Building a Gimbal KAP Rig

Hello everyone, over the past few months I have been experimenting with ready-made low cost 3 axis gimbal camera mounts intended for ‘first person view’ on entry level drones.

I have converted two of these gimbals for KAP use, one for my Canon S110 and the other for my Ricoh GR.
Here are pictures of the converted rigs showing the 3 motors, the camera and the control boards.

Canon S110 Gimbal KAP Rig"
CANON S110 GIMBAL KAP RIG

Ricoh GR Gimbal KAP Rig
RICOH GR GIMBAL KAP RIG

The rigs weigh 650 grams and 700 grams respectively and are attached to the kite line via a short picavet for torsional stability.

The ready-made 3 axis gimbal camera mounts cost about $70 each and come complete with motors and control boards.

http://www.goodluckbuy.com/3-axis-brushless-gimbal-camera-mount-with-32bit-storm32-controller-for-gopro-3-4-fpv-black.html

The conversion was not a simple process but required a lot of careful ‘shop’ work building aluminium brackets and fibreglass panels etc. The most difficult part of the project was balancing the cameras in all 3 axes through repeated adjustment and several re-builds!

The electronics and on-board computers are not a ‘plug and play’ solution either so there are significant hardware and software challenges in converting them for KAP. – Not for the faint hearted!

But I believe that the effort was worth it. I now have two very stable camera platforms for action KAP, video and twilight photography.

I get a very high percentage of keepers, where low shutter speeds and low ISO shots are now possible. Gimbal stabilisation is ideal for rigs with live video feed where photographic composition is important.

I have also experimented with bigger motors, attempting to build a KAP gimbal rig for the NEX 5R but the total weight exceeded 900 grams and became a bit impractical for KAP.

I’ll post some details of my experiences after a bit more flight time but after a month or so of testing I can say that these rigs can be very useful.

Al
«13

Comments

  • edited February 2016
    Congratulations,
    It is a very good project for learning the working of gimbals in our hobby !
    You find too a very easy system to use the zoom of a compact canon camera .
    The V on the axis of the head of the servo , is it used with aerial modeler on ailerons systems ?
    José
  • Very impressive stuff! I look forward to the field results
  • edited February 2016
    Thanks for your comments José and Hans.

    I took the Canon S110 rig out for a test flight over the kite surfers at Woodman Point over the weekend and can give you some feedback on its performance.

    At full zoom (120 mm full frame) there was visible movement on video recordings although this was not noticeable while actually shooting.

    The rig was in the air for 2 hours in a 15 to 20 knot south westerly lifted by a FlowForm 16. I zoomed in and out many times during the flight and found the added flexibility quite useful to better frame each shot.

    This shot was taken at 1/640 of a second, ISO 200 with the zoom at 45 mm and is the full un-cropped 12 MP’s.

    Powering Along

    The zoom is activated by a very small servo attached to the front of the camera with double sided tape and as José points out, the servo horn can easily be cut into a ‘V’ shape to straddle the zoom lever. I have not seen this used in model aircraft but it seems feasible as José suggests.

    Canon S110 Zoom Servo

    Gimbal rigs with small motors such as this one depend on perfect balance to reduce motor torque and power requirements. I initially balanced the rig without the zoom servo motor and connector, but found that the off centre addition of 5 – 10 grams was enough to upset the delicate balance, sending me back to the workshop for another re-build!

    I will make some tuning adjustments to the Gimbal computer PID settings in the next day or two in an attempt to improve the video performance and let you know how I get on.

    Al


  • Please do. This is really REALLY cool.

    Tom
  • great and interesting matter !!
    SMAC from Italy
  • edited February 2016
    Great to see this, with great results too.

    Before I get one tossed into a container and shipped I'd like to be a bit clearer about what I'm getting into.

    I'm looking at a reasonably priced one of these on eBay (eBay item number:222026267316) with this spec:
    Name: CNC metal brushless gimbal
    Material: CNC Aluminium alloy
    Motor: Model 2208 kv80 Brushless motor
    Control Panel: BGC3.0 Brushless control panel
    Software: BGC2.2
    Strap: for Gopro camera strap
    Compatible camera: for GOPRO 3 3+ 4
    Battery plug type: JST plug / XH-2.54 plug
    Voltage: 12V/3cell Lipo battery
    Compatible model: for DJI phantom I, II and other DIY drones
    Color: silver Black
    Size: 80 x 80 x 80mm
    Package size: 125 x 125 x 105mm
    Net weight: 121g
    Pacakge weight: 162g
    I notice the BGC3.0 Brushless control panel can be bought seperately at hugely varying prices: how much quaity imapct would a fancy board have on the outcome?

    On the question of balance can you say how you did the adjustment, is it a case of 'tuning' the gimbal or does this mean more major surgery? I ask because the gimbal is designed for a 70g s**tcam and the Ricoh is 245g- does this move the centres out of the adjustment limits? How much programming (and what firmware/software) is involved in getting it to work with the new mass?

    Since the demise of the gyro servo ( I have but one left now) this looks like the best route for active stabilisation, I'm nervous about it as I can see this man in his shed getting in a right muddle getting nowhere at great expense if this is an 'expert hack' project in the hands of an amateur. Is it poissible to get this working with 'knife and fork' metal work skills?

    Thanks for sharing this- it's an inspiration!
  • Thanks for your comments Tom and Smac.

    Bill - As I mentioned a bit earlier, this project is not for the faint hearted!

    There are metalwork and computer skills that probably exceed the average hobbyist’s repertoire. Over the next few posts I’ll work through some of the hurdles I faced and this may help you decide on the way forward.

    I began with the choice of ready made Gimbal.

    This key decision was determined by the availability of good, practical on-line information. The 'Goodluckbuy gimbal' identified above has many adherents in the amateur drone world. In particular those who use RC groups to report their trials and tribulations!

    I ended up watching this long video several times to ‘understand what I was getting into’:

    https://youtube.com/watch?v=OpRAKMewn8I


    It goes into great detail about how to get the gimbal working with a GoPro or similar action camera.

    As you will see there is a fair amount of computer work required just to get started, but as you know 'this is not an out of the box solution’.

    I then gathered as much information as I could about the motors supplied with this unit and tried to work out their suitability for heavier loads. Comparing brushless gimbal motors however is no easy task, so I took a leap of faith and ordered the goodluckbuy unit anyway, figuring if nothing else, I could use it with my Xiaomi Yi.

    The balance in all three axes is important since it minimises the work required by each motor. The GoPro weighs less than 100 grams so the motors are probably sized to manage a reasonably balanced arrangement with some room for error. However, once we increase the weight and the dimensions we are adding considerable load and turning moment to the mix.

    If the load is too great and the rig is subjected to turbulence the motors it will be ‘bumped’ out of alignment with the computer commands and camera stabilisation will be lost.

    This concern prompted me to consider bigger motors but further investigation shows that the overall weight climbs quite steeply once we head down that path. So, I decided to knuckle down and work hard to achieve balance in all three axes ‘no matter what’!

    I began this exercise starting with the pitch (or tilt) axis of the camera ensuring all ancillary cables were attached and the camera was turned on to extend the lens. From the Ricoh GR rig photo above we can see that the camera is tilted (pitched) forward at about 45 degrees. We know it is balanced when it is able to maintain any angle it is placed in from zero through to 90 degrees without power to the motor.

    The position of the other camera axis is arbitrary at this point but comes into play in balancing the roll (or twist) axis where we must choose the length of the rear horizontal arm so that the camera remains horizontal without motor power.

    This brief explanation should give you some idea of the type challenges which await, and there are several more points raised in Bill's post, but I’ll stop there to 'catch my breath' and see what you think of the issues so far.

    Al

  • edited March 2016
    Thanks Al, after watching Andy RC explain I think the firmware/software it is in the realms of the possible.

    I wondered about replacing motors etc. The software procedure is well covered in the youtube link, what a joy it is to have such knowledge on tap!

    It's now clear there are specifics to adhere to make the best of the knowledge gained by FPV practitioners. The Storm32 BCG controller is a 'must have'. OlliW has done a great job with the firmware and control software, choose the wrong board and the project could be toast!

    Extension plates for mounting the bigger camera might be just about do-able with a vice and hand tools....

    ...looks like I could be scheduling some quality shed time!

  • Bill, if you jump at this project and don't have this tool in your toolbox already, look into getting a hand nibbler. They're dog slow, but very easy to use with a good deal of precision. I use a saw and file for a lot of my sheet metal work, but for inside slots and large holes I'll reach for a nibbler most of the time. (And then a file.)

    Tom
  • Impressive RC KAP rig!

    Looking forward to the Kite Aerial Video (KAV).

    The added stabilization will help big time with low light conditions (sunsets / sunrises) for both stills and video.

    The additional benefit of shooting with higher f stops to improve image quality while keeping low ISOs and mid level shutter speeds should be a sweet combination.

    Well done!

    WW
  • edited March 2016
    Jim - The stabilisation is a big help for Action KAP as well, since the shutter speed only has to cope with the moving subject but not the moving camera. The kite surfer shot above would normally require 1/2000 sec in shutter priority mode, but I’m now shooting in aperture priority and quite happy with ISO 200, 1/640th sec result.

    Bill – Yes, I reached similar conclusions on the importance of the Storm 32 board and when you consider the cost of the individual components, the goodluckbuy product is quite reasonably priced. I went one step further though, and ordered the plastic case for the control board to ensure a level of protection in the event of a crash or two.
    goodluckbuy.com/storm-32-bgc-brushless-flight-control-housing-protective-shell-protector-case.html

    A final word on gimbal balance: The ready-made unit was designed to have a small amount of adjustment in the pitch axis so there is help at hand for that final millimetre or two if you need it.

    Here are a few more issues that you may wish to consider before you schedule that ‘quality shed time’:

    Since the KAP gimbal has larger cameras and longer mounting plates it follows that the control wires will not be long enough and new wires will be required.

    The brushless gimbal motors used in the ready-made gimbal have hollow drive shafts allowing the motor, IMU and any camera control wires to pass through them on their way to the controller. This ensures that gimbal balance is not disturbed by any trailing wires and makes for a very neat job provided you can get the wires through the 5 mm hole!

    So my next headache was to find suitable wire as my Canon S110 rig has 15 conductors passing through the yaw motor’s hollow shaft! I eventually found some plastic coated multi-strand wire with an outside diameter of 0.8 mm which did the trick.

    Then there is the issue of the tiny connectors linking the MPU with the control board. These connectors are so small that I decided to leave the original wires in place on the connectors for at least the first few cm’s before soldering the extensions and placing heat shrink tubing over each wire.

    In order to minimise the rig’s weight I discarded the metal plates that sandwich the control board in the readymade rig and substituted fibreglass plates made from PCB material.

    Most of the other changes I made are fairly self-evident from the photos. One thing that may not be entirely clear though, is that despite the two rigs looking similar, they are in fact custom-made for each camera.

    Al
  • Thanks to all for these clear explanations and comments.

    I'm keeping strong interest in gimball rig. in 2013 I have already purchased a set of gimball motors and accessories. I have only done a few tests in the first months, enough to understand I will need much more time on that project.
    I'm fully convinced that the weight balance is a real trick and the software is another deal.

    Also I'm using zoom command since years on P&S Ricoh digital cameras and now on Olympus micro 4/3 camera for pole and kap shooting.
    See that page on the command for Ricoh. The page for Olympus is still at work.

    I have observed during my gimball tests that the zoom was enough to unbalance the rig. It is discussed here the importance of wires and a few 5 to 10g offset are challenging. What is the upset of the zoom on the Canon S110?
  • Christian – Thank you for your interesting question.

    The S110's zoom lens is quite light and positioned close to the centre line of the camera (the pitch fulcrum) so there is little change in turning moment applied when zooming in or out. Experiments show that the rig is able to maintain stabilised flight in all zoom positions.

    This contrasts with the change caused by adding the zoom servo motor as discussed above. This addition is considerably further from the roll fulcrum centre line leading to a greater turning moment and noticeable imbalance. (Turning Moment = Force x Distance).

    During my experiments with the Sony NEX 5R I was using larger gimbal motors but still noticed the effect of the bigger lenses further from the pitch centre line.

    Al
  • After spending only a few hours with Windows Movie Maker I can’t claim to be much of a video editor and since I had to double up as the photographer, choreographer and director as well, there are bound to be some shortcoming in this video.

    Never the less, I have put it together with the best intentions to give you an idea of what the Ricoh GR Gimbal Rig can do in 15 to 20 knot winds.

    I find it best to pause the video initially to allow my slow speed internet connection to catch up but after a 20 second pause it runs well. Your mileage may differ!

    https://flickr.com/photos/skysnaps99/24827121824/in/dateposted-public/

    All comments welcome, Al

  • Al,

    Super impressed with the stability and the ability to follow the action (kite boarders) with smooth pan and tilt on your Kite Aerial Video (KAV).

    Well done with the 3 axis gimbal camera mounts. This represents a clear breakthrough with some of the drone tech being reapplied to the KAP world!

    Agree with Christian and your comments above - getting the center of gravity nailed is important.

    After watching the video .... I may drop KAP for kite boarding ;-)

    The ability to shoot at 1/640 or 1/500 is also a big boost that opens up higher f stops and sharper IQ.

    I am currently shooting f 7.1 on my auto KAP - Sony A6000 rig for the best IQ but with out the gimbal mounts .... I need to slow down the shutter speed to keep low ISOs....

    Love the zoom option with manual control. I used zoom brackets with my Canon S100 using CHDK / SDM scripts to switch between video, stills and zoom brackets. .... miss this with my fixed prime Sony A6000.....trade offs....

    Again - well done....keep the innovation coming!

    WW
  • This video is the best I ever saw shot a from a kite line.
  • edited March 2016
    Al,
    The clip is magnificent, the gimbal really works well, the kite surfers look like they are at the top of their game too!

    Bravo!
  • Jim, Christian and Bill – thank you very much for your encouraging words – they are greatly appreciated.

    This video clip demonstrates the marvellous technology in the gimbal computer. The smooth and fluid movement of the camera in tracking the kite surfers owes a lot to the airborne computer sitting under the picavet suspension and not just the cameraman sitting on the shore.

    The gimbal software is configurable in several ways. One of these is via a slider which varies the maximum acceleration and speed which the radio control can exert on the yaw, pitch and roll motors.

    This means that an anxious cameraman cannot exceed the speed and acceleration limits set in the computer This has the effect of dampening the movement of the camera, minimising any sudden change in direction or subject overshoot.

    By careful programming of the gimbal computer these parameters can be optimised before a shoot. For this video the gimbal computer was set up for fluid camera movement and the controller was switched to joystick for pitch and yaw.

    For top down photography the gimbal would be set for a quicker response and the joystick for pitch and roll.

    Isn't technology marvellous?

    Al


  • edited March 2016
    Al,

    I have begun experiment with a dummy load and find the GLB gimbal copes well with a 35cm pendulum suspension, this is good start. The OliW software reads the X,Y,Z reactions in real-time and you can see the clipping caused by excessive travel or over limit loads.

    I have not added RC yet..one step at a time eh?

    Given your successful adaption of it for the Ricoh I'll follow the same route.

    As a pioneer can you answer some questions on adaption for the Ricoh GR (APSC model)?

    1. How big are the extension pieces for the support arms, any chance of CAD of them?

    2. Did you re- position the camera IMU and if so did this impact on performance?

    3. Can you give more detail on the extension of the control and power wires? My example has no colour coding, all wires are black- did you re-wire all the way back from the motor to the board or piece in extensions?

    4. Power supply and smoothness, the gimbal seems to draw a fair amount when it's working with big movements, what sort of battery life are you getting?

    5. How did you achieve video relay and shutter release for the Ricoh? (I don't have one yet so this may be obvious when I get one)

    Your rig is a really neat job, an example to us all!

    B

  • Al,
    You use a servo motor to move the little zoom stick on the top of your Canon Camera S110, so you use manualy the zoom Do you use CHDK to trigger the camera ?
    I check with one of my canon Camera with intervalomter mode, CHDK seems to stuck the zoom motion .
    Is there one instruction to be added to allow it ?
    José

  • There will be manual zoom only with mecanical push on the camera button .
    We can't mix the two modes on zoom and triggering .
    And allowed zoom with triggering with CHDK with the instruction remote zoom activated ?
  • Bill,

    I'm glad you have had success with the pendulum. In my experiments even the slightest torsional ‘give’ resulted in a motor driven correction followed by overshoot and further correction.

    No CAD drawings I'm afraid, but I’ll go through some preliminary info before digging into the details. However I believe that it is very important to work outwards from the camera with all accoutrements attached so that the centre of gravity is fixed once and only once. This includes the IMU in its final housing. The final choice of video and shutter control lead must also be established before fabricating the brackets. Each rig will be slightly different and since balance is so important I believe this is the only way to achieve success.

    My rig uses 3 new aluminium brackets and the 2 original motor clamping pieces. You will see that the original roll motor clamp is about 65mm long and the original pitch clamp is about 50mm long. I re-used these but swapped the roll clamp to pitch and the pitch clamp to roll. You will see that this simplifies construction.

    The camera support arm is a single piece of 1.6 mm thick x 25 mm wide aluminium bent to form an ‘L’ shape. The IMU is then fixed to the base of this wherever convenient provided the axis remains true to the original. Mine sits inside a piece of aluminium ‘U’ channel close to the bend in the centre of the arm. The camera is mounted off centre to ensure correct balance.

    Rather than provide dimensions I suggest you work with the camera, leaving a gap of 5 – 10 mm between the camera and the motor, (8 mm in my case). You will need to position the camera on the support bracket such that the centres of gravity axes align perfectly with the motor axis.

    The wiring is not as complicated as it seems. Each motor has 3 wires but they are interchangeable, ie it does not matter which wire of the 3 is connected to which pin on the control board. The IMU wires however must be connected exactly as in the original. In my case I rewired all the way to the control board. However since the IMU connectors are so tiny I added the extensions after a few cm’s of original cable whereas I rewired from the motor terminals.

    I use 2 x 500 mAh lipos weighing 50 grams each. One is for the gimbal and the other for the video transmitter and RC receiver etc. I usually leave the rig in the air for 30 to 35 minutes between battery changes.

    James Gentles makes a product to manage the Ricoh GR’s video and shutter release. Please check his site to ensure you get the correct one.

    I think I have fielded all of your questions so far but if not please ask, I’m happy to help where I can.

    Al
  • José,
    On my Canon S110 I typically use Aperture priority and zoom and shoot without problem. I use an SD card configured with USB Remote CDHK and trigger the camera remotely.

    In video mode I use a different SD card with CDHK set differently and it is a bit more complicated. In CHDK USB Remote I set the Switch Type to 2 and the Control Mode to video.

    Before launching the KAP rig I switch on the camera and press the red video button to start video. I then press my remote shutter and it stops the recording. I then launch the rig and press my remote to start and/or stop recording and use the servo zoom when required.

    I hope that helps.

    Al
  • edited March 2016
    Al, I guess but it is actually strange :
    Without any SD card (no CHDK ) inside the Canon camera, the mechanical motion with the servo on the zoom is working normally.
    With CHDK working, the view inside the screen is locked on wide angle: no action available on zoom.
    I am a CHDK user for usual settings on Canon Camera but only one expert on CHDK can find where it can't work !
  • edited March 2016
    Al,
    One of my friends Eric L told me that he try too to make the same thing with his Canon S110 , with CHDK.
    He saw that what ever the mode you choose (Auto,P,Tv,Av,M) or with CHDK
    (interval.bas, shoot.bas, hdr.bas...) the zoom stick on his camera is not active.
    But when he use KAP UAV 3.5, the zoom can be available from 0 to 100% whatever the choosen mode !
    Do you use KAP UAV 3.5 ?
    José
  • José,

    I don't use the KAP UAV script but just set the remote CHDK parameters using this reference:

    chdk.wikia.com/wiki/USB_Remote

    Al
  • edited March 2016
    Bill,

    I have taken a few extra photos to provide some more detail on the way I have constructed the rig. I hope these are useful. The batteries are not shown in these shots, but they sit on the upper deck on the velcro strips.

    Al
    Ricoh GR Gimbal KAP Rig (view 1)
    VIEW 1

    Ricoh GR Gimbal KAP Rig (view 2)
    VIEW 2

    Ricoh GR Gimbal KAP Rig (view 3)
    VIEW 3
  • edited March 2016
    Thank you Al.

    Your work is so neat!

    The photos are very useful, I see you cut down the Gentles USB [type 122?] cable and threaded it through the motors- no mean feat!

    You retained the ferrite ring- is it really nessecary?

    I see from the 323bcg tool the board has shutter control for SonyNex, Canon and Panasonic cameras, it really is a bargain.

    The work here is in containing 'rig fever' whilst assembling materials. The pendulum is a composite of thermo plastic tube, ally tube and ally bar, I have nice smooth dissipation curves on the feedback display; I suspect I could recude the cycles by lengthening it.

    I'm thrilled with the smoothness of the movement even with a dummyload.

    B
  • edited March 2016
    Al,
    It is a very good design of the rig !
    The cable are going through the axis of the gimbals, it is very good to be safely connected !
    You can earn weight by using a video transmitter using only one Li po battery element :1S !
  • edited March 2016
    This is the 35cm pendulum. It disipates movement in about 7 cycles

    image"

    The rig was released from a 30 degree apex for these 2 traces:

    image"

    B
Sign In or Register to comment.

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

In this Discussion