Continuous rotational servo

[Oliver]

Does anyone know of the existence of a "continuous rotational servo with position control" ? Basically it is a servo where the arm can rotate 360 degrees continuously yet the position where you want the arm to be can be controlled.

I would think such a servo would be used in RC sailboats.

The use for it is to position a surface on a mobile object to a fixed point no matter what the position is. For example a satellite dish on a truck that needs to be fixed in position to a satellite while the truck is in motion.

How the servo is controlled is not important, however that it can rotate 360 degrees continuously is, that the position can be controllled, and that it is the same size as an RC servo is too.

If anyone has any knowledge, your help would be greatly appreciated.

Thanks, Oliver

[jeffrey g]

www.servocity.com

they have many options and mention witch regular servos can be modified for 360 rotations.

This modification is done by removing the pot and replacing it by a resistance of "x" + removing the limiting tab in the gearing..

You'll find many posts on aérial photography with this info

here is n other link : http://www.uoguelph.ca/~antoon/gadgets/servos/servomod.html

Jeffrey

[Oliver]

Thank you Jeffrey.

[Andrew Fernie]

Those sites are interesting - I will keep them both bookmarked.

However, I don't think that either provides what you are looking for:
1. ServoCity has multi-turn (but not continuous rotation) servos with feedback. They use a multi-turn pot for this.
2. ServoCity have continuous rotation servos, but they don't provide position feedback.
3. The modifications on the U of Guelph site provide continuous operation, but no feedback (they remove the pot and replace it with two fixed resistors that will always set the position feedback in the servo loop to zero.) This is effectively the same as item 2 above.

The trick to getting position feedback is to find an encoder that supports continuous operation without a discontinuity (i.e. a jump from full positive to full negative at some point, and no dead area.) This used to be done with AC encoders that provided two outputs representing the sine and cosine of the angle, nowadays you are more likely to find an optical encoders.

One other cheap approach that I can think of would be to use two pots for the feedback. Modify both to allow continuous operation, but recognize that they will both have dead areas where you can't get a good signal. Mechanically arrange them such that the dead areas don't overlap. Then you just need to read two inputs, and work out an algorithm to select the one with good data, and probably have some form of average calculation when they are both valid to smooth out any bumps in the data.

What do you want to do with the servo and the position feedback? Maybe there is a simpler solution if we know what limitations you can put up with.

Andrew

[Michael V]

Those sites are interesting - I will keep them both bookmarked.

However, I don't think that either provides what you are looking for:
1. ServoCity has multi-turn (but not continuous rotation) servos with feedback. They use a multi-turn pot for this.
2. ServoCity have continuous rotation servos, but they don't provide position feedback.
3. The modifications on the U of Guelph site provide continuous operation, but no feedback (they remove the pot and replace it with two fixed resistors that will always set the position feedback in the servo loop to zero.) This is effectively the same as item 2 above.

The trick to getting position feedback is to find an encoder that supports continuous operation without a discontinuity (i.e. a jump from full positive to full negative at some point, and no dead area.) This used to be done with AC encoders that provided two outputs representing the sine and cosine of the angle, nowadays you are more likely to find an optical encoders.

One other cheap approach that I can think of would be to use two pots for the feedback. Modify both to allow continuous operation, but recognize that they will both have dead areas where you can't get a good signal. Mechanically arrange them such that the dead areas don't overlap. Then you just need to read two inputs, and work out an algorithm to select the one with good data, and probably have some form of average calculation when they are both valid to smooth out any bumps in the data.

What do you want to do with the servo and the position feedback? Maybe there is a simpler solution if we know what limitations you can put up with.

Andrew

:confused::confused::confused::confused::confused:
Speak ah the english :)
Andrew if I didn't already know you were an engineer, I would know now.
love it!

[Oliver]

Hello Andrew,

That helps a lot. Yes the servos mentioned do not quite do the trick. The purpose for the servos are actually for a robot that draws any shape circles around a fixed position. An arm moves around the fixed position with a pin at a 90 degree angle. The movement of the arm as like that of a drafting table meaning the pin has a top and bottom that are always in the same compass reading. However, the idea is to have a surface on that pin that will rotate and stay facing the fixed center position. The movement of the arm is made by tiny controlled up and down, and side to side movements, and it is actually creating a polygon rather than a circle. Also, the "circle" is not necessary a circle but could be an ellipse, square, etc. Therefore the position of the servo would have to be precisely controlled.

Thanks, Oliver

[Andrew Fernie]

Hello Andrew,

That helps a lot. Yes the servos mentioned do not quite do the trick. The purpose for the servos are actually for a robot that draws any shape circles around a fixed position. An arm moves around the fixed position with a pin at a 90 degree angle. The movement of the arm as like that of a drafting table meaning the pin has a top and bottom that are always in the same compass reading. However, the idea is to have a surface on that pin that will rotate and stay facing the fixed center position. The movement of the arm is made by tiny controlled up and down, and side to side movements, and it is actually creating a polygon rather than a circle. Also, the "circle" is not necessary a circle but could be an ellipse, square, etc. Therefore the position of the servo would have to be precisely controlled.

Thanks, Oliver
Oliver,

I can't visualize the mechanism - do you have a sketch?

Andrew

[Jeremy Cartlidge]

Does it look something like this?

2991

[Ronald Longtin]

I have very limited skills in electronics, however a friend who specializes in control systems suggest a multi turn pot as a solution.

Apparently they can give you accurate feedback over 10 complete turns and are available in small sizes. He mentioned Bourns as a possible source.

[Oliver]

Sorry , I ws away. LOL, no it doesn't quite look like the image. Is that a cement mixer ?
Andrew, to help visualize the set up, imagine a clock with a pin sticking up at the end of the arm. At the end of the pin is a weight, like a pendulum and on that pendulum is a small flat surface that you always want facing the center of the circle. This is all happening at low speed.
In regards to the continuous rotational servo, the problem has been solved. Instead of using a servo to keep a flat surface aimed at the center, The flat surface is mounted on a quickly rotating motor and the moments where the surface is facing the center are sufficient.
Thanks for your help.

[beto9]

Are you sure you are not bent on destroying the Universe?