only need to make a few small modifications.  I had to use an old soldering iron to melt several small indentions at the top of each leg piece to accommodate the motor mounting bolts.

_WDG8640

 

With the modifications done I began fitting them to the frame.

_WDG8644

 

Here is one leg finished.

_WDG8646

 

All finished.

_WDG8666

 

The legs seem flexible yet very sturdy and provide more than enough height to allow for the gimbal to hang underneath.  I am very thankful that Mr. Bendtsen was able to assist me in this endeavor as there is no product like this on the market.

This collaboration also shows a very important part of what makes any project like this successful. That is with help and collaboration from the community as well as the rise of at home design and manufacturing, one can accomplish just about anything.  May the makers of the world continue to unite and thrive.

 

More On The Gimbal

Here are a few more details regarding the camera gimbal build video I posted previously.

An important part of this project is making sure that the multi-rotor is built at minimal cost yet remains of substantial enough quality to perform all needed tasks.  That being said, finding a high quality gimbal at a low price can be very hard.  Gimbals range in price from $40 dollars into the thousands.

After much research I came across a servo controlled 2 axis gimbal that looked like it could easily be modified to mount on the frame I selected for the build.  It was also very inexpensive – around $37 dollars.  The gimbal can be found here.  they have several nice pictures of the fully constructed gimbal.  It is important to note that servos were not included nor did they state what size of servo would be needed.  When it arrived it looked a little less constructed.

_WDG8515

 

There were also no instructions included.  Though I shouldn’t complain.  I wasn’t certain I would receive the order at all being as the vendor’s name is Goodluckbuy.  So I started putting things together using the photos on the vendor’s site.  Slowly it started taking shape.

_WDG8519

 

Things went mostly according to the plan my mind had made, though i did struggle with some quality issues.  The ball joints and bolts designed to be the actuating arms didn’t actually fit and had to be tapped and glued together.  But that was about the only snag I came across at this point.

_WDG8521

 

After several hours and lots of tinkering I had something that resembled the completed product.

_WDG8523

 

You can see the aforementioned arms in this rear view.

_WDG8531

 

All in all the build went very smooth and mounting it to the multi-rotor frame was very simple.  Just four holes 3.5 mm in diameter and four M3 bolts with locking nuts.

For the price, I am so far very satisfied with this gimbal and I am fairly confident it should get the job done.

Proof of Concept

Not long after returning from Cuba I began testing my concept.  I had only recently started experimenting with quad copters and had a working one which I had built several months before.  On it I had installed a mount for a Gopro camera.  The Gopro is capable of taking series of timed images and has a 7 mega pixel sensor and as such should be perfect for capturing images for this project.

Before I conducted my first test I did some reading online about using Gopro cameras for this type of application.  There were limited sources and from what I could find people had used them with some success or stated out right that it would not work.  The latter reasoning that the extreme fish eye, variable exposure, and CMOS sensor would render it useless for this application.

Despite this I pushed forward with my tests.  My first subject being my house.

DCIM100GOPRO

The experience taught me many things.  Most importantly that my concept could work and could work in tight spaces.  I also found that using Agisoft Lens I was able to successfully create a lens calibration XML for the Gopro which meant that the collected images could be processed using Agisoft Photoscan.  I processed the images on the lowest settings and the results were encouraging enough to move forward.

ProofOfC

 

It is important to note that this was processed on the lowest settings with very few photos.

So with some lessons learned I pressed onward.  I realized that the multi-rotor I used for this test was not ideal for several different reasons.

  • It is heavy and because of this it had a flight time limited to about 7 minutes of safe flying.
  • The motors and the speed controllers which they attached to are not ideal.  The former being not well balanced and the latter being not as responsive as I would like. The importance of balance and responsiveness will be explained later.
  • It was unstable and a lot of this can be attributed to my relative inexperience with the controller I was using but also to the vibrations which existed in this multi-rotor as tested.
  • The use of the Gopro, though successful, might not prove ideal and the camera mount on the multi-rotor would not accommodate cameras of any different size.
  • Finally during later testing this multi-rotor crashed and was basically destroyed.

These early tests and the information and experience I gathered from them shaped the idea of where I wanted to go from there and what I wanted to accomplish.  I would build a new multi-rotor and it would be designed to be higher quality, lighter, and more effective while keeping price to a minimum.  It should feature:

  • A lighter and more resilient frame.
  • Higher quality components.
  • A servo stabilized 2 axis camera gimbal.
  • The ability to carry a larger variety of cameras i.e. Gopro or compact digital.
  • The ability to remotely trigger the camera shutter.
  • Flight times of over 10 minutes on one battery.
  • Tuned and effective controller for the multi-rotor.
  • The ability to broadcast live video from the multi-rotor to either a ground station or video goggles for first person flight.

With this in mind I started researching parts and kits.  The results of which I will be posting as the project continues.

 

The Build

For the most part the multi-rotor I will be using has been built and in coming weeks I will be posting video and pics of the whole process.  I will be discussing a range of topics from, parts needed, back ground of the hardware used, theory, and best practices.  I will also discuss engineering issues and how I overcame them.

The idea being that when I am done I will have an aircraft that is stabilized by an Arduino based controller board.  The aircraft will feature a 2 axis, servo stabilized, camera gimbal capable of carrying a compact digital camera.  The camera shutter will be triggered by an additional servo mounted on the gimbal and controlled by the transmitter.  The multi-rotor will also feature an additional camera that will transmit live video to a ground station which will in turn broadcast to a pair of video goggles so that the aircraft might be flown as if the pilot is sitting in the aircraft.  This allows much more precise flying.  The aircraft will also broadcast live telemetry information to a laptop which acts as a flight recorder and will give the pilot the ability to tweak configurations on the fly.

If all of these systems can be made to work without error this multi-rotor will be a valuable tool for gathering photo based data to be used in the application of photogrammetry on a large scale.

The Project

Based on my epiphany in Havana I have endeavored to achieve the following.

  • I will build, from parts only, a multi-rotor remote controlled aircraft as a stabilized camera platform.
  • I will use traditional photography in addition to the multi-rotor to capture images of large scale structures.
  • These images will in turn be used to render textured 3D models  of said structures.
  • I will document this entire process.  Including the multi-rotor build, multi-rotor calibration and programming, use in the field, and image processing.

I hope to use this process and resulting information as a means to inspire and educate others who are interested in using these tools themselves.  I believe that this system if properly implemented will have a large impact on many fields such as conservation, restoration, cultural heritage, and many others.

I will also endeavor to make the project accessible as possible in regards to price point, and as a guide to put interested parties on a path to success.

This will include:

  • Explanation of concepts and terminology.
  • Tips for getting started
  • Skills needed
  • Complete build example
  • Methods of use
  • Processing files
  • Safety concerns
  • Possibilities