Selecting Servos

[Matt Denton]

I have had many people asking me for advise on servos over the years, hence this area of the forum.

I will add my views on certain makes and models of servos that have proved them selves in the past within Animatronics and Robotics, along with name and shame those which have failed.

A few notes on selecting servos:

1) Scale
Pick the largest servo you can use within your application. Eg. You may find a standard size servo with the same torque rating as a 1/4 scale servo, however, the gearbox in the 1/4 scale servo will usually be much stronger than the smaller standard size servo gearbox.

2) Make
Generally speaking, the more expensive the servo is, the better quality it will be, however, you do tend to pay for the name with certain makes of servo.

3) Analogue/Digital
There are pros and cons for both these servo types. Digital servos usually give better resolution than analogue servos, and usually have higher torque ratings than their analogue counterparts. However, this increased torque usually comes from driving the motor harder which uses more current and generates more heat. Digital servos are often noisier than analogue's which is not a problem for most robot projects, but within animatronics can cause problems.

4) Brush-less Servo Motors
There are a new range of servos coming out which we will probably see more of over the next few years. These servos utilise electronically commutated brush-less motors. As well as the other benefits of digital servos, the coils on a brush-less motor are on the outside, which gives several benefits such as low start up inertia and good heat dissipation.

5) Linearity
Some cheap servos use cheap feedback potentiometers which can give very poor linearity over the servo range travel. This means that servo centers will vary more than usual, but more importantly the distance the arm travels in either direction may be different. This is not so much of a problem within general R/C use, but within robotics can soon become a headache. For example, if the servo neutral centers vary (which they will) adding an offset to calibrate each servo is not too difficult, however, if the servo travels a different amount in each direction, you will need to calibrate the center point and end points of the servo and use interpolation do generate a linear motion... yawn! Good servo linearity is one of the most important servo features.

6) Torque
There are many standard size servos which now offer huge torque ratings of 15 kg/cm of torque or higher, however, it's always worth looking at the speed of the servo. Many of these high torque servos have been designed with R/C models in mind, and as such they need to be fast as well as strong. This sounds great, but the way that fast servos achieve high torque is by using much more current, this is fine for a few servos, but if your running an 18DOF hexapod, you'll be eating batteries galore! In my opinion, depending on the application Its better to pick slower servos for robotics use. This has a couple of benefits, such as lower current consumption, cheaper price, less jitter and often stronger gearbox's.

[Tony]

good topic, i have headache on the torque and price!

[paulp]

There are a few other points of interest that I have come across.

Hitec and Kondo both produce humanoid robots, Robonova and the KHR. There are several others such as Manoi now coming onto the scene. These were originally seen as research projects and priced accordingly.

The servos they employ have several features that may be of interest.

They usually run off higher voltages. As high as 10.8V in the Kondo kits. The electronics in the servos are upgraded to handle the higher voltages and current though not always as well as they could be, As I have found to my expense. The F.E.T's blow regularly and are not always easy to replace. Being surface mounted you usually only get one shot at fixing them. The Motors tend to polarise and bind quite badly after a while.

The range is usually 180 degrees rather than the normal 90 degrees. This limits the number of controllers that can run them as they only generate a pulse range between 1000uS and 2000uS. To run 180 degrees requires 500uS to 2500uS

Some such as the KRS4024 from Kondo has a 270 degree range. This has problems near the extremes of the range due to reaching the limits of the potentiometers.

Some have Learning Modes. When the pulse is set to 0uS they go into a holding mode rather than sleep and start to send a pulse back to the controller to indicate the current position of the servo. The controller can then tell the exact current position of the servo. This does require a servo specific controller from the manufacturer.

[robotguy]

Some have Learning Modes. When the pulse is set to 0uS they go into a holding mode rather than sleep and start to send a pulse back to the controller to indicate the current position of the servo. The controller can then tell the exact current position of the servo. This does require a servo specific controller from the manufacturer.

Why would it require a special controller to get this information? What is special about the data being sent back to the controller? Is it PWM? What is it?

Inquiring minds want to know!

8-Dale

[paulp]

I believe it is PWM. It would make sense as there arent any additional data lines. The controller needs to have a learning command that forces it to read from the output pins which on most multiplexed outputs such as Matts design and the Lynxmotion SSC32 isnt possible.

I forget to mention there are also ther Bioloid servos which are addressable and connect together in a daisy chain affair. Dont know much about them but they are supposedly quite good.

[Matt Denton]

Some have Learning Modes. When the pulse is set to 0uS they go into a holding mode rather than sleep and start to send a pulse back to the controller to indicate the current position of the servo. The controller can then tell the exact current position of the servo. This does require a servo specific controller from the manufacturer.

Why would it require a special controller to get this information? What is special about the data being sent back to the controller? Is it PWM? What is it?

Inquiring minds want to know!

8-Dale

If it's anything like the hi-tec feedback system, then yes the returned signal is PWM. I can tell you its pretty useless for any thing useful! Much better with a system such as bioloids half duplex 1 mbit rs485 interface and do away with PWM altogether.

[paulp]

Didnt realise they were RS485. Now that I understand......learn something new everyday

[Matt Denton]

Didnt realise they were RS485. Now that I understand......learn something new everyday

Actually I think it's only their larger servos that have the 485 interface, the smaller ones I believe are 1 wire TTL.

[paulp]

You may be right, I seem to remember some complaints about the bioloid servos and the daisy chain nature.

Something to do with having to provide more ports due to wiring complexities. Imagine trying to wire 18 servos on a Hexapod in a serial format.

[winxamitosis]

hi there, i joined recently after being gobsmacked by the hexapod creations.
as you'll be able to tell in a few seconds, i'm very new to all of this, so i have a few questions.

what on earth are...

F.E.T.'s
P.W.M's
1 wire TTL's
pulse ranges between 1000uS and 2000uS

see? told you i was new. i also have a feeling that leaping straight in and getting a phoenix kit (or waiting for the hexapod
kit matt's mentioned he looking into (and yes i have voted on some of the options)) feels like that i could very well be out of my league on most topics here (and construction)

but as servo's are the muscles that make all this move, i thought i'd have a read.
some good stuff here (i think) but the acronym's are losing me slightly.

please help a noob

cheers
Winx
p.s i absolutely love the hexapod stuff, and found the terrain reading version fascinating. keep up the good work!
(edited for spelling)