Servo performance (tester?)

If you can describe which parameters of a servo's performance are critical to the design, we can suggest ways to test them.

Some are very simple:

- If the weight of the servo is design-critical, a scale can be used to measure them.

- If the maximum sustained torque is design-critical, it can be looked up in the manufacturer datasheet.

Are you takning about hobby-airplane servos?

I don't understand this question.

I'm not sure exactly what characteristics of the servo you want to test but Dale Wheat of the Dallas Personal Robotics Group makes a little atmel-based diagnostic/calibration board for R/C servos that might be of interest.

-Steve

This is a problem of terminology. The classic "servo tester" merely tests the servo, as in "does this thing work?" Most testers will sweep the servo back and forth some pre-defined amount, but that's about it. It sounds like you want something to test the operational parameters of the motor, such as speed, holding torque, bandwidth, and so on. I don't know of anything like this outside of the manufactuter's inhouse test rig.

Most folks take the specs of the servo at face value. Perhaps you can look at the specs, derate them 10% or 20%, and hope for the best?

-- Gordon Author: Constructing Robot Bases, Robot Builder's Sourcebook, Robot Builder's Bonanza

Bryn wrote:

Thankyou gentlemen for your responses....to clarify

The one of the top end motors I am looking at is the Hi-tec HSR 5995TG otherwise known as the X-SERVO

Stall torque 30.0kg/cm Transit speed 0.12sec/60°@7.4V Size 40(L)×20(W)×37(H)mm Weight 62g

Servos performance; This is given by the manufacturer as stall torque and transit speed under no load (see above). So when selecting a suitable servo these two performance parameters are all I have to go by.

Now I may select a motor with a transit speed of 0.1sec/60 degrees and a stall torque of 10kg/cm. If I apply a load to this motor of say 50% its stall torque, 5Kg at 1 cm from centre of rotation, what would the transit speed be? By how much has the transit speed degraded? what is the relationship between load applied and transit speed degregation?

It would be a fairly simple to measure this once I have bought the servos but obviously I dont really won't to buy a load of servos test them and then make my selection (I only need eight servos for this project). I have contacted most of the major servo manufacturers however they seem very reluctant to give anything away.

Also its worth adding the performance of these motors under load will have a direct influence on the design of the robots appendages (wings). I am trying to maximise the available torque (to give a good range of results) whilst maintaining the highest possible transit speed (to minimise wing size).

My main concern is transit speed, so I may be better buying a slow rated motor but with a higher stall torque, than a motor with a fast transit speed with medium amounts of torque. The thinking being that the slower rated motor will actually be quicker than the faster rated motor under load.

Any thoughts much appreciated...

I think that if you make a simple free-body diagram and look at the inertia of the wing/air-mass reflected back to the servo, that you will find out very quickly that the limit to the speed you can make the transition is the maximum torque available at the servo.

I don't think you will be able to move any substantially-sized flapping wing in 100 milliSeconds using a hobbby-airplane servo.

I wonder if you are starting from the wrong end. The power output of one of the big HiTecs is around 6 watts - this isn't very much, and I'm guessing that you may well need more. Typical servos lose a lot of power in their gearbox, and you will lose more power in starting and stopping. Typical servos are also not designed to run at maximum power for long periods either.

I'd look for a way to run motors continuously, and consider devices like the electronically commutated 'cobalt' motors, some of which run to many hundreds of watts. A custom 'gearbox' (toothed belt drive), driving a crank if you want a 'flapping' action. If you want a variable action, consider something like moving a pivot point instead.

Dave

hi bryn, i'm working on the similar project (neural network controlled robot(using servos)here i dont want to use microcontroller)

coming to your transit speed issue...thinkng from control signal point of view...a servo needs the command signal(i.e the specific width corresponding to specific postion)repeated 50 times a second(in general)...which implicates that the servo needs atleast a second(or

3/4 second) to move from one extreme to another(given extreme commands)...and you have talked abt the servo with high transit speeds(does it mean less repeatation frequency of signal)...i'm new to robotics and servo control...and need a guidance from experienced people like you to get started with my project...though i have built analog signal controlled circuit for servo...not yet sure when implemented with loaded servo(wing weights)

waiting for your insights and guidance

-sanjay kumar boddhu,