Siemens SINUMERIK 840D sl Function Manual page 442

K3: Compensation
5.6 Friction compensation (quadrant error compensation)
MD32580 $MA_FRICT_COMP_INC_FACTOR
(weighting factor friction compensation value with short traversing movements)
This weighting factor specified in the above machine data automatically takes effect when
friction compensation is activated (conventional QEC or QEC with neural networks) acting on
all positioning movements that are made within an interpolation cycle of the control.
The input range is between 0 and 100% of the calculated compensation value.
Control of learning process duration
As described in previous sections, the duration of the learning process is dependent on
several parameters. It is mainly dependent on the following values:
● Coarse quantization ($AA_QEC_COARSE_STEPS)
● Measuring time for determining the error criterion ($AA_QEC_MEAS_TIME_1 up to
● Number of learning passes
● Detailed learning active [yes/no]?
● Fine quantization ($AA_QEC_FINE_STEPS) (only if "detailed learning active = yes" is
● Directional compensation active [yes/no]? ($AA_QEC_DIRECTIONAL)
● Duration of reversing movement
The setting "Detailed learning active = yes" causes a significant increase in the time required
for learning. It must therefore only be used where precision requirements are high. It is
necessary to check whether these requirements only apply to certain acceleration ranges. If
so, detailed learning only needs to be performed section by section (see "Section-by-section
learning y/n?"). The number of learning passes must be reduced in any case.
If the reference NC programs mentioned above are used with the recommended parameter
values, the following times have been determined for the learning process time:
● Detailed learning not active: approx. 6.5 min
● Detailed learning active: approx. 13 min
442
$AA_QEC_MEAS_TIME_3)
selected)
Function Manual, 09/2009, 6FC5397-1BP20-0BA0
Extended Functions