Resolution in an absolute sense is difficult to answer with lack of accurate measuring possibilities.
The encoder gives 12 bits per half turn so (1:2 gear) so ~ 0.044 degree steps.
For azimuth I get about 0.15 degree from the setpoint, but there is a little play between outgoing axis and encoder gear.
Azimuth only has to fight with the inertia. It already moves with a PWM input of 10 (out of 255).
Elevation is worse as it has the full load, so gravity and inertia and torque depending on amount of balancing.
The whole structure has flexibility here and that is after the encoder. From the encoder I also get at about ~0.15 degree to setpoint but with a long settling time as it is the I term from the PID that has to get it there. Not clear what the accuracy in motion (tracking) is. The rotor only starts moving with about 115 on the PWM. So there is quite a lot of stiction and friction, which makes it jump and overshoot when it tries to correct small setpoint errors.
Overall I would judge it is more like 0.5 to 1 degree accurate. I cannot easily judge if this is the tripod, the ABS 40mm pipe, or the rotor. For the pipe I want to try aluminium but you need accurate pipe to get it through the bearings. The ABS I had to turn down on the lathe.
You can also checkout: https://tysonpower.de/blog/diy-az-el-antenna-rotator-for-under-150
It really depends on the weight of the dish.
I’m also considering a diseqc rotor like:
(dutch magazine: http://www.kunstmanen.net/index.php/2012-10-01-09-31-09/archeif)
Diseqc I believe does something like 0.1 degree steps.
I’ll create some build blog somewhere the next 2 weeks or so and update here.