Well, before I start I need to say that I just received my belts & pulleys and didn’t had a chance to actually test it on the field, so I’ll update bottom part of this page once I do it. The accuracy is a multi-component challenge. It depend on the way and accuracy you have polar aligned the mount, the accuracy of the clock and the software, the accuracy of the stepper motor, the accuracy of the gearing system and etc. Some of those components vary because of the temperature, weight of the telescope/camera equipment and etc. Thus you can never be sure your system is dead accurate. For the purposes of visual observing however most of those components have small variations, which is different when you do astrophotography. Let me give you an example:

The Arduino have variable speed of “Slewing To”. Because of the code I wrote it can sometime go above the steps requred to the object or stop before it reaches them - e.g. ±4 micro steps.

Now in my setup, I’m using 144 tooth gear, 3:1 belt and pulley, 200 steps/rev. stepper motor on 1/16 micro steps mode. This means (see the XLS file from Download section) that I have 0.9375 arc sec./micro step x 8 micro steps = 7.5 arc sec. is the Error.
Well, considering that the Earth rotates with 15 arc sec/sec. it is not an issue.

Another consideration could be the Field of View:
In most cases I’ll use C8 (2032 mm. focal length, 203 mm. aperture) with let’s say 26mm (52 deg) eyepiece for finding the object. This actually gives me a Field of View = 0.67 deg. = 2412 arc sec. So, the error in my case would be 0.31%.

In my code, I introduce 4 possible errors (or at least those I can think of):
1) Error from the clock for tracking – microseconds and time to execute;
2) Error from calculating LST and HA – rounded values and size of variables;
3) Error from SlewTo command - ±4 micro steps;
4) Error from Polar alignment - Vixen polarscope have 3 arc min accuracy!