Improving tilt resolution on camera gimbal

hexacop

HexaCop
Hi guys,
tell me if I'm wrong, but till now every gimbal on the market I have seen uses a potentiometer directly coupled with the tilt axis. This leads to a limitation of possible resolution and as we all know the results are not very satisfying specially when the tilt direction changes.
I have rebuild my simple gimbal and made a short video, let me know your thoughts...
 
Last edited by a moderator:

Macsgrafs

Active Member
Hexacop, great video. I use a gimbal without external pots, just pulleys. I'm not using my MK FC board as the control for it, but a 2nd spektrum receiver controlled by a seperate DX7. This all goes through a FY30A, but its still not 100%. After I land I notice the camera move down a touch, but only after I land!!!! Strange stuff ;)
So what is your next route from where you are now?

Ross
 

hexacop

HexaCop
Next test will be outside, hope the whether gets better.
Definitely the precision and return rate to initial position is much better even I use a second gear to drive the potentiometer.
After that I will use the Picloc 3x and do the same test again, I'm sure that will give me even better results.
 

Bartman

Welcome to MultiRotorForums.com!!
hexacop, great video but please tell me your shop isn't always that neat!! :)
have you found that the MK camera control will always give the gimbal a little last extra movement at the end of a movement that throws it just a bit off?
if the MK FC is generating the steps, how would a geared potentiometer increase resolution? i understand that a geared potentiometer could more accurately detect the steps but does it really increase resolution (as far as control commands are concerned)?
Bart
 

DennyR

Active Member
The initial response that you see is a typical signature of the ADX 620 gyro, it is too high on the full scale range to see the initial movement. I think it is only 8 bit ADC.
Nothing can be done about that. A stand alone imu with a greater sensitivity is the only way. IT IS COMING.
 
Last edited by a moderator:

hexacop

HexaCop
Today I was able to test the modified camera gimbal with the Gopro HD Hero. The dampers underneath the camera are optimized for the weight of the GH2 and not for the Gopro but still satisfied with the stabilization.
 
Last edited by a moderator:

Bartman

Welcome to MultiRotorForums.com!!
Today I was able to test the modified camera gimbal .....................

great video hexacop. it maybe looks like it could use another couple of points added to the nick compensation value but it's very smooth. how does it compare to video shot before the geared potentiometer was installed?
 

Macsgrafs

Active Member
Superb Hexacop, love it. You certainly give it some greif at the start & it was very steady, I'm impressed. Do you mind sharing what mount, servos etc you are running?
What does that say in the snow next to that building?

Ross
 

hexacop

HexaCop
Probably the video below can be used for comparison, it was made 3 weeks ago with the same camera gimbal and the GH2.
But there I was using the Picloc 3x pro and not the FC for stabilization!

Let me play a little bit devils advocate ;-)
To my opinion the weakest point of all the current camera gimbals is the cheap analog potentiometer used to determine the position of the camera.
I is so funny to see all this very expensive and nice camera gimbals but all of them duplicate the same mistakes.
Most have realized that camera gimbals did not work very good without additional gears.
So we add nice gears and we get very a powerful drive by increasing the torque of the servo but reducing the speed.
To compensate the first mistake we get the fastest DIGITAL servo we can find on the market and then we remove the internal potentiometer.
Now we introduce the next mistake, the potentiometer is mounted on the output axis and what is the result???
A LARGE Servo with external gear, very powerful but SLOW, and DECREASED PRECISION.
By the way the potentiometer is an analog device so forget the DIGITAL precision of the 'not so expensive' servo.
You can get extremely powerful and slow servos on the market and you would not need to modify them...

Let's get back to the potentiometer, as I said the weakes part of the expensive gimbal. In the picture below you see two potentiometers, a small one and a large one. What do you think, which of the two has better precision and resolution when you mount it on the output axis of your gimbal?

View attachment 2828
So i hope you got the point, I did nothing else then replaced the tiny potentiometer by a larger one.
And I'm also convinced you can build as many controllers you want around this servo, it will not get better as long you are not fixing the cause of the problem.
 

Attachments

  • potentiometer1.jpg
    potentiometer1.jpg
    33.9 KB · Views: 352
Last edited by a moderator:

Macsgrafs

Active Member
Very true, larger diameter pot will hyave more track to travel for the same value, thus increasing resolution...how simple, but very true.
So where did you get your pots from to make your gimbal so accurate?

Keep up the great work hexacop, you are making big leaps forward.

Ross
 

Bartman

Welcome to MultiRotorForums.com!!
Hexacop,
I actually had my mount working pretty well with direct drive from the servo in the way of 1:1 pushrods and control arms but the problem was that any deadband in the servo was 100% transferred to the gimbal and was very obvious in the video product. In switching to gear reduction I was pursuing more of a mechanical filter than improved control. The high torque servos have plenty of power and speed to control the gimbal but as soon as they start to develop slop the video goes down the tubes. Also, if the pushrods weren't dead perfect in alignment and tension the servo would buzz. I had great performance, followed by OK performance, followed by better performance with the servos off than on. Also, balance of the camera on the mount was much more critical with the direct drive.
I admire the work people do to try to improve their equipment. Maybe stepper motors will provide some improvement but I think the real improvements will come when we have a camera controller that can monitor attitude while also anticipating future movement by interpreting the signals coming into the receiver and factoring them into the overall control efforts. At this point everything is reacting to the motion of the helicopter so there is always lag and it's full speed ahead for the control algorithms regardless of what is coming next. It then becomes a process of trying to tune camera mount settings to match flying styles, photo equipment, environmental conditions, etc.
Why did you switch back to the MK camera control from the Picloc 3X? I was going to try a 3X at some point in the near future.
Regards,
Bart
 


hexacop

HexaCop
Superb Hexacop, love it. You certainly give it some greif at the start & it was very steady, I'm impressed. Do you mind sharing what mount, servos etc you are running?
What does that say in the snow next to that building?

Ross

I can read in the snow "Free Tibet"
Did not realize that from the ground as I was there only for 5 minutes...

We have only one Picloc 3x and I share it with my friend, that is the reason I'm testing currently without the Picloc.
I'm sure Picloc will even improve the result with this setup.
 

Attachments

  • multifeatures.gif
    multifeatures.gif
    23.5 KB · Views: 352
Last edited by a moderator:

Macsgrafs

Active Member
I can read in the snow "Free Tibet"
Did not realize that from the ground as I was there only for 5 minutes...

We have only one Picloc 3x and I share it with my friend, that is the reason I'm testing currently without the Picloc.
I'm sure Picloc will even improve the result with this setup.

More details on multi turn potentiometer and why it does perform much better than one turn ...
View attachment 4174

He must be a good mate to share your picloc with ;)
That pot looks just the job, so where can us mere humans get them from?

Regards
Ross
 




Macsgrafs

Active Member
Thanks Hexacop (really wish I knew your real name). Have you had a chance to try with the picloc yet...now that will be a film a would love to see, can you do me a favour, find somewhere really beautiful to shoot, with wonderful scenic views & fly your abosulute smoothest possible with the picloc. Imagine you are going to be paid for your film. Fly with passion & confidence, then post the video...I cant wait ;)


Thanks for the info on the pots you are running, appreciated.

Ross
 

Bartman

Welcome to MultiRotorForums.com!!
No worries Bart, seems you are using this type of pot http://www.maplin.co.uk/miniature-potentiometers-2203 4K7 linear of course ;)
I thought hexacop was running multi turn pots..have you tried them at all Bart or are you happy with the standard pot you are currently using?

Ross

Ross,
As long as the replacement pots are 5k ohm and linear they should be fine.
I haven't tried the multi-turn pots yet. If the controller only has so many steps to offer, and the current pots can provide adequate sensing of the steps, then I don't see the reason for the added weight and complexity of a multi-turn pot. I understand that by stretching out the range, the positions can be sensed more finely but I don't know if the current pots are really inadequate enough to mreit replacement. I'm about at the limit of what I can do with my current mount design and the last of the prototype versions was assembled yesterday and today. Next comes the first of the post-prototype versions and that is all drawn and at the cut-shop to have the first batch of parts made. I actually have a "post-prototype"-prototype in pieces on the bench and I'm hoping to have it assembled this week so I can green-light the cutting of additional parts/kits. When I see what the mounts can do I'll know if my pots are doing their jobs. Eventually I'll post some pics and it'll be obvious that I've gone a different route to achieve smoothness. It's killing me to keep it under wraps but I'm just trying to decide the best way to proceed.
Bart
 

Macsgrafs

Active Member
Bart....you are such a tease ;) ;)
I am looking forward to seeing your gimbal...that almost sounds rude doesn't it ;) Have you got a price for your new one yet?
I'm currently on a 4 year old+ gimbal & feel I could be flogging a dead horse if I continue to try & get it working smoothly!

Regards
Ross
 

Top