Siemens SINUMERIK 840D sl Function Manual page 427

Quantization of characteristic
The quantization, and thus the resolution, of the characteristic is defined via the two
quantities fine quantization ($AA_QEC_FINE_STEPS) and coarse quantization
($AA_QEC_COARSE_STEPS). The finer the resolution, the higher the memory requirement
and the longer the duration of time required for the learning phase.
The number of memory locations required and the total number of quantization intervals is
calculated by the formula:
Number of memory locations = $AA_QEC_FINE_STEPS * ($AA_QEC_COARSE_STEPS+1)
Up to 1025 memory locations per axis can be reserved. In this way, a sufficiently high
resolution is achieved for high precision requirements.
The following 3 diagrams illustrate the meaning of the characteristic values for coarse and
fine quantization, and their effect on the teach-in period, as a function of the parameter
"Detailed learning active y/n". Three cases are distinguished for better understanding.
Case 1:
Coarse quantization > 1; fine quantization = 1 (special case; usually the fine quantization is
in the region of eight):
In this case, the interpolation points of the characteristic are determined solely by coarse
quantization (see diagram below).
Figure 5-21
Case 2:
Coarse quantization > 1; fine quantization > 1; "Detailed learning" is deactivated (this setting
is the default):
In this case, discrete linear interpolation is used for fine quantization between the
interpolation points of the coarse quantization.
The learning duration is identical with 1 because learning only occurs at the interpolation
points of the coarse quantization.
Extended Functions
Function Manual, 09/2009, 6FC5397-1BP20-0BA0
Coarse quantization of characteristic
5.6 Friction compensation (quadrant error compensation)
K3: Compensation
427