| |
| playfair | Key Veteran - Location: Rochester, NY - |
ALL of us have pondered the difficult question of servo selection.
Besides the obvious trade-offs of price vs. performance, the broader issue is how to interpret the given specifications to make the best decision.
Manufacturers only provide us with an unloaded speed and a stall torque, neither of which is directly useful in choosing the best servo for a particular application.
Part 1 of this topic will deal with investigating these specs at more detail. A few others have done similar testing or made circumstantial assumptions, but I wanted to answer some questions and be able to dig a little deeper.
I started off by building a test fixture consisting of a bracket to support the servo, an electric resistance brake, and an external potentiometer.
 
The brake (brass colored cylindrical thing in the above right picture) varies in torque based on input voltage. I simply calibrated it at a given moment arm length using a force gauge.
The above line equation is then used with a known servo arm length to give us ounce-inches of torque.
The servo is controlled using a standard transmitter/receiver, with the total movement triggered by a switch. Endpoints were adjusted to achieve 60 degrees of travel (which varied even within manufacturers).
To acquire the data, I wrote a program in Labview. It records position and current in relation to time. The complete rig (including power supplies) added little to my already cluttered work bench!
The following screen captures show the behaviors for a S9252 digital and a S9402 analog servo at 5v pushing 16 oz-in for 60 degrees.
The upper graph is position in degrees, the lower is the current signature.
The pulses before and after the movement are the "buzzing" noises we hear as a servo tries to hold position.
Besides the obvious frequency differences, notice how the target endpoint is gradually reached by the analog servo, but the digital servo will use more current to get-to-the-point and decelerate!
This brings us to the good stuff- A chart comparing several servos I had (or borrowed). Though the unloaded speeds for all of them were similar to the given specs, I ended the test once the transit time dropped severely. I couldn't bare to see those little guys moaning so much, plus that extreme is definitely NOT where we want to operate!
Manufacturer specs:
S9252- .14sec@92ozin (4.8)
DS8317- .15/.12@69/84 (4.8.6.0)
S9351- .16/.13@153/192 (4.8/6.0)
Z9000- .06@128 (6v)
HS5975- .13/.10@91/109 (4.8/6.0)
BLS451- .13/.10@118/147 (4.8/6.0)
One observation that stemmed from measuring degrees of travel was noting the servos didn't necessarily return "home" as the loads increased. I started tracking this behavior, and formed the following graph showing the "accuracy" of the same servos.
A few degrees of this "slop" is likely due to the support structure flexing, but the trends should be relative.
This information is swell and all, but how does it fit into the "big picture"? If you've stayed interested this far, you are surely asking, "What does my servo see for a load during flight?"
Go to SERVOS: An in depth look - Part 2
The sky is our canvas |
| 04-14-2008 05:33 PM | | | |
)
