The cause for this is the way that the DSP type II servo loop was implemented in this specific design. The internal conversion loop is continually "monitoring for change” from the DPRAM commanded angle register. On initial power-up (or soft reset), all variables, including the initial angle are “zeroed”. If the initial commanded angle is “0” degrees, then there is “no change” for the loop to work from. As a result, the loop doesn’t administer the "change from previous” variable and will occasionally initially output 225 or -135 as observed.

The “workaround” is to program an angle one or two LSB from zero, initially. This initialization-only action will “kick-start” the loop and administer the initial “change in the variable” the loop needs. Programming one LSB (i.e., 0.0055 degrees) and then back to zero degrees is enough to “kick start” the conversion. Once the loop is “kick-started,” there is no issue.