I whipped up a jig using some scraps of wood and plastic that I had. I used one of the servo horns that came with the servo, and my highly sophisticated coupler consists of a scrap of 1" round aluminum that was drilled to accept the shaft of the encoder, with a setscrew to hold things in place. The encoder is a surplus Oak Grigsby one from somewhere. It has 128 pulses per revolution, or 512 edges when you start to do quadrature on it. Bigger...

I noticed some anomalies with the data so I decided to document these separately. Details...

The first test I did was to position the servo at the position given by a 1 millsecond pulse, and then moved it to the position with a 2 millisecond pulse. Recording only the B channel, I recored the time that each change in encoder position occured and plotted them. I tested 2 Hitec HS-645MG servos and one HS-311 servo. Bigger...

In this test, I set the pulse width to 500 usec, and incremented it by 100 usec, recording the encoder position for each pulse width. It's interesting to note that the two HS-645's didn't quite move the same amount and the two extreme positions were at different places. HS-645-1 started to move before HS-645-2 but stopped sooner. Bigger...