Blanco TCS Upgrade

unkindnesskindΠολεοδομικά Έργα

15 Νοε 2013 (πριν από 3 χρόνια και 8 μήνες)

78 εμφανίσεις

Blanco TCS Upgrade

Final Configuration,


and Integration Test Results


Nov
-
2011Thru Feb
-
2012


M.Warner

E.Mondaca

R.Cantarutti

G.Schumacher

References

1
-

Motor and Tachometer Vendor Specifications:


ftp://ftp.ctio.noao.edu/pub/warner/blanco/Blanco_motor&tach.pdf

2
-

Current Driver Data Sheet:



ftp://ftp.ctio.noao.edu/pub/warner/blanco/az40a8.pdf

3


Original 4m Telescope Servo Analysis:



ftp://ftp.ctio.noao.edu/pub/warner/blanco/blanco_4m_analysis.pdf

4


TCS CDR (Telescope Lumped Mass Model):



ftp://ftp.ctio.noao.edu/pub/warner/blanco/tcs_cdr_warner.pdf

5
-

Blanco TCS Upgrade Project Report


ftp://ftp.ctio.noao.edu/pub/warner/blanco/B4Upgrade.pptx

6


TCS Upgrade Test Results


ftp://ftp.ctio.noao.edu/pub/warner/blanco/TCSUpgrade_test_results.pdf

2

General Telescope and Drive Specifications


The Dynamic Telescope requirements for both axes,
needed to meet DECAM mission (2deg in 17sec):


Maximum Jerk = 0.05[deg/s^3]


Maximum Acceleration = 0.05[deg/s^2]


Maximum Velocity = 0.5[deg/s]

Note: This limits where used for all test trajectories presented in
this repot.



Tracking Requirements:


Jitter : 0.1”rms maximum on both axes


Settling Time: 5sec


Error band: 0.1”, 1 Second Mean


Drift: 1” per min




3

Final System Configuration


Hardware Overview


Manual Console Signals


Main Driver Chassis



Telemetry


cRIO
-
V and
cRIO
-
P Hardware


Safety Interlocks


Web
-
browser based Engineering Panels


TSC Kernel


Pointing Kernel Model


Pointing Tests Results


Trajectory Generation using algorithms developed for
robotic control.



4

M1

HA Tape

Encoder

HA Absolute

Encoder

Telescope Mount

Kernel

PC

M2

Tach1


Tach2

HA (West)

Current Driver

M3

DEC Tape

Encoder

DEC Absolute

Encoder.

M4

Tach3


Tach4

cRIO
-
V

HA (East)

Current Driver

DEC (North)

Current Driver

DEC (South)

Current Driver

HA Inc.

Encoder

HA Inc.

Encoder

cRIO
-
P

Torpedo

(1KHz)

System Overview

Ethernet


(20 Hz)

Main Chassis

Computer Rack

5

OLD CONSOLE

PLC

cRIO

-
V

Guide

Set

Search

West

East

North

South

Hand
Paddle

CONSOLE SIGNALS

Current
Monitor

Manual Slew
Control

6

Main Chassis

Power Supply
(2)

Braker

and

Contactors

Current
Amplifiers (4)

Filters (4)

Fuses (8)

7

cRIO
-
V Telemetry Configuration

Power Supply

(2)

Current Amplifier

(4)

cRIO
-
V

Brakes Power Supply Box

Current Transducers

(4)

Power

Supply

Enable

Logic

Fuses

Main Chassis

To Brakes

Power Supply Voltage Telemetry (2)

Current Amplifier Status (4)

Current Amplifier Voltage Output (4)

Current Amplifier Current Output (4)

Brakes

Voltage

To Motors

8

Hardware Interlocks

Power Supplies

(2)

Current Amplifiers

(4)

Main POWER

Interrupt

Main Brakes

Panic

Stop

PLC

Enable

Software

Enable

25[A] Fuses

ON/OFF Buttons

To Motors

Main Power Supply

(3 phases, 480V)

9

cRIO
-
V Box: Front Panel

10

cRIO
-
V Box: Rear Panel

11

cRIO
-
P Box: During Assembly

12

cRIO
-
V Safety Interlock System

13

cRIO
-
V Engineering Web Browser Based Display

Real time graphic

display of all motor current,


tachometer, and demand

signals.


Facilitate telescope balance

adjustments, by monitoring

motor current graph


Status of all safety interlocks

14

cRIO
-
P Engineering Web Browser Based Display

Real Time graphic

display of all positions and

demand signals, including

position error RMS value


Status of all safety interlocks

15

TCS Kernel


TCS Kernel main function is to generate the
demands to the mount Servo system.


Demands are generated using software based on
a pointing model.


The pointing model parameterize real behavior of
the telescope structure.


The demands are computed at 20Hz and fed to a
trajectory generator, before passing them to the
Servo.


The Kernel transformation functions are also
utilized to compute the guider (X,Y) position to
acquire a guide star.

16

Pointing Kernel Concepts

(P.T. Wallace 2002)

Roll index error

Pitch index error

Vertical deflection

OTA/pitch
nonperp

Roll/pitch
nonperp

Roll misalignment W

Roll misalignment N

IA

IB

VD

CA

NP

AW

AN

POINTING

MODEL

Target

position

Pointing

origin

Mount

axes

Pointing

origin

Target

position

Mount

axes

GENERIC MODEL

TARGET [
α
,
δ
]

AIM [
x
a
, y
a
, z
a
]

BORESIGHT [
x
b
, y
b,
z
b
]

TELESCOPE [
x
t
, y
t
, z
t
]

[1, 0, 0]

Equatorial Telescope:

Roll

-
hour angle

Pitch ≡ declination

Light deflection, aberration

Precession
-
nutation

Earth rotation

[h,
δ
]

Refraction

Mount orientation

roll

Roll/pitch non
-
perpendicularity

pitch

guiding

[
ξ
,
η
]

Vertical deflection

OTA/pitch non
-
perpendicularity

Site location, UT1

Weather, color

AW, AN

IA

NP

IB

GA, GB

Rotator angle

VD

CA

X, Y

B

A

Pointing origin

Offsets

-
ha

dec

17

POINTING TESTS RESULTS


IH +669.0271

Hour Angle Index


ID
-
153.5064

Declination Index


NP
-
57.1753

Non
-
Perpendicularity between HA and DEC


CH +60.1232

Collimation: non
-
parallel between optical and mechanical axis


ME +167.3562

Polar axis error in elevation


MA +25.9390

Polar axis error in azimuth


HXSH
-
105.4795 Horseshoe flexure east
-
west (dynamic non
-
perpendicularity)


HHSH2 +62.5299 Horseshoe flexure HA


HDCH4
-
22.8837 Horseshoe flexure DEC


HDCH3 +51.5131 Declination gearbox

A pointing test was performed utilizing the Mosaic instrument, in order to

d
erive a new model utilizing the tape encoders. Based on that data the new model is:


Note the large value of the ME coefficient. Historically this value was in the

r
ange of 25
-
30. It means that the mount did sag sometime after 2001, when

t
he previous pointing test was done.


The new model was tested with the same instrument, by setting a star in

t
he guider box and observing the drift. None was observed for about 10

m
inutes, an indication that on one side, the model produces the correct

m
otion demand and on the other, the new servo control handles the

m
ount correctly.

18


DEC drift eliminated, with Pointing Model
Correction.


DEC Tracking Error=0.013”rms

Pointing Model

Correction

19

Kernel Dataflow

Demanded Guider Position

Environment Info (t, p, h)

Focal Plane Configuration

Mirror Position

Guider tracking
errors

Offset Requests

Demanded Mount Position

Pointing Coefficients

Mount

Demand

Guider

Demand


Absolute Time


20 Hz

Astrometry.

Pointing
Model.

Demand
Computation.

Internal
State.

KERNEL COMPONENT

TRAJECTORY

GENERATOR

cRIO
-
P

POSITION

VELOCITY

TCS

Jerk/
Accel
/Velocity


Limited Trajectory

TCS

20

Optimum Jerk/Acceleration/Velocity Limited
Trajectory Profile

This method generates the smoothest possible motion within the
dynamic constraints imposed, allowing a seamless transition during
track
-
slew
-
track motions.

21

Optimum Jerk/Acceleration/Velocity Limited


Trajectory Generation

(from Reflexes Type IV Motion Library)

FROM

cRIO
-
P

FROM

KERNEL

TO

cRIO
-
P

0.5 0 deg/s

0.05 deg/


0.05 deg/


22

HA Trajectory: Track+2 Deg
Slew+Track

Trajectory generation for

2 degrees + track to track


motion
:

Max
Vel
: 0.5 deg/sec

Max Acc: 0.05
deg/sec²

Max Jerk:
0.05 deg/sec³

Reflexxes

Type IV Motion Library

Torsten

Kröger

23

Track

Slew

Track

Test Results Overview


Servo Models


Matlab

Model and
LabView

Implementation


Open Loop Bode Plots


Velocity Loop Test


Velocity Loop Servo Model


Slew Trajectory and Model Verification


Slew Velocity Loop Step Response and Closed Loop Plots, Model Verification


Friction Loop Plots


Position Loop Tests.


Telescope Models for
Tach
-
>Tape Transfer Function


Position Loop Test Results for Slew Trajectory


Tracking jitter performance on the sky


Final Integration Tests.


Full Range Friction Plots, for establishing a baseline before
DECam

Integration.


Parametric study on Settling time v/s Trajectory Limit

24

Servo Models


Servo Models are based on Lumped Mass
model described on Ref. 3, and coded into
Matlab
.


DEC Velocity Slew Compensation is based on
original analog compensation modified to have a
larger DC gain.


HA Velocity Slew Compensation is based on
modified original analog compensation.


Tachometer to Tape Transfer Function was
derived from measured sine wave sweep data.


25

HA Servo Model

(Slew Baseline Model)

Velocity Loop (
cRIO
-
V)

Position Loop (
cRIO
-
P)


s + 0.5

HPCOMP1 =
-------


s


66.6 s^2 + 682.6 s + 16650

HVCOMP1 =
--------------------------------


s^2 + 9.3 s + 8.3

Tachometer

Tape Encoder

26

DEC Servo Model

(Slew Baseline Model)

Velocity Loop (
cRIO
-
V)

Position Loop (
cRIO
-
P)


s + 0.5

HPCOMP2=
-------


s


60.37 s^2 + 9170 s + 84070

HVCOMP2 =
-----------------------------------


s^2 + 53.12 s + 52.12

Tachometer

Tape Encoder

27

Position Loop Implementation in
LabView

code

running at 1Khz loop cycle

Servo Compensation Filters

cRIO
-
V

Command

Tape

Encoder

Inputs

Kernel

Command

HA

DEC

28

Velocity Loop Implementation in
LabView

code

running at 1Khz loop cycle

Tach

Inputs

cRIO
-
P

Command

Motor

Amplifier

Current

Command

Servo Compensation Filters

HA

DEC

29

Velocity Open Loop Bode Plot

30

Position Open Loop Bode Plot

(Combines Lumped Mass and Tape Measurements)

31

Velocity Loop Tests


Velocity Loop Test


Baseline 2 Deg Slew Trajectory performance was
compared against Model


Slew Velocity Loop Step Response and Closed Loop
Plots, Model Verification.



32

HA 2 Deg Baseline Trajectory Test (Position and
Velocity)

33

HA 2 Deg Baseline Trajectory Test (Velocity Error)

34

DEC 2 Deg Baseline Trajectory Test (Position and
Velocity)

Tachometer Jitter

35

DEC 2 Deg Baseline Trajectory Test (Velocity Error)

Tachometer Jitter

36

HA Slew Velocity Loop Step Response

37

HA Slew Velocity Closed Loop Bode Plot

38

DEC Slew Velocity Loop Step Response

39

DEC Slew Velocity Closed Loop Bode Plot

40

Position Loop Tests


Position Loop Test


Telescope Models for
Tach
-
>Tape Transfer Function
based on actual measurements


Baseline Slew Trajectory performance was compared
against Model



HA Tracking Jitter Measurements


41

HA Tachometer to Tape Transfer Function

used on Position Loop Models

42

DEC Tachometer to Tape Transfer Function

used on Position Loop Models

43

HA 2 Deg Position Loop Trajectory Test

(Comparison to Servo Model)

44

DEC 2 Deg Position Loop Trajectory Test

(Comparison to Servo Model)

45

HA Track
-
2.25 deg Slew
-
Track Test

using Kernel Generated Trajectory

HA Position

HA Position Error

46

HA Track
-

2.25deg Slew
-
Track Test

using Kernel Generated Trajectory (Detail)

+/
-

0.2”

25sec

+/
-

0.4”

23sec

HA Position Error

47

DEC 2 deg Slew Test

using Kernel Generated Trajectory

DEC Position

DEC Position Error

48

DEC 2degTrack
-
Slew
-
Track Test

using Kernel Generated Trajectory (Detail)

Final Baseline Compensation

+/
-

0.2”

24sec

DEC Position Error

49

Long Term Track Jitter

50

HA Sidereal Tracking Power Spectral Density

5 min Sidereal Tracking Test, using

baseline HA Notch Track Filter #3

Total Tracking Error = 0.067”
rms

Tape Periodic Error = 0.02”
rms

51

Final Integration Tests.


Full Range Friction Plots, for establishing a baseline
before
DECam

Integration.


Friction Plots consists of measuring the torque needed to move
the telescope across it’s full range of motion, at a constant
speed ( 0.5 deg/sec).


Is a good diagnostic for friction, and any dynamic interferences
present on the mount.


Parametric study on Settling Times v/s Trajectory Limit


Different jerk, and acceleration limits where used for a 2deg
track
-
slew
-
track motion, settling time was defined as a track error
< 0.2”.


52

HA Friction Final Baseline

53

DEC Friction Final Baseline

54

HA Parametric Study

Trajectory Limits v/s Settling times

These tests confirms that baseline values are near optimal, maximum

acceleration is limited by the maximum torque (current) available to

accelerate the HA axis inertia.

55

DEC Parametric Study

Trajectory Limits v/s Settling times

These tests confirms that improvements can be obtained with a

more aggressive trajectory. As the DEC axis inertia is smaller,
some margin exists in the maximum torque (current) available.


Note: These tests where performed with original Velocity
Compensation.

56

Conclusions


The Velocity and Position Loop tests validated the
historical lumped mass model used.


The use of a novel dynamic limited trajectory
generator, provides a framework for safe, and
efficient use, of the telescope control system
resources.


The new system provides all the telemetry needed
to characterize the telescope under the new loads
imposed by
DECam
, and allows further
optimization if needed.


57