Camera leveller

[camerart]

Hi,

The POT has been removed from a standard Servo, and with a weight on it's spindle corrects level ok, but it is too big. I checked the voltages on the +- and wiper of the POT and it isn't straight forward 0v to 5v with wiper, so I thought it might be too complicated.

The Pendulum does work as you say and also does oscillate too much.

I'm sure we would get there in the end, but I sensed your doubts, so this morning I made a PIC chip circuit, with A/D inputs and PWM out.

I have tried a quick damping test and have got it to oscillate 1 or 2 degrees, which is fine for me, but I'll play some more with the program, and I'm sure I can improve it.

Does this make more sense?

Cheers, Camerart.

 

[NickS]

(1) The POT may be logarithmic scale

(2) I think you will be much happier with the flexibility of using a pic. What are you using for a level sensor now(accelerometer, pendulum)? The pendulum is inherently problematic with only 2degrees of range so I hope you are using an accelerometer.

(3) The way to get rid of the fine oscillation right at the level point is to add some hysteresis to the control. Its kind of like a dead zone right in the middle where you don't try to correct. Then if the level breaches your set dead zone threshold it will then return to normal operation.

 

[camerart]

Hi Nick,

I don't understand Hysteresis, but I get what you mean.

I'll stick to PICs then. There are 2 X axis, so I can switch between between 'READs' then allow each Pendulum to stop.

I can get it easily to get down to 'say' 2 degrees (within the Swing), then I thought I would simply read how far it was away from centre, and only move that distance once.

My time is up, So Cheers, Camerart.

 

[camerart]

The Pendulums/Servos are working fine, although the PIC chip program needs a little tweeking, Thanks.

C.

 

[NickS]

Glad to hear its working.

 

[camerart]

Thanks for your help.

 

[camerart]

Hi,
I've just re-read this thread
I hope that over the years, I've got a bit better at asking and explaining questions.
I would like to thank NickS again for his endless patience.
I hope to explain a little how it works: I dismantled two small clock that had balance wheels in them. I cut out all but one of the spokes of the wheels, leaving a pendulum, suspended on jewels, so they were pretty free.
I place these on two axis forward/backward and left down/left up. Each spoke/pendulum passed between the gap in the opto but long ways, i,e, up and down the slot, instead of the usual in line with the slot, which gives a slower electrical change. Especially as the gap was cut near the suspension centre. This worked ok, and I took it to show the camera being levelled, to my 90 year old neighbour. He 'was' a clock maker. He said why didn't I simply suspend the camera, on two axis so it levelled itself without all the electronics. I tried that, and of course it worked, and that's what I used in the end.
All the best, Camerart.

 

[Minder]

It sound to me as though you are describing a slot opto?
For future ref. I would think you would need a PID based servo system such as demonstrated in the PIC AN964 Pendulum App.
I put one together a few years ago.
There is the code in the app in assy and C.
M.,

 

[camerart]

Hi,
It's taken me ages to get around to explaining this, but here goes.
Here are 4x images of a small clock mechanism, with all but one spokes removed from the balance wheel and the last one used as a pendulum. (Anything similar could be used)
The spoke moves down the slot of a slot photoelectric sensor, and slowly cuts off the light instead of the abrupt way it usually does.
I hope you can see
Camerart

 

[Bluejets]

For future reference, take a peek at the open source headtracker project using Arduino. It may give you a few alternative ideas and have some advantages. Perhaps look at the associated calibration which is also available freely.

 

[camerart]

For future reference, take a peek at the open source headtracker project using Arduino. It may give you a few alternative ideas and have some advantages. Perhaps look at the associated calibration which is also available freely.

Hi B,
If I was going to do it again, I would update my methods, but this was only to illustrate how I did it using slot photoelectric sensors.
In the thread I explained that a 90 year old neighbour, asked why I didn't use gymbols as in a boat, which I did, so no electronics.
Thanks.
C

 

[camerart]

Hi,
I've just re-read this thread, and can see I wasn't good at explaining.
I hope this sketch clarifies things a bit more. It works quite well.
The pendulums are out of clocks. They are balance wheels, with all but one spoke (Pendulum) cut out.
C

 

[Bluejets]

There was a hack many moons ago for rc servos to change from the standard feedback pot to optical encoder.
Why I'm not sure, I just vaguely remember it. Perhaps a tighter deadband.?
See what I can find but again, why bother for a simple device as described here.
The servo itself is hardly going to do the job.
Perhaps an external feedback arrangement but ten that introduces other requirements.

Edit...here is a magnet encoder hack on a fairly standard large scale servo.
https://www.instructables.com/id/Hack-your-servo-v200-Add-10-bit-incremental-a/

 

[camerart]

There was a hack many moons ago for rc servos to change from the standard feedback pot to optical encoder.
Why I'm not sure, I just vaguely remember it. Perhaps a tighter deadband.?
See what I can find but again, why bother for a simple device as described here.
The servo itself is hardly going to do the job.
Perhaps an external feedback arrangement but ten that introduces other requirements.

Edit...here is a magnet encoder hack on a fairly standard large scale servo.
https://www.instructables.com/id/Hack-your-servo-v200-Add-10-bit-incremental-a/

Hi B,
This is a very old thread, and I was only clarifiying, the pendulum.

I used 2x opto outputs, into a PIC, which then drove 2x servos, 1x on the horizontal axis, and 1x on the forward backward axis.

I used this method for a while, then a 90 year old, asked me why I don't simply use gravity on the camera, instead of the pendulums. I re-designed the camera gimbals, and this was much simpler than my original complicated effort.

I used them on buggies, that followed a wire, along a path, where the buggy stopped, took a photo, then waited for a set time, before repeating, so that a video could be made from the images. Good fun.
C