Tm450 Acopos Control Concept and Adjustment

8/3/2019 Tm450 Acopos Control Concept and Adjustment

1/51


2/51

2 TM450 ACOPOS ControlConcept and Adjustment

Prerequisites

Training modules: TM410 MotionComponents Basis

Software: AutomationStudio 2.5 or higher

AutomationRuntime 2.80 or higher

ACP10-SW V1.160 or higher

Har dware: 8V1xxx.yy-2


3/51

Table ofcontents

1. INTRODUCTION 4

1.1 Objectives 5

2. THE BASICS OF CONTROL TECHNOLOGY 6

3. CASCADED CONTROLLERCONCEPT 10

3.1 Overview 10

3.2 Set value generator 11

3.3 Predictivepositioncontroller 15

3.4 Speed controller 18

3.5 Current controller 22

4. THEORETICALLYDETERMINING THE CONTROLPARAMETERS 23

4.1 Speed controller 23

4.2 Positioncontroller 24

5. PROCEDURE FORSETTING THE CONTROLLER 26

5.1 Generalinformation 26

5.2 Speed controller 295.3 Positioncontroller 35

5.4 Limitparameter 41

5.5 Overview 44

6. SAVING THE CONTROLLERSETT INGS 45

6.1 NC INITparameter module 45

6.2 ACOPOS parametertable 45

7. SUMMARY 46

8. APPENDIX 47

ACOPOS ControlConcept and Adjustment TM450 3


4/51

Introduction

1. INTRODUCTION

The high level ofperformance of a servo drive 'scontrolleris crucial for thequality,precision and dynamiccapabilitiesduring a process. This means

that the controlconceptas well as the controllersettings are decisivefactors.

Fig. 1 ACOPOS servo drive and motor

During thistraining we will first learn the basicsbefore taking a step-by-step lookat the controlconcept of the ACOPOS servo drive.

We will then take some time learning how to determine the optimumcontrolparameters. Some of the comp onents from MotionComponents

(trainingmodule TM410) will help us throughout thistrainingmodule.



5/51

Introduction

1.1 Objectives

The courseparticipant will become familiar with the structure and theeffectiveness of the ACOPOS controlconcept.

Each courseparticipant will be able to adjust and optimize the controlparameters.

Fig. 2 Overview



6/51

The Basics of Control Technology

2. THE BASICS OF CONTROL TECHNOLOGY

To begin our training, we would like to briefly review a few basicsbeforediving intothe details of the ACOPOS controlconcept.

Let'sstart by askingourselves the followingquestion:

Why does aservo drive need a controller?

The goal of a servo drive is to reach a positionas fast and as accuratelyasposs ibleusing a motorand variousmechanics.

Fig. 3 ACOPOS servo drive and motor

The first thing the controllermust know iswhere to send the motor. This isgenerallyreferred to as the set position. Incontroltechnology, a referencevariable isused to dothis.

The controllermust receivesome information to find out where the motorispresentlylocated.This informationisobtainedusing an encode r. Theencoderis a measuringelement, which provides the controllerwith theinformation(actualposition of themotor) via feedback.

Fig. 4 ACOPOS servo drive, motorand feedback



7/51


A lag error or control deviation occurs when the set position and theactualposition do not match.

This lag error iscompensated for by the ser vo drive. This is why it outputs

amanipulated variable. This manipulatedvariableaffects the controlledsystem (in thiscase, the motorand the subse quent mechanics)so that theactualpositionreaches theset position.In our case, the actualpo sition isthe control variable.

All of thistogetheriscalled the closed control loop.

Fig. 5 ACOPOS servo drive, motor, closedcontrol loop and disturbancevariab les

There are even more externalfactors that affect thissystem. These types of

factors arecalleddisturbancevariables, and alsomust be comp ensated for by the controller. A suspended load is one example of a disturbancevariable.

Differencesbetween closed and open loop controllers:

Unlike a closed loop controller, an open loop controllerdoes not havefeedback. Thismeans we cannot determinewhen, how or whetherthe goalhasbeen reached. An open loop controller isgenerallyused for

conventionalfrequencyconverters.



8/51


But now wha t happens within a closedloop controller?

A closed loop controllercan be made up of one or more parts of transferelements.These transferelementsreact differently to input values.

The controllers on the ACOPOS servo drive are generallyma de up ofaproportionelement (P-element) and an integral element (I-element).

A P-element imm ediatelyreacts to an input value jump with a proportionaloutput value jump. The size of the jump on the output isdetermined by afactor.This is labeledas thefactor"kv".

Fig. 6 Reaction of aP-element

The output value increasescontinuously in the form of a ramp if the inputvalue of an I-element jumps. The slope of this ramp and the rate at whichthe output value increasesdepends on the time. We will label the time asintegralaction time "tn".

Fig. 7 Reaction of an I-element



9/51


A PI controllercontains a P- and an I- section. The output values of bothelementsareadded together. As a result, the PI-controllerreacts to ajumping input value asfollows.

Fig. 8 Reaction of a PI-controller

The P-sectionallows the controllerto react immediately to a change in the

input value. A remainingcontrollerdeviationwould occur if only one P-sectionwas used alone.TheI-section integratesthis,therebycompensating for the deviation.

The ACOPOS servo drive isequipped with a high-performanceprocessor.Amo ng othertasks, thisprocessoralsocalculates the controlleralgorithms.This is why the term "digitalcontrol" isused. The differencebetween anana log(continuous)controllerand a digital(discrete time) controlleris thatthe controllerdeviationsisscanned in a correspondingtimeframe (cycle).The duration of this cycle shouldalways be the same (= jitter-free), butshouldalso be as short asposs ible to receive the changes in the controllerdeviationasfastasposs ible.



10/51

Cascaded Controller Concept

3. CASCADED CONTROLLERCONCEPT

3.1 Overview

The cascaded controllerconcept is the core of the ACOPOS servo drive.The positioncontroller,speed controllerand current controllerarecascaded starting with the setvaluegenerator. As a result, the manipulatedvariable of the higher-levelcontrollerbecome s the referencevariableforthe lower-levelcontrollers(e.g.: the positioncontrollerdetermines the setspe ed for thespeed controller).

Fig. 9 Cascaded structure



11/51


3.2 Set value generator

Fig. 10 Principle

3.2.1 Basismovements

As seen earlier, the referencevariable isprovided for the positioncontrollerby a set value generator. This is done with a cycle time of 400s.

The job of thisset value generatoris to create a movement profile after acommand forexecuting a basismo vement. The course of this profiledep ends mostly on the basismovement parameters(targetposition,acceleration,etc.).

Fig. 11 AutomationStudioonline help, basismovements



12/51


The following figure illustratesthis type of profile and the parametervaluesused:

Fig. 12 Trace display of the position,speed and acceleration(derivative of the speed)

We can see from the chart that jumps occur at each end of theacceleration.Thesejumps are calledjolts.These occur when there is a

bend in the respectivespeedcurve.

Generallythis type ofbehavior is not desiredbecause the motor must

generate ahighertorque. Furthermore, a high load isplaced on themechanics and the entiresystem vibrates.

This is why the set value generatorin an ACOPOS servo drive isprovidedwith joltlimitation.



13/51


3.2.2 Jolt limitation

As we have already learned, the jolt occurs due to a change that causes abend in thespeed curve. If the speed isslowly increased at the beginn ing

or slowlyreduced at the end of the accelerationphase, the rectangularacceleration profile becomes a trapezoid. The followingchart illustratesthis

point forus (the basismovement parameters were set the same as in Fig.12):

Fig. 13 Trace display of the position,speed and acceleration with active jolt filter

A jolt filter time can be set for the ACOPOS servo drive to generate amovement profile with jolt limitation.

The jolt filter time is the time required for acceleration from zero to themaximum valuedefined.Joltlimitationuses a linear filter duringruntime.

In Fig. 13 we can see that the jolt filter time "t_jolt"has been set to 0.03seconds. Themeasurement cursors in the lowerchart diagram indicate thatthe rise time of theacceleration is the same as the jolt filter time.



14/51


An active jolt limitationextends the time for the set value generation(seeFig. 14).However, in many cases the positioning goal can be reachedsooner because thesettling time of the system isconsiderablyshortened

by the lowerjolt load.

Fig. 14 Timing diagram of a movement with (bold line) and without(dotted line) jolt limitation

Note:

The "t_jolt"parameterislocated in the limit values:

Fig. 15 AutomationStudioonline help system Limit values

A value between 0.0 and 0.2 seconds can be defined for "t_jolt".



15/51


3.3 Predictiveposition controller

The referencevariable of the positioncontrolleriscreated by the set valuegenerator.The ACOPOS servo drive receives the controlvariable(currentmotorposition) via themotor'sencoder system and a correspondingencoderinterfacecard. The controldeviationisdetermined from these twovariables, which results in a new manipulatedvariable for the lower-levelspe ed controller.

Fig. 16 Block diagram of the positioncontroller

The positioncontrolleris implementedas PIcontrollerwith anti-windup(manipulatedvariablelimitation) and "predictive" feed forward (inputcontrol).



16/51


The proportionalelement with the factor"kv"causes an imm ediatechangein the setspeed in the event that a lag error occurs. Changes in the setvalue or disturbancevariables can cause these type ofcontroldeviations.

The I-element with integralaction time"tn

" isused to compensate for

stationarydisturbancequantities (e.g. suspended loads).

The manipulatedvariablelimitation isimplementedusing the parameters"p_max"and"i_max".These values limit the maximum effect of the P-section and the I-section.

The feed forward is the predictiveelement of the positioncontroller.

Fig. 17 Prediction, feed forward

A feed forwardspeed ("v_feed")results from a differentiation of the setposition(s_set).ThePI-controllerestablishes a correctionspeed ("v_corr")from the lag error (s). These two speeds are added togetherto produce theset speed (v_set) for the subsequent speedcontroller. The feed forward isthenpredictive if the se t position issent to the PI-controllerwith a delay(t_total). The set value should be delayedas long as the delay time of thecontrolledsystem. This is why the set speed (t_total t_predict) isintroduced to the spee dcontrollerfirst. If the correspondingsetposition ofthe PI-controller isthen accepted, thecontroldeviation isthen smalleras aresult of the already fed set speed value.Thispresents a load on the PI-controllerbecause it still has to compensate for the remainingdeviation.Withoutspeed input control, the PI-controllerwouldhave to take care ofthe set speedby itself. This improves the referencebehaviorand thedynamicproperties of the drive.



17/51


Calculation of "v_corr" which results from the lag error "s":

First, the speed "v_p"resulting fromthepropor tional gain iscalculated

and limited to"p_max":

v_p = kv * sif ( v_p > p_max )

v_p = p_maxelse if ( v_p < -p_max )

v_p = -p_max

This value and "i_max"are used tocalculate "i_limit"and the speedresulting from the integral gain "v_i"islimited to this value:i_limit = i_max - |v_p|

if ( i_limit < 0 )i_limit = 0

v_i = f(v_i, s,kv,tn)

if ( v_i > i_limit )

v_i = i_limitelse if ( v_i < -i_limit )

v_i = -i_limit

Fig. 18 AutomationStudioonline help, calculation of "v_corr"

Finally "v_corr= v_p + v_i" can be calculated.

The set speed isconverted from the configurablemeasurement system(unit/sec) to the physical motor encoder system (rev./sec)before being

pas sed onto the speedcontroller.

Lag error monitoring isalsoperformed in the positioncontroller. An E-Stopisexecuted if the controldeviationexceeds a configurablethresholdvalue.



18/51


3.4 Speed controller

3.4.1 Speed controllerGeneralfunction

The speed controller's job is to determine the differencebetween themanipulatedvariable of the higher-levelpositioncontrollerand themea sured speed. A manipulatedvariable for the lower-levelcurrentcontrolleristhen generated using a PI-controllerwith anti-windup.

Fig. 19 Block diagram of the speed controller



19/51


The set speed value from the positioncontroller isfirst fed through aninterpolator.Thisisnecessary because the positioncontrollerruns at acycle time of 400s, whereas thespeed controllerruns at 200s. The valueisthen limitedusing the maximum motor speed.

The actualspeed isdetermined by differentiating the encoderposition. Thevalue isthensent through the speed filter. The value isconverted from"incr./sec" to the unit "rev./sec"before it can be subtracted from the setspeed.

We will take a more detailed look at the functionality of the speed filter alittle later.

The P-element with the factor"kv"allows the controllerto reactimm ediately to anycontroldeviation.This makes thiselementdecisive for

the dynamics of the speedcontroller.

The I-element with integralaction time "tn" isused to compensate forstationarydisturbancevariables (e.g. load torque).

The output of the PI-controllercan be filteredusing a notch filter. The

function of thenotch filter will be explained in a later section. Before

the set current can beprovidedas the referencevariable for thecurrent controller, the value is limitedusing a torque limiter. This torquelimitationalsodetermines the anti winduplimits for the speed controllerI-element.

3.4.2 Torque limiter

The torque limiter ismostlyused to protect the motorand the ACOPOSser vo drive from the followingrisks:

The ACOPOS servo drive cannot output mor e current than the motorcan handle(motorpeak current).

The motor'sstator current cannot exceed the ACOPOS peak current.

By default, the torque limiter is initialized using the smallerof the followingtwo values:

Motor peak currentACOPOS peak current



20/51



3.4.3 Signal filter

Signalfilters can be used on ACOPOS servo drives to separate highfrequencydisturbances (e.g. signalnoise) from a desiredsignal and tosuppress a resonancefrequency.

Disturbance in the encodersigna l can be caused by one ormore of thefollowingposs ibilities:

Coupling of disturbances on the commun icationpath (encodercable).

Quantizationnoisewhen converting the analogsignal to a digitalform. Thismostlyoccurs when evaluating a resolversignal with lowresolution.

With warp-resistant drive mechanics (e.g. direct load coupling to the motorshaftusingfasteningdevices) the mechanicalsystem can be subject tooscil lations due to the closedcontrol loop ("two-massoscillation").Thesetypes of systems generallyhave a resonancefrequency in the range from700 to 1500Hz. This isfurtherdependent on the followingfactors:

Rigidity of the mechanicalsystemMass inertia of the mechanicalsystemPhysical layout of the system.

Speed filter

As described in section 3.4, the actualspeed isfiltered,beforebeingprocessed in thespeed controller.This filter is a "speedfilter" and functionslike a low pass. High-frequencydisturbances can be filtered out from thespe ed signalusingthis low passbehavior.Thisallowsus to achievehighercontrollerquality.

Caution:

Parts of the desiredsignal will also be filtered out if the limit frequencyof the lowpass filter isset too low!


21/51


Notch filter

The frequencyrange of the set current, which causes the mechanicalsystem tooscillate, can be filtered for the current controller.

Caution:

The notch filtershould only be used on mechanics with a rigid coupling(e.g. direct drive). This filter should not be used for connectionssuch asbelts or gears!

Furthermore, it can only be used if the existingmoments of inertia arealwaysconstant!

The resonance frequency of the system couldshiftas a result ofmechanicalwear.This means that over time the defined filter can loseitseffectiveness .

The notch filter is only effectivewhen the resonance frequency is in therange from 700 to 1500Hz.

The filter has the highestamount of damping at the "notchfrequency"entered (=resonance frequency of the mechanicalsystem).There is arange (band width) aroundthisnotch frequency in which the dampingis

lowerthan 3dB. The smallerthe band width isset, the strongerthedamping is in the notch frequency.

The notch filter can be used (once all of the requirements for use havebeen met) to increase the controllergain factors,withoutcausing the entiresystem tobecome unstable.



22/51



3.5 Current controller

The current controller ismade up of PI-controllers (like the position andspeedcontrollers). The correspondingparameters are automaticallydetermined by the servo drive using the mo tor parameters and the specificACOPOS parameters. This means that we do not have to spend time in alater sectiongoing over how to set these values.

The current controlleruses itsmanipulatedvariable to control the IGBTs(Insulated Gate Bipolar Transistor).These then output a pulse widthmodulated (PWM) currentsignal to the motor.

The current controllerfunction at different cycle timesdepending on thePWM switchingfrequency:

PWM frequency Display unit Cycle time

5kHz 8V128M.00-2 200s

10kHz 8V101x.yy-2

8V1090.00 -2

8V1180.00 -2

8V1320.00 -2

8V1640.00 -2

100s

20kHz 8V1022.00 -2

8V1045.00 -2

50s


23/51

TheoreticallyDetermining the Control Parameters

4. THEORETICALLYDETERMININGTHE CONTROL PARAMETERS

In the previoussections we learned how the ACOPOS servo drivecontrollersarestructured and inter-related. Now we want to find the

correspondingvalues for thecontrolparameters. These values can becalculated or determinedempirically ifsome of the requirementshave notbeen met.

We can use the followingformulas to determine good output values for thecontrolparameters in the event that we don't already know the system'stotal moment of inertiaand if the load is fixed to the motor. Inmostcases, we will achieveeven better controllerbehaviorby making fineadjustments to the parametervaluesmanually.


Replacement time constant TI of the current controlloop:

TI

2 1 0.000075 2 SwitchingFrequency

Summation of the individual time constants to a replacement time constantT_v:

T v TI Td ead v Tfilter

Tdead_v = 0.000175 s (encoderinterfacedead time, speed determination undscann ing)

Tfilter ("t_filter"parameter)= Filter time constant of the speed filter

Proportional gain of the speed controller:

kv JT

2_v kt


24/51

J = total moment of inertia (Jmotor + Jbrake + Jload )

kt = Torque constant of the motorbeingused [Nm/A]

Integralaction time of the speed controller:

tn 4 T _ v



25/51



4.2 Position controller

Summation of the individual time constants to a replacement time constantT_p:

T_p Tinterpol 4 T v Td ea d p

Tinterpol = Dead time resulting from interpolator(0.0001s)

4 x T_v = Replacement time constant of the speed control loop

Tdead_p = Dead time resulting from scanning(0.0002s)

Proportional gain of the positioncontroller:

kv1

2 T _p

Integralaction time of the positioncontroller:

tn 4 T _ p


26/51

=


Example:

Configuration of the 8MSA2S.E0 motor with ACOPOS 8V1010.00-2

withoutload:

kt = 0.46 Nm/A

J= 0.06 kgcm

Switchingfrequency= 10000 Hz

Speed controller:

TI 2 1 0.000075

2 SwitchingFrequency 2 0.000075

1 0.00025 sec

2 10000

Tv

TI

Tdead v Tfilter = 0.00025 0.000175 0 0.000425 sec

kvJ

T

2=

_v kt

0.0000060.000425

20.46

0.136As / Rev.


27/51

tn 4 T_v = 4 0.000425 0.0017 sec

Positioncontroller:

T_p Tint erpol 4 T v Tdead p = 0.0001 4 0.000425 0.0002 0.002 sec

kv1

=1

2501

2 T _p 2 0.002 sec

tn 4 T_p = 4 0.002 0.008 sec



28/51

Procedure for Setting the Controller

5. PROCEDURE FORSETTINGTHE CONTROLLER

Inthissection we will learn about a possibility for determiningcontrolparameters, which has been proven time and time again in the field. This

will allow you to checkandmake fine adjustments to values that havealready been calculated.If thisisnotposs ible,suitablevalues can bedeterminedempirically.Inthiscase, the valuesspecified in thecorrespondingnotes can be used as start values.

5.1 General information

The controlparameters only really have to be determinedwhen themechanicsarealready put together. When dealing with a machinewherethe axis is loadedwithdifferentma sses, the parameters mustbe tested

both without load and with the highest load. The parameters shouldalso

be tested at differentspeeds and accelerations.

These tests couldresult in the need for a compromise.

Globally valid control parameters cannotbe used because all mechanicshavedifferentfeatures.

To determineparameters from cascaded controllers, it isbest to start fromthe bottom(last) controllerand work up. In our case, thismeans that wewouldstartby settingthespeed controllerfirst followed by the positioncontroller. The current controllerisautomaticallyconfigured by theACOPOS servo drive.

Fig. 20 Orderfor se tting the controllers



29/51


It isusually the goal to set the controlleras "tough"as poss ible. Acontrollercan be considered"tough"when a disturbancevariable iscompensated for asquickly and perfectlyas po ss ible.

Example:

We want to manuallyrotate a flywheel mass which ismounted to themo torshaft.

The controllers are set "soft" if the flywheel mass can beeasilyrotated.The controllers are set "hard" if the flywheel mass isdifficulttorotate or cannot be rotated at all.

Howe ver, sometimes it is not the goal to set the controlleras tough asposs ible.This isbecause a tough controllercan cause quick heating, ahigherload on the mechanicsandthereforemore wear. This is the reasonwhy a compromisemust often be found whendetermining the parameters.

We will be using the MotionComp onents test window to help us set thecontrolparameters. This allowsus to change and initialize the control

parameters online.Thisalsomakes itposs ible to startpositioningmo vements using any basismovementparameters. We can alsoconfigure,start and evaluate the trace in the MotionComponentstest window.

The behaviorof the controllerduring a typical mo vement can be closelydeterminedbyselectingsuitableparameters for tracing.Therefore, a basismo vement (e.g. relativeorabsolutemovement)isusuallystarted and therespectiveparameters are recorded.

Furthermore, the controlparameters should be selectedso that oscillationof therecorded valuesisminimal.



30/51


Examples:

Fig. 21 Strong oscillations

Fig. 22 Almost no oscillations

The user must then checkto make sure that all of the requirementshavebeen met(e.g. lag error within the tolerance)once all parameters havebeen set.

If thisis the case, make sure that there are reserves for the gain factorbecause thesystem can behave differently due to mechanicalwear.Therefore, a correspondingreserve (approx 1/3) should be taken from thedeterminedvalues.



31/51



The following three parameters can be set for the speed controller.Theseare located in a subgroup of the controlparameters:

Fig. 23 AutomationStudioonline help system Controller

Note:

The followingparameters can be configured for the trace when settingthe speedcontroller:

Set speed: ACP10PAR_SCTRL_SPEED_REF (ID 250) [1/sec]Actual speed: ACP10PAR_SCTRL_SPEED_ACT (ID 251) [1/sec]Current controller:Setstator current of quadrature component

[A]

The scan rate should be set as low as poss ible(approx. 0.2 to 4msec).A triggerevent can be used to start the trace as accuratelyas poss ible(furtherinformationcan be found in the AutomationStudio Onlinehelp).

The "kv" and "tn"parameters of the positioncontrollershould beinitialized withthe value "0"so that only the speed controller isactive

Note:

The value from "currentcontroller:set stator current of the quadraturecomponent"is the torque generatingcomponent of the set current.

The peak value of the current isdisplayed in the MotionComponentstrace. Thisvaluemust be divided by the factor2 (1.414 ) to compareit with the specifications of the motor parameters and the ACOPOSser vo drive parameters.



32/51


5.2.1 Proportional gain "kv"

The most importantparameterof the three speed controllerparameters isthe gainfactor"kv".Thisparametersignificantly determines the dynamic

properties of thiscontroller. The goal is to set the value as large asposs iblewithoutcausing the system tooscillate.

Note:

1/5 to 1/10 of the nominalmotor current (In) can be used as startingvalue forthisfactor.

Example:

Differencebetween the set and actualspeed: 1 Rev./sec

Value of the factor"kv": 2 As/Rev.

Output value of the P-element: 2 A

5.2.2 Integralaction time "tn"

For most applications it is not necessary to use an integralaction time in

the speedcontroller("tn

"= 0s). However, this time value should not be settoo low if an applicationdoes require an I-section (e.g. to compensate for

load-sidedisturbances, poor (soft) loadcoupling, high speedprecision).Otherwise, the tendency for oscillation isincreasedinthespeed controller.

Note:

100msec (0.1sec) can be used as starting value for the integralactiontime ifyou want to use the I-section.The value can then be graduallyreduced.



33/51


5.2.3 Speed filter

The filter time constant "t_filter" for the spee d filter is the lastparameterinthe speedcontrolparameters. It can be used to set the limit frequency with

the unit [sec] (e.g. 1kHzequals0.001sec).

Improvement to the controllerbehaviorusing the speed filter can only beachieved insystems with a high mass mome nt of inertia and encodersystems with a low resolution (e.g. resolver).Whereas ifencoder systemswith a high resolution are used, the speedfiltergenerallycannot create anyimprovements in the controllerbehavior.

Note:

You can start with a value of 0.8msec (0.0008sec). This value can bethenbeincreased in smallsteps until the controllerbehavior(lessoscil lation)has improved.Generally, the usablevalues are in the rangefrom 0.8msec to 2msec.

5.2.4 Notch filter

Note:

As we already know, the notch filteralsoworks in the speed controller.However,there is no entry for this in the axisstructure because it is notused too often.Therefore, it must be directlyconfiguredusing

parameterIDs:

Speed controllernotch filter:Frequency: ACP10PAR_FILTER_F0(ID226 )

Speed controllernotch filter:Bandwidth:ACP10PAR_FILTER_B0(ID227 )



34/51


The resonance frequency of the system can be determinedusing thefollowingsteps:

Initialize the positioncontrollerparameters with the value "0",

deactivate the spee d filter (t_filter = 0) and deactivate the notch filter(set both parameters to "0").Switch on the controllerThe proportional gain of the speed controller is increased until themechanicsoscil late.Record the actualspeed using the trace (scan rate = 200s). Switchthecontrolleroff againwhen the trace has finished.Analyze the frequencyspectrum in the trace using FFT (Fast FourierTransformation).

Set the most notablefrequencyoccurring in the trace as notch

frequency for the notch filter.Enter a minimum band width (e.g. 25Hz).Switch the controlleron again and determine a critical proportionalgain for thespeed controller.If the critical value has been increased,determine if the behaviorhasfurtherimproved due to the variation of the band width and thenotch frequency.Otherwise,determine a characteristicnaturalfrequencyagain.If furtherimprovement is no longerposs ible, the determinedvaluescan beentered in an ACOPOS parametertable for example.

Note:

The notch filter cannot be used if a distinctresonance frequencycannotbedetermined.



35/51


Fig. 24 FFT analysis.



36/51


Exercise:

Project:Name

Hardware: Test rack, flywheel mass mounted to the motorshaft.

Specifications:

The basismovement parameters to be used are alreadydefined in theNC INITparameter module.

Positioningpath "s"= 50000 Units

A stationarymanipulatedvariable issimulated in the exampleproject

usinganadditiveset torque.

Parameters for the trace:

Max. trace duration: 2 seconds

Scan rate: 0.0008 seconds

Task:

Use the flow chart (Fig. 32) to set the speed controllerin steps.

What happens in thiscase if the value "0"isentered for thepositioncontrolparameter"kv"?

Can the notch filter be used and defined?

Can thespeed filter be used and defined?

Would you define an integralaction time "tn" for the speed

controller?



37/51


5.3 Position controller

The parameters for setting the positioncontrollerare located in a subgroupof the controlparameters:

Fig. 25 AutomationStudioonline help system Controller

Note:

The followingparameters can be configured for the trace when settingthe positioncontroller:

Positioncontroller: Actual speed [Units/sec]Positioncontroller: Lag error [Units]Current controller: Set stator current of quadrature component[A]

The scan rate should be set as low as poss ible in thiscase as well.



38/51


5.3.1 Proportional gain "kv"

The value from the "kv"factorshouldalso be set as large as poss ible forthe positioncontrollerwithoutcausing the controllerto oscillate.

Note:

You can start with a value of 50 /sec and increase it gradually.

Example:

Lag error: 15 units

Value of the factor"kv": 100 /sec

Output value of the P-element: 1500 Units/sec

5.3.2 Integralaction time "tn"

The similarconditionsalso apply for this value as for the integralactiontime of thespeed controller. For some applications it issufficient to justuse one trueP-controller. An I-sectionmust be used if an axishas tocompensate for a stationarymanipulatedvariable (e.g. hanging load).Furthermore, it may be necessary to use theI-section of the positioncontrollerif the speed controllerwas only able to be set soft.

The integralaction time does not have to be set for the positioncontrollerifalreadyset for the spe ed controller.

Note:

You can alsouse a starting value of 100msec ifnecessary in thiscase.

An I-section in the positioncontrollercauses an overshoot when thetargetposition isreached and thereforeshould only be used in specialcases.



39/51


5.3.3 Total delay time "t_total"

In a single-axisapplication,thisparametershould be initialized with thesam e valueas the predictiontime.

The delay via the networkcan be compensated using the "t_total" in multi-axisapplications.

The po tential value range isalsoas large as in the parameter"t_predict".

5.3.4 Prediction time "t_predict"

Thisparameterisrequired for the feed forward. A value "t_predict= 0s"disablestheoffset.

The prediction time mak es itposs ible to compensate for the lag errorduringtheacceleration and decelerationphase to nearly"0".This

parameterisgenerallyset aftercorrect valueshave been determined for"kv" and "tn".

Note:

The po tential value range of the "t_predict"parameteris 0.0 to 0.06seconds.

Generally, a large prediction time is not really necessary if the speedcontrollercould be set hard.

Definition rule:

t_predict4 J

0.0002(kvSpeedController kt)

J = Moment of inertia on the motor[kgm]kvspeed controller= Proportional gain of the speed controller[As/Rev.]k

t

= Torque constant of the motorbeingused [Nm/A]

Note:

The positioncontrollerruns at a cycle time of 400s. This is why thevaluesfor"t_predict" and "t_total"shouldalways be selected in a way sothat they are equal to a multiple of 0.0004 seconds. Otherwise, they arerounded off by the ACOPOS servodrive.



40/51


The followingcharts display the lag error fordifferentprediction timevalues.

Correspondingspeedprofile:

Fig. 26 Correspondingspeed curve for the following lag error curves

"t_predict"set too low:

Fig. 27 Prediction time set too low

"t_predict

"set too high:

Fig. 28 Prediction time set too high

"t_predict"set correctly:

Fig. 29 Prediction time set correctly



41/51


5.3.5 Maximum proportionalaction "p_max"

The influence of the proportional gain can be limitedusing the "p_max"parameter. This can be done to prevent manipulatedvariables that are too

large.

Note:

The value for these parameters can be calculatedusing the followingformulas:

p_maxImax 2 UnitFactor

kvSpeedController

Imax = Motor peak current [A]

kvspeed controller= Proportional gain of the speed controller[As/Rev.]Unit factor= Unit scaling[units/rev.]

5.3.6 Maximum integralaction"i_max"

The maximum influence of the integralsection can be limitedusingthe"i_max"parameter. This can be done to prevent a "windup".

Note:

The value for these parameters can be calculatedusing the followingformula toachieve a requiredholdingtorque:

i_ max

M1.1

kt UnitFactorkvSpeedController

M = Requiredholdingtorque [Nm]kt = Torque constant of the motorbeingused [Nm/A]kvspeed controller= Proportional gain of the speed controller[As/Rev.]

Unit factor= Unit scaling[units/rev.]



42/51


Exercise:

Sameproject and hardware asbefore.

Specifications:



A stationarymanipulatedvariable issimulated in the exampleprojectusinganadditiveset torque.




Task:

Use the flow chart (Fig. 32) to set the positioncontrollerin increments.This setting will be based on the valuesdetermined for the speed

controllerin the previousexercise.What happens if you are not using an integralaction time "tn" forthe positioncontroller?



43/51


5.4 Limit parameter

When the controlparameters are optimized, two parameters must still beset for the limit values.

5.4.1 Jolt filter time "t_jolt"

The value for the parametercan be determined by recording the lag errorsduringapositioningmo vement without jolt filter time. At the end of thismo vement, it willbecomeevident that the system must first "settledown".Afterbeingdetermined from thetrace, the settling time (time until theoscil lations level out) can now be used as jolt filtertime "t_jolt".

Fig. 30 Determining the jolt time



44/51


5.4.2 Lag error stop limit "ds_stop "

Fig. 31 AutomationStudioonline help system Limit values

A warning isoutput if the current lag error value set for the "ds_warning"parameterisexceeded. An emergency stop isexecuted if the value of the"ds_stop"parameterisalsoexcee ded.

Note:

A controllede-stop ramp isgenerated when an e-stop occurs. The setspe ed isdecelerated to "0"using the current initialized limit values.The

controlleristhenswitchedoff.

Note:

The value for "ds_stop" can be determinedusing the followingcalculation:

kvSpeedController

Imax

kvPositionController2 UnitFactor

Imax = Motor peak current [A]kvspeed controller= Proportional gain of the speed controller[As/rev.]kvpositioncontroller= Proportional gain of the positioncontroller[1/sec]Unit factor= Unit scaling[units/rev.]



45/51


Task:

Sameproject and hardware asbefore.

Specifications:



A stationarymanipulatedvariable issimulated in the exampleprojectusinganadditiveset torque.




Task:

Set the limit value correctly and checkthe result.



46/51


5.5 Overview

Fig. 32 Overview of the process for se tting the controlparameters



47/51

Sav ing the Controller Settings

6. SAVINGTHE CONTROLLER SETTINGS

The data justdeterminedmust be saved to the controllerso that it isavailable to the ACOPOS servo drive each time the machine isstarted. All

parameters available in theaxisstructure can be saved in an NC INITparameter module. The remainingvaluescan be entered in a parametertable.

6.1 NC INITparametermodule

The followingoptions are available for saving the values in the NC INITparameter module.

Entering the valuesdirectly in the correspondingmodule.

Fig. 33 NC INITparametermodule

Saving the determinedvalues imm ediately in the NC test windo w.

Once everythinghas been saved, the projectshould be compiled and thecurrentversionshould be transferred to the controller.

6.2 ACOPOS parametertable

The parameters and values (e.g. for the notch filter) can be entered in anACOPOS parametertable.

Fig. 34 ACOPOS parameter tab le



48/51

Summary

7. SUMMARY

By understanding how the parameters that we will be using work, we arenow able todeterminecorrect values.

The cascaded controlconcept allowsus to optimize the controlparametersand filterparameters step-by-step.

The parameters determined are usuallydifferent.However, the procedurefor obtainingthese valuesisalways the same.



49/51

2

Appendix

8. APPENDIX

kt

= __________Nm/A

J = __________kgm

Switchingfrequency= __________Hz

Speed controller:

2 0.000075

1 1 0.000075

2 SwitchingFrequency 2

Tv

TI

Td e a d

v

Tfilter 0.000175

kv JT

2 2_v kt

tn 4 T_v 4

Positioncontroller:

T_p Tint

erpol

4 Tv

Td ea d p

0.0001 4 0.0002

kv1 1

2 T_p 2

tn 4 T_p 4


50/51



51/51

Documents

Tm450 Acopos Control Concept and Adjustment