The LMD18245 from National Semiconductor appears to have a problem when the motor current direction pin is changed while the current through the device is not 0 Amps. This bug was found by RadekCX, a CNCZONE member and verified by alan@fromorbit.com.
The ELS revison 1.03i and earlier also had this bug but it didn't appear to affect the motor speed in either direction as described by others. In either case, though it seemed prudent to fix the problem and make the waveforms applied to the motor identical for each motor direction. Below are scope photos showing the before and after waveforms.
Revision 1.03i Motor Moving left
Above is the way the current waveform for A (Blue) and B (Yellow) is supposed to look in both motor directions.
Revision 1.03i Motor Moving right. Notice how the current doesn't sharply drop to 0 amps.
By adding two more lookup tables to the code and testing motor direction the ELS now has the following current profile identical in both directions.
Revision 1.03m Motor Moving Left
Revision 1.03m Motor moving Right