Robotic arm Servo questions

[superfly501]

Hello Society of Robots,

I want to make a simple lever arm.  Is attaching 2 servos onto a pivot joint to increase the torque a good idea, or would they be hard to align, making them oppose and stall each other out?

Also for the rotational motion, how would a modded continuous servo, say hitec-422, be for positional control with an encoder? I'm not expecting super accuracy or anything. Can continuous servos hold a position at all? Too bad hitec 360 servos are so pricey. Thanks for any help.
 

[Soeren]

Hi,

I want to make a simple lever arm.  Is attaching 2 servos onto a pivot joint to increase the torque a good idea, or would they be hard to align, making them oppose and stall each other out?

A stronger servo would be a better (and probably cheaper) solution.

Also for the rotational motion, how would a modded continuous servo, say hitec-422, be for positional control with an encoder? I'm not expecting super accuracy or anything. Can continuous servos hold a position at all? Too bad hitec 360 servos are so pricey. Thanks for any help.

If you describe exactly what you wanna make (using plenty of metric/imperial units), it will be easier to decide what will be the best solution.
Terms like '"super accuracy" doesn't really help, as we all have different frames of reference, which further changes with different projects (might be 0.001°, 1°, 10° or whatever, depending on purpose and scale).

[superfly501]

Sorry about my vagueness. Well I was planning to put the arm joint on a disk about 10 cm in diameter. The arm being about 40cm fully extended. Then having a servo underneath the disk so the arm could rotate 360. Right now I just want to get basic motion going, but eventually I want to have sensors attached to the arm so it could pan around and detect flames, color, or distance and be able to move back to where it detected the thing I was looking for.  The accuracy I was aiming for was +/- 5 degrees for rotation, but up to +/10 isn't too bad.

Ideally, I wanted a 360 servo so I could know my position, but to save money I was planning on using a continuous HS-422 and using the encoder method posted on the site http://www.societyofrobots.com/sensors_encoder.shtml. But I'm not sure how accurate that is, or how good continuous servos are at stopping. Can they hold position at all once modded for continuous, or are they basically geared motors?

[Soeren]

Hi,

I was planning to put the arm joint on a disk about 10 cm in diameter. The arm being about 40cm fully extended. Then having a servo underneath the disk so the arm could rotate 360.

Do you want to rotate the disk (with the arm mounted), or just the arm?

Right now I just want to get basic motion going, but eventually I want to have sensors attached to the arm so it could pan around and detect flames, color, or distance and be able to move back to where it detected the thing I was looking for.

OK.
Either you use a "simple" quadrature encoder and count the ticks from the max. reading, or you use an absolute encoder and just store the max. position read from the disc (i.e. keep the angular reading taken when the signal maxed out).

For some of what you want to detect, an analog sub-system might be faster.

The accuracy I was aiming for was +/- 5 degrees for rotation, but up to +/10 isn't too bad.

5° is 72 positions. With an absolute encoder, you could make it either 5 bits wide (11.3° resolution), 6 bits wide (5.6°) or 7 bits wide (2.8°), with a relative encoder, youhave to decide whether you want a quadrature track or you want to mount the readers displaced for the quadrature signal.

When you decide what kind you want and the exact measures, I can make you a vectored .pdf of it, ready for printing, if that will be of any help.

If you just wanna go towards the position, you don't need much precision, as the angle of detection will widen as you get nearer.

Ideally, I wanted a 360 servo so I could know my position, but to save money I was planning on using a continuous HS-422 and using the encoder method posted on the site http://www.societyofrobots.com/sensors_encoder.shtml. But I'm not sure how accurate that is, or how good continuous servos are at stopping. Can they hold position at all once modded for continuous, or are they basically geared motors?

They're nothing but geared motors when the feedback is removed, so the holding power is just the gear ratio times what it takes to overcome the (very small) motors inertia - not much.
A better solution is to use a DC-motor and a worm gear, as it cannot be rotated backwards (from the load to the motor), so it locks when the motor stops.

If you just want to record the position of max. signal, it shouldn't matter how fast it stops, since you can go slow when returning (as you now know where it is).

I don't see any need for large holding power, as the rotational plane is level with the ground (I assume).

[superfly501]

Thanks for taking the time to write all that.

The disk will have the arm mounted onto it. I'll probably stick to just having simple ticks on the encoder, and count them. I think I'll be able to make that myself.

If you just wanna go towards the position, you don't need much precision, as the angle of detection will widen as you get nearer.

I don't quite get what you mean about the angle detection, could you explain that?

Also for the encoder does it actually have to be transparent paper? I'd assume with plain paper there'd be a big enough difference between white and black for the IR detector to notice the difference.

[Soeren]

Hi,

I don't quite get what you mean about the angle detection, could you explain that?

Place yourself a 100 yards from a phone pole or similar and it will block only a small fraction of your field of view. Now walk all the way to right before your nose hits the pole and note how it will "widen" in your field of view, perhaps blocking it completely at the end of your experiment.

Also for the encoder does it actually have to be transparent paper? I'd assume with plain paper there'd be a big enough difference between white and black for the IR detector to notice the difference.

There's two ways to read an encoder disc. Reflecting the IR light off one side, or shining it through the disc.
Transparent "paper" (acetate film) will let a large amount of light through non "inked" parts, but for reflective use, the transparent parts can look black (depends on reading angle, refractive index etc.).

From what you write, I assume you want to read it with a reflex reader and thus the best option is plain white paper - LASER printed, but if you have an inkJet and you need more contrast, a photo copy is a good way to increase the contrast.


That said, optical reading is all about contrast (difference between light and dark - the higher the contrast, the more sloppy you can be with the hardware (not that it should be taken as an excuse for sloppy work )

For see-through encoders I often use PCB material (mostly fibre glass carrier board, but the cheap pertinax works as well, although it's rather brittle), by etching the pattern, just like when making a regular PCB. The good thing about PCB material is that absolutely no light will go through the copper foil and you get very crisp edges.

For reflex reading, the ultimate contrast optimum will be with silk finish photo paper, as this is very reflective, but without glare. Print on a LASER toner, as it's way "flatter" than inkJet ink, but with carefully chosen LED- and transistor current, you can decode quite low contrast ratios if you don't have much ambient influence.