Checking servo linearity and centering

[blegas78]

Ok, well maybe asking a question about the quality of a specific servo will not get much feedback, so here is a more general question since I already have a few micro servos that I would like to test out.

Since I have never checked the centering and linearity of standard servo motors, what are good methods for doing so? I assume this to be very simple, such as calibrating the servo at 0 degrees and 60 degrees, then setting the servo position to 120 degrees to see how far off it is.

What sort of error is acceptable/good/excellent for linearity/centering error?

[Matt Denton]

For starter I presume you have read this: http://hexapodrobot.com/forum/viewtopic.php?f=21&t=13

I have a calibration rig, essentially a rotary encoder with a small gear attached, I then attach a large gear onto the servo horn of the required servo and mount the two so that they mesh nicely. The gearing up helps resolution on the encoder, currently I'm using a 128ppr type. I then have a small piece of firmware on a pic that reads the encoder position converts to degrees and squirts it out to a serial port.

So plugging in the servo I center and reset the encoder pos, then send equal +500u / -500u Sec pulse to drive the servo in each direction, and of course not down the degrees travelled. I repeat this for 3 or more servos and then take an average for my servo range value.

For linearity, I would usually be happy with +/-2 degrees worse case. I tried some cheap servos once that gave readings of say +44, - 53 and all were different!

Hope that makes sense.

[blegas78]

Thanks, that makes perfect sense. I mostly just didn't know what "good" linearity meant quantitatively.

And wow, +44/-53 is pretty incredible...