Siemens SINUMERIK 840C User Manual

Page 356

Advertising
background image

09.95

6 NC Machine Data (NC MD), NC Setting Data (NC SD)

6.7 Axis-specific MD 2 (axial data 2)

Torque distribution torque compensation controller

(SW 4.4 and higher)

1344*

Default value

Lower input limit

Upper input limit

Units

500

0

984

Active on
NC Stop

With MD 1344*, the input variables of the torque compensation controller are weighted to
permit a parameterizable torque distribution over both drives according to the respective
moments of inertia.

The standard parameterization of 500 _ produces a distribution according to the values in
drive machine data 1725 (normalization torque setpoint interface) for both axes. If a different
distribution is to be used, MD 1344* must be calculated according to the following formula:

M

desired

slave

MD 1344* =

––––––––––––––––––––––––––––––––––––––––––

· 1000 ‰.

MD 1725

slave

M

desired

slave

+ M

desired

master

*

–––––––––––––

MD 1725

master

M

desired

is the desired torque distribution between master and slave.

2nd compensation time constant (as from SW 4)

1364*

150

0

9999 9999

0.1 ms

Active on

NC Stop

MD 1236* defines the recovery time for the compensation setpoint pulse when operation is
without its adaptation. MD 1236* applies to neural and conventional quadrant error
compensation. If the adaptation of the recovery time constant (neural QEC only) is selected
(MD 1812*, bit 2 = 1), MD 1236* defines the filter time constant in the middle of the working
range, MD 1364* defines the value for acceleration 0.

Learning rate (as from SW 4)

1368*

Default value

Lower input limit

Upper input limit

Units

100

1

500

%

Active on
NC Stop

The learning rate for neural quadrant error compensation defines how quickly the neural
network learns the optimum characteristic during the neural quadrant error compensation (MD
1812*, bit 0 and 1 set) active learning phase. A high learning rate means a short learning
phase, which in the worst situation can cause instability. A low learning rate is recommended
for learning processes during normal operation as otherwise the characteristic will be changed
if even the smallest disturbance occurs when the speed goes through zero. Reproducibility is
then no longer possible.

©

Siemens AG 1992 All Rights Reserved 6FC5197- AA50

6–171

SINUMERIK 840C (IA)

Advertising
This manual is related to the following products: