SMALL - espritmodel.com SMALL - Telemetry SMALL - Radio
Closed Thread
Thread Tools
Old Jul 19, 2013, 01:50 AM
Registered User
Joined Jun 2013
136 Posts
Having magnetic encoders gives us the advantage of being able to mount the IMU on the centre of mass of the platform. Movements can be counteracted for before they reach the camera platform.

This is of course just in theory and results would vary depending on different forms of dampening.

I would be interested in developing a Kalman filter of this platform and perhaps even fusing it with control inputs to give it some form of predictive control.
PaulPhill is offline Find More Posts by PaulPhill
Sign up now
to remove ads between posts
Old Jul 19, 2013, 02:34 AM
Registered User
evvaldis's Avatar
Lithuania
Joined Jan 2007
179 Posts
I still can't see any logical explanations why magnetic encoder + IMU is better than just one IMU on the camera.
If you put highend encoder to the axis and use external imu for horizon, you will get the same accuracy as with IMU on the camera. Because IMU gives you horizon!!
IMU on the camera is good because it gives ACTUAL position of the camera, not position of the copter or something like this. Also accelerometer gets less movements, therefore horizon becomes more accurate.
One more problem with magnetic sensor is that it can't see flexibility of the gimbal, it can't see actual angle between IMU and camera.
evvaldis is offline Find More Posts by evvaldis
Old Jul 19, 2013, 02:41 AM
bob the builder
france, near GENEVA
Joined May 2009
14 Posts
Quote:
Originally Posted by evvaldis View Post
I still can't see any logical explanations why magnetic encoder + IMU is better than just one IMU on the camera.
If you put highend encoder to the axis and use external imu for horizon, you will get the same accuracy as with IMU on the camera. Because IMU gives you horizon!!
IMU on the camera is good because it gives ACTUAL position of the camera, not position of the copter or something like this. Also accelerometer gets less movements, therefore horizon becomes more accurate.
One more problem with magnetic sensor is that it can't see flexibility of the gimbal, it can't see actual angle between IMU and camera.
.

It permit to fix the magnetic field of the motor, in stead of getting "forward and then backward" loop as you get with the imu, even if its pretty fast.


continue Brainstorming, its becoming pretty interesting now
andreas-aero is offline Find More Posts by andreas-aero
Old Jul 19, 2013, 02:46 AM
Registered User
evvaldis's Avatar
Lithuania
Joined Jan 2007
179 Posts
Quote:
Originally Posted by PaulPhill View Post
Having magnetic encoders gives us the advantage of being able to mount the IMU on the centre of mass of the platform. Movements can be counteracted for before they reach the camera platform.
I have tested different IMU placement and there was no difference at all (~+-20cm to all sides). One important rule is that IMU have to be placed tightly.
As I have said, IMU gets almost none movements, therefore it is easier to calculate horizon, also accelerometer gets less distortion while sitting on the camera.
P.S. I see that even movi uses IMU on the bottom of the camera, and it works without problems. I think this means something
evvaldis is offline Find More Posts by evvaldis
Old Jul 19, 2013, 02:48 AM
Registered User
Poland
Joined Jul 2011
126 Posts
andreas-aero, this "pretty fast" is currently 500 Hz I believe. So, It's not "much forward and then backward", because every 2ms motors are corrected. So it's just a matter of movement during these 2ms - I believe this is a very very small movement your eye won't notice...

Perhaps we'll see it better in the future if we have some gfx charts on PC.
acc007 is offline Find More Posts by acc007
Old Jul 19, 2013, 02:49 AM
Registered User
evvaldis's Avatar
Lithuania
Joined Jan 2007
179 Posts
Quote:
Originally Posted by andreas-aero View Post
.

It permit to fix the magnetic field of the motor, in stead of getting "forward and then backward" loop as you get with the imu, even if its pretty fast.


continue Brainstorming, its becoming pretty interesting now
With magnetic sensor you will get loop between motor and sensor, so it is almost the same as imu-motor loop.
evvaldis is offline Find More Posts by evvaldis
Old Jul 19, 2013, 03:43 AM
bob the builder
france, near GENEVA
Joined May 2009
14 Posts
Quote:
Originally Posted by acc007 View Post
andreas-aero, this "pretty fast" is currently 500 Hz I believe. So, It's not "much forward and then backward", because every 2ms motors are corrected. So it's just a matter of movement during these 2ms - I believe this is a very very small movement your eye won't notice...

Perhaps we'll see it better in the future if we have some gfx charts on PC.
Yup, "pretty fast" could be replaced by "fast enouth" :P
But i'm not sure you can read I2C bus at 500Hz ?


Anyway, a friend of mine who work in a drone making company is working on BLG concept, and he tried magnetic sansors WITHOUT ANY IMU and IMU alone under the camera.

He found it easier to tune with only magnetic sensors thant with a MPU 6090 ( not sure, the one with integreted compas)
Also, results are comparable from is point of view.

Any way, I think that mag sensor could be great for the autopan fonction, but probably not nescessary for the other axis.


Don't complicate it too much, cuz the frame will be harder to build if we have to integrate a magnetic sensor on each axis.
Also, many more cables, not pratical at all on little rigs ( don't care on dslr ones...)
andreas-aero is offline Find More Posts by andreas-aero
Old Jul 19, 2013, 05:50 AM
Registered User
Joined Apr 2013
91 Posts
Have you guys seen this? I believe he is using Alexmos boards or a variation of some sort.

http://www.kickstarter.com/projects/...liser?ref=live
natious is offline Find More Posts by natious
Old Jul 19, 2013, 06:01 AM
Registered User
evvaldis's Avatar
Lithuania
Joined Jan 2007
179 Posts
Quote:
Originally Posted by andreas-aero View Post
Yup, "pretty fast" could be replaced by "fast enouth" :P
But i'm not sure you can read I2C bus at 500Hz ?
Now evvgc works on 500hz, I have tested 1000Hz and it works even better.
Maybe will post 1kHz code today or tomorrow.
evvaldis is offline Find More Posts by evvaldis
Old Jul 19, 2013, 07:18 AM
Ready to fly MRs
Sid3ways's Avatar
Swanton, OH
Joined Jul 2007
5,066 Posts
Quote:
Originally Posted by fr3style View Post
Hi Guys! Just bought this and one Contorller. any one know if this will work on a Canon T2i with light lens also on a Canon 7D with light lens

http://www.goodluckbuy.com/3-axis-ds...otography.html
Please keep us updated on this setup.
Sid3ways is offline Find More Posts by Sid3ways
Old Jul 19, 2013, 07:49 AM
Registered User
Joined Jun 2013
136 Posts
Is there a way to setup the eclipse environment without the use of cygwin?
PaulPhill is offline Find More Posts by PaulPhill
Old Jul 19, 2013, 08:26 AM
Registered User
United States, GA, Atlanta
Joined Dec 2010
2,061 Posts
Quote:
Originally Posted by PaulPhill View Post
Is there a way to setup the eclipse environment without the use of cygwin?
the only thing that cygwin is used for is *cs-make and cs-rm*.

If you can get another make on your system, you can use it, go to project->properties->c/c++ build and tell it to not use the built in make (cs-make). You may have to give a path to the other make if you do this.

Also, you can go copy your del.exe to cs-rm.exe that will fix the delete feature for clean, etc.

The other options are to use mingw or see if you can find a gcc tools for windows (which is pretty much what cygwin is).

Alan
aadamson is offline Find More Posts by aadamson
Last edited by aadamson; Jul 19, 2013 at 10:58 AM.
Old Jul 19, 2013, 09:24 AM
Registered User
Joined May 2013
45 Posts
Quote:
Originally Posted by evvaldis View Post
Ho ho.
Since people are working on eclipse version, I have modified few things in keil version.
Not much changed, I took 1.2 board from AirBot and tried ultrasonic operation. Changed PWM timers prescaler from 5 to 3 (now PWM is 18kHz), and it works pretty nice and SOUNDLESS.
Also changed "return from collision speed", and because of that PID configuring should be easier than before.

Inside ZIP you should find sources+HEX for reflashing.
Just Tested out your updated FW, it works great! Zero Noise!

My PiD settings need to be changed a little now, But overall it seems like the yaw and roll have a little more strength.
Bergeninc is offline Find More Posts by Bergeninc
Old Jul 19, 2013, 10:01 AM
Registered User
Joined Jul 2013
24 Posts
A very interesting development is constantly ahead ... :-)

EvvGC - Testing infinite "Pitch", regardless of the "Roll" ... (2 min 23 sec)
hcmpromotion is offline Find More Posts by hcmpromotion
Old Jul 19, 2013, 10:20 AM
Registered User
evvaldis's Avatar
Lithuania
Joined Jan 2007
179 Posts
Quote:
Originally Posted by Bergeninc View Post
Just Tested out your updated FW, it works great! Zero Noise!

My PiD settings need to be changed a little now, But overall it seems like the yaw and roll have a little more strength.

Thanks a lot for the feedback

This weekend I hope to test more 1000Hz loop speed, also to test AutoPan+AutoTilt feature, also will improve GUI.
Will post sources+hex here
evvaldis is offline Find More Posts by evvaldis
Closed Thread


Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
Discussion Open Source Brushless Gimbal edge0xc Multirotor Electronics 3046 Sep 01, 2014 06:40 PM
Idea Open Source Gopro Brushless Gimbal icais Multirotor Talk 16 Mar 28, 2014 12:07 PM
Discussion SkyLight SBG - GoPro 2 Version Released (+ GP3 Beta + 3 axis Alpha) !! ed_jza80 Multirotor Talk 494 Jan 03, 2014 10:09 AM
Sold open source brushless gimbal controller jy0933 Aircraft - Electric - Multirotor (FS/W) 57 Mar 26, 2013 07:02 PM
Sold [closed] open source brushless gimbal controller jy0933 Aircraft - Electric - Multirotor (FS/W) 16 Feb 26, 2013 09:19 AM