LHC Collimation low level equipment

1
LHC Collimation low level equipment

• R. Losito - AB/ATB
• LHC Collimation Motorization review
• 4/11/2005

2
OUTLINE

• This presentation is the result of the work of
  many people, in particular
• Arnaud Brielmann
• Fabrice Decorvet
• Pierre Gander
• Jerome Lendaro
• Alessandro Masi

3
OUTLINE

• Introduction
• Why Stepping Motors?
• Stepping motors characteristics and test
• Why LVDTs and resolvers?
• Calibration of LVDT and resolvers
• Motor Drivers functional specs
• EMC, are we going to be really a problem?
• Conclusions

4
Introduction

• Up to 108 collimatorsincluding tertiary,
  scrapers, absorbers, phase 2, etc
• Distributed over 6 points with a large
  concentration in point 3 and 7

• Jaw positions are correlatedprimary secondary
  tertiary
• Also during movements!they have to stay in sync

5
Introduction
Side view at one end
Vacuum tank
Movement for spare surface mechanism (1 motor,
2 switches, 1 LVDT)
CFC
CFC
Temperature sensors
Microphone
Reference
Reference
Motor
Motor
Sliding table
Gap opening (LVDT)
Resolver
Resolver
Gap position (LVDT)
switches for IN, OUT, ANTI-COLLISION
6
Why stepping motors

• The jaws have to be positioned with an accuracy
  which is a fraction (1/10) of the beam size
  (200µm!!).
• Stepping motors, by construction, translate
  switched excitation changes into precisely
  (0.1) defined increments of rotor position
  (steps).
• The most relevant parameter is not the absolute
  accuracy (precision with respect to the
  theoretical nominal orbit), but the repeatability
  of the position (precision with respect to the
  position of the beam measured during
  calibration!!)

7
Why stepping motors

• The jaws, in fact, will be initially positioned
  using a beam based alignment (measure of beam
  loss vs. jaws position).
• Then, during ramping, squeezing etc the jaws
  will be moved according to predefined functions,
  and assume that the beam will always behave in
  the same way.
• Interlocks shall be raised by beam loss survey
  (the beam does not behave correctly), and by jaw
  position survey (the jaw movement does not behave
  correctly).

8
Why stepping motors

• The repeatability of the movement is ensured by
  the accurate positioning achievable by stepping
  motors (provided they do not loose steps).
• Therefore, strictly speaking, a position feedback
  is not required by the system functionality
• Position survey is needed for interlocking
  (somehow releasing the specs on position
  measurement accuracy, and making them more
  realistic in the harsh environment of the LHC)

9
Preliminary test of repeatability

• This table shows the results of measurement of
  end positions over 200 tries for each case.
• More accurate measurements will be repeated in
  next weeks to understand error in absolute
  position
• Repeatability is in the order of the accuracy of
  the metrology palmer.

10
Why stepping motors

• The stepping motor has therefore to be chosen
  such that
• It has sufficient margin on nominal torque to
  give enough confidence that it will not loose
  steps.
• It has sufficient angle resolution to provide the
  required linear resolution (5 µm).
• Most constraining requirement
• It must be Radiation hard (23 Mgray/year)!!!

11
Stepping motors characteristics and test

• We choose
• 2-phase STEPPING motors, 200 step/revolution
• Angle resolution 1.8 0.1
• 2mm pitch leadscrew
• Torque sufficiently higher than MAX load torque
  (1.2 Nm x 3)
• And we have
• 2mm/200 step (10 0.5) µm !!!!
• Static Error due to load negligible SINCE
• pull-out torque gtgt load torque
• With a Ministep Driver we can do 400, 800, 1600,
  3200 steps/revolution (5, 2.5, 1.25, 0.75 µm)

ministepping
12
Stepping motors characteristics and test

• Radiation hardness is the most constraining
  parameter, since no company (in Europe) provides
  Rad-hard stepping motors as an off-the-shelf
  product
• We contacted SCK-CEN, involved in the programme
  aimed at establishing the radiation tolerance of
  remote handling components for the maintenance of
  ITER.
• They found a recipe to build motors that survived
  an irradiation test of 100 Mgray without any
  change in the characteristics (insulation,
  current, torque).
• This recipe is unfortunately covered by a
  confidentiality agreement inside ITER, but we got
  addresses of companies that were involved in the
  construction of this motor and different tips,
  included in our specs

13
Stepping motors characteristics and test

• For radiation hardness we required
• Materials compatible with high radiation levels
  (30 Mgray). Firms were requested to provide
  either the list of materials or a certificate
  that their products survived irradiation test
• Dry lubrication (graphite incrustation or
  similar).
• Special bearings.

14
Stepping motors characteristics and test

• Main requirements

15
Stepping motors characteristics and test

• Main requirements (2)

16
Stepping motors characteristics and test

• Test specified in the contract
• The contractor shall perform on each motor the
  following tests
• Geometrical dimensions (ensure mechanical
  tolerances are respected)
• Insulation resistance phase-to-phase and phase to
  shielding
• DC Resistance, AC impedance at 1 kHz.
• Detent torque
• Holding torque
• Pull-out torque curve
• Pull-in torque curve

17
Stepping motors characteristics and test

• Definitions and test procedures are based on the
  standard IEC 60034 - 20 1
  Rotating electrical machines Part 20-1
  Control motors Stepping motors
• detent torque
• maximum steady torque that can be applied to the
  shaft of an unenergized permanent magnet or
  hybrid stepping motor without causing continuous
  rotation

18
Stepping motors characteristics and test

• Definitions and test procedures are based on the
  standard IEC 60034 - 20 1
  Rotating electrical machines Part 20-1
  Control motors Stepping motors
• holding torque
• maximum steady torque that can be applied to the
  shaft of a stepping motor energized by a specific
  current without causing continuous rotation

19
Stepping motors characteristics and test

• Definitions and test procedures are based on the
  standard IEC 60034 - 20 1
  Rotating electrical machines Part 20-1
  Control motors Stepping motors
• pull-out torque
• maximum torque that can be applied to the
  rotating shaft of a stepping motor driven at a
  given pulse rate under specified drive
  conditions, without causing the motor to miss
  steps

20
Stepping motors characteristics and test

• Definitions and test procedures are based on the
  standard IEC 60034 - 20 1
  Rotating electrical machines Part 20-1
  Control motors Stepping motors
• pull-in torque (the standard does not give a
  definition, but provides a procedure to measure
  it!!!)
• maximum torque load at which a stepping motor
  (driven at a given pulse rate under specified
  drive conditions) from a stall (rest) position
  can start the rotation without missing steps

21
Stepping motors characteristics and test
22
Stepping motors characteristics and test

• What is the reality?
• We built a stepper motor test bench (we can
  reasonably test 40 motors/week)

Torquemeter
Stepping Motor under test
Motor AC Brushless
Encoder
23
Test Bench Calibration
The residual torque over a complete turn has to
be subtracted to the stepping motor detent torque
measurement

• Torquemeter Technical Characteristics
• Non linearity Error Hysteresis lt1
• Sensor Repeatability lt0.1

• Improvements applied
• Use of sensor calibration curve (10 points)
• 100 oversampling factor in the torque acquisition

Performances reached

• repeatability less than 1 mNm
• Sensitivity 5 mNm
• non-linearity In the range 0-200 mNm the error
  is lowered to less than 0.1
• Overall accuracy 5 mNm on the detent torque

24
LEP Collimator Motor n. 1 -Detent Torque
The max detent torque is 30 mNm
Two consecutive magnetic poles equispaced of 7.2
degree (the rotor contains 50 magnetic poles)
25
LEP Collimator Motor n.2 -Detent Torque

• Acquisition carried out over a complete
  revolution
• Test speed 0.1 rpm
• Sampling frequency torquemeter 5 KS/s
• Acquisition decimation factor100

For this motor the max detent torque is 60 mNm
26
LEP Collimator motor Pull-in curve at nominal
current
The Motor is powered at the same current value
used for the test at TT40 1.4 A (the motor
nominal current is 1.2 A)
At the collimator nominal speed (400 step/s) the
dynamic torque is less than 0.8 Nm
27
New motors for the LHC Collimator prototype
Detent Torque
The detent torque is in agreement with the motor
technical data (detent torque 80 mNm)
Hard point, maybe due to misalignment of the
motor axis
Max Detent Torque75 mNm
28
New motors for the LHC Collimator prototype
Pull-in curve
Motor phases current 2 A
At 400 step/s the dynamic torque is 1.6 Nm
29
New motors for the LHC Collimator prototype
Pull-in curve
At 400 step/s the dynamic torque is 1.7 Nm
Resonance
30
Conclusions on Motor characteristics measurements

• We have full control of the motor
  characteristics, each motor will be measured
  before installation
• A huge difference has been found between rated
  Torque and measured Torque. After discussions
  with several producers, this does not seem
  unusual.

31
Conclusions on Motor characteristics and test

• The best bid comes from a company which
  participated in the ITER program (motors at 100
  MGray)
• They will subcontract the manufacture of the
  motor
• They made three proposals

32
Conclusions on Motor characteristics and test

• The third option seems the most convincing (no
  special redesign, just change the diameter of the
  rotor)
• We asked them to build asap (before FC) 2
  prototyopes (not Rad-Hard) to check that they can
  reach these performance. We should get them in
  two weeks time
• We must take into account, however, a spread 20
  on nominal torque (the margin is never enough!!!)

33
Why LVDTs and resolvers?

• Possible position sensors

• Analog
• Linear
• Resistive
• Capacitive
• Inductive
• Hall Effect
• Magnetoresistive
• Magnetostrictive
• LVDT
• Rotative
• Resolver
• RVDT

• Digital
• Encoders, both linear and rotative.
• Magnetic
• Optical
• Contact

34
Why LVDTs and resolvers?
Position Sensors pros and cons

• Analog
• Signal is continuously changing with position
  ? infinite resolution
• Ouput Signal prone to significant reduction of
  SNR for transmission over long distances
• Careful filtering and signal processing has to be
  performed (expensive electronics)

• Digital
• Signal is in bit, resolution limited by
  quantization error (½ bit)
• Excellent performance with respect to long
  distance transmission.
• In static condition, the electronics only need to
  detect 0 and 1, BER negligible.
• No special effort on conditioning electronics,
  (cheaper)

35
Why LVDTs and resolvers?

• Analog
• Classification based on characteristics found on
  web for several producers (Accuracy lt 50mm).
• Linear
• Resistive Not Accurate
• Capacitive Accurate, single sided
• Inductive Accurate, single sided
• Hall Effect Typically semiconductor, Rad hard?
• Magnetoresistive Not Accurate
• Magnetostrictive Accurate (radar, but close
  electronics)? not Rad hard
• LVDT Accurate, double sided (ZERO!!!)
• Rotative
• Resolver Accurate, absolute measurement on
  one revolution
• RVDT Accurate, lt 360

36
Position Sensors

• Analog
• LVDT and Resolvers
• Absolute position reading
• Radiation hardness They are both made with the
  same technology of a motor
• If the motor survives, the LVDT and the Resolvers
  can survive.
• Lifetime infinite since contactless (no
  mechanical stresses)
• Resolution is determined by ADC in conditioning
  electronics (and NOISE!!!).
• 16 bits on 40 mm 1.2µm resolution

37
Position Sensors

• LVDT

• Figure from MSI technical bulletin n. 0103

38
Position Sensors

• LVDT

• Ratiometric conditioning allows to filter out the
  noise, to be independent from temperature and
  excitation drifts. (at the first order)

39
Position Sensors

• LVDT
• LVDT is used in position control applications for
  homing purpose (its zero is used as reference
  for positioning).
• Repeatability is intrinsically infinite, only
  depends on mounting (on us!!!). Commercial
  (expensive) models (non Rad Hard) are guaranteed
  for a repeatability of 0.15µm
• Linearity depends on uniformity of material and
  of windings.
• NON LINEARITIES DO NOT AFFECT ZERO and
  REPEATABILITY.

40
Position Sensors

• LVDT
• Measured errors by one of the market leaders
• Zero Drift 5 µm/ 30 mm.
• After one year it tends to reach a stable
  position. It can probably be reduced by applying
  ageing thermal cycles.

41
Position Sensors

• Standard LVDT Front-end electronics

42
Position Sensors

• LVDT
• Measured errors by one of the market leaders
• Zero Drift 5 µm/ 30 mm.
• After one year it tends to reach a stable
  position. It can probably be reduced by applying
  aging thermal cycles.

43
Position Sensors

• LVDT
• Adding a compensation to thermal drift of
  electronics
• (521223212232)½ 18 µm uncertainity at zero
• 51 µm ACCURACY at full scale ON LONG TERM

44
Position Sensors

• LVDT
• Error chart

BEAM AXIS
0
-L
L
45
Position Sensors

• Resolver

• Figure from catalog of TYCO resolvers

46
Position Sensors

• Resolver

47
Position Sensors

• Resolver
• The resolver itself gives down to 4 arc minute
  (0.4 µm on 1 revolution)
• Signal conditioning electronics similar to LVDT
  by using the same worst case ppm error we get
• Over 1 revolution 30 arc minutes ?
  3 µm ACCURACY
• We have to add something for SNR degradation over
  1 km and for mechanical plays.

48
Positioning Strategy

• Hypothesis
• We can make an accurate calibration before
  installation
• We have enough time to order and receive the
  sensors before installation.
• Mechanical plays are negligible (we measured on
  the prototype a play of 20 mm)
• Yearly calibration in the tunnel does not require
  human intervention in the tunnel to mount
  calibrated reference position sensors (no dose to
  personnel for the intervention)

49
Positioning Strategy

• We use Resolvers for absolute angle measurement
  (1 resolver revolution1 motor revolution)
• We use LVDTs for homing and back-up on jaw
  position and for direct measurement of gap
  opening
• We can get (in the best case lab measurements)
• 20 µm accuracy on jaw position
• Since real life includes additional drifts,
  Noise, fatigue etc, a factor of at least 2 has
  to be considered for the accuracy

50
Positioning Strategy

• Calibration of the LVDTs (and resolvers) cannot
  be correctly performed in the tunnel due to high
  level of radiation, it is impossible to foresee
  the mounting of a reference position sensor
  (metrology palmer).
• This will be done once before installation. A
  database of calibrations will be established

51
Positioning Strategy

• Yearly check of the calibration of LVDTs will be
  performed
• Check the electrical zero
• Go to the mechanical stops (with the motors)
  reducing the torque of the motor to a minimum
  value not to damage the stops (the motor driver
  must be fully programmable)
• Check the reading with respect to the database
• Calibrate intermediate points by advancing with a
  given number of motor steps and looking at he
  resolver and LVDT readings.
• Check the coherence within all results of the
  previous points.

52
Positioning Strategy

• Yearly check of the calibration of LVDTs will be
  performed
• If a large incoherence (say, gt100 µm) is found,
  then set an alarm and, eventually exchange the
  collimator with a spare.
• All that will go through a learning process to
  establish the details the procedures and the
  limits.

53
Conclusions on positioning strategy

• Having a reasonable ratio (gt3) between motor
  pull-out torque and load (jaw) torque should
  ensure the motors do not loose steps and a
  negligible position error due to stepping motors
• Resolvers will check that the motor does not
  loose steps
• Position sensors will be used to check (not to
  steer) correct behavior of the motors.
• Position measured with Analog sensors (LVDT,
  Resolvers) can give (hopefully) 50 µm
  accuracy
• Repeatability is expected to be even better

54
Drives functional specs

• Ministep option (at least 800 steps/rev)
• They have to be fully programmable we will have
  to modify remotely the number of steps and the
  nominal current (to go gently on the mechanical
  stop and to verify the margin of safety)
• There are on the market drivers that can provide
  a real-time protection against motor stall
  (loosing steps) and resonances. (features called
  Encoderless stall detection and Active
  Damping)
• These are proprietary algorithms based on
  realtime measurement of voltage and current.
• Will this be compatible with EM Compatibility?

55
Motivation of EMC studies
56
Where are we in LHC?

• 30 Collimators spread over 700 m

57
Where are we in LHC?

• 88 Collimators spread over 410 m

58
Problems we might create

• Stepping motors work with pulses!!!

• Switching amplitude
• 2 to 5 Amps

59
Problems we might create

• Even worst, stepping motors work with chopping
  drivers!!!

60
Problems we might create

• Noise measured with antenna specified in
  IEC-60478-5

Large scale experience with industrial stepping
motor controllers and resolver read-out systems
at SPS and LEP Colchester, R J Gras, J J
Jung, R Koopman, J Vouillot, J M Feres, J
López, BCERN-SL-95-123 BI
61
Problems we might create

• Even worst, stepping motors work with chopping
  drivers!!!

Large scale experience with industrial stepping
motor controllers and resolver read-out systems
at SPS and LEP Colchester, R J Gras, J J
Jung, R Koopman, J Vouillot, J M Feres, J
López, BCERN-SL-95-123 BI
62
Problems we might create

• Filtering did not work for LEP (but we will try
  again)

Large scale experience with industrial stepping
motor controllers and resolver read-out systems
at SPS and LEP Colchester, R J Gras, J J
Jung, R Koopman, J Vouillot, J M Feres, J
López, BCERN-SL-95-123 BI
63
Problems we might create

• true DC works

Large scale experience with industrial stepping
motor controllers and resolver read-out systems
at SPS and LEP Colchester, R J Gras, J J
Jung, R Koopman, J Vouillot, J M Feres, J
López, BCERN-SL-95-123 BI
64
Problems we might create

• CNGS T40 Target Station
• The same cable is used (over few meters) for

DC motor power (PWM) Strongly perturbed
sensor signal
60V, f 8kHz
Sensor signal -8V to 8V DC Perturbation 15V
65
Problems we might create

• In the tunnel motors cables are as much as
  possible separated from signal and control cables

66
EMC conclusions

• We will test some drivers motors with BDI
  equipment (maybe in point 3 where we have already
  cbles?)
• According to the result of the test we will
  decide what is the limit to specify
• We will try ourselves to fulfill this
  specification with standard drivers and active
  filters while sending out an official Invitation
  to Tender

67
Conclusions

• Experimental evidence that all specified
  movements can be achieved with the specified
  stepping motors, for each device and orientation
  involved. In particular, it must be demonstrated
  that motors can be operated precisely, synchronic
  and reproducibly (without losing steps) while
  respecting the auto-retraction functionality.
• Answer
• Motors cannot loose steps if their static and
  dynamic torques are higher than the load torque.
  We have specified a factor of 3 on the worst case
  (in most cases the margin is at least 5).
• Synchronicity within the same collimator is
  reached through the use of 4-way controllers,
  that ensure msecond level synchronization.
• Synchronicity among different collimators can be
  ensured within 9 msec (Siemens) or less by
  going to real-time architectures (compatibility
  with collimator infrastracture to be checked).

68
Conclusions

• Experimental evidence that all specified
  movements can be achieved with the specified
  stepping motors, for each device and orientation
  involved. In particular, it must be demonstrated
  that motors can be operated precisely, synchronic
  and reproducibly (without losing steps) while
  respecting the auto-retraction functionality.
• Answer
• Autoretraction should work with the specified
  detent torque.
• All the motor characteristics can be measured at
  CERN, to ensure that the motors comply with the
  specs

69
Conclusions

• If experimental evidence cannot be provided for
  all devices or all specifications, then provide
  engineering calculations which demonstrate the
  expected full compatibility of the specified
  stepping motors with the LHC requirements. Each
  de-vice and orientation should be commented on
  separately.
• See Alessandros and Olivers talks

70
Conclusions

• Comments on required margins, taking into account
  high radiation environment, aging, wear, thermal
  cycles (bake-out), etc.
• Is a factor of 3 enough?

71
Conclusions

• Supporting arguments for using one general type
  of stepping motor instead of several specialized
  stepping motors.
• Why not?
• There are evident cost savings in choosing only
  one motor (and drive)
• For example, engineering cost on one type of
  motors has been quoted 150KCHF

72
Conclusions

• Specifications for uncontrolled mechanical
  movements with the specified stepping motors.
• Uncontrolled movement (excluding power cuts,
  drive failure) can only happen with wrong setting
  from the operator or the supervisory system
  protections should be set into the communication
  protocol.
• Uncontrolled movement can be due to a motor
  loosing steps. To reduce this risk we will take a
  sufficient margin (at least 3) on load torque.
  We will check every shutdown the margin using a
  programmable drive.
• Autoretraction takes 1.5 seconds on 30 mm

73
Conclusions

• Estimation of mean time between failure of the
  package mechanics plus motor, for example a
  lifetime test on the third collimator prototype
  (for example 10,000 cycles of jaw in/out).
• Motors will be tested for 15 Million revolutions
  at nominal torque.
• A lifetime test can be performed on the 3rd
  prototype. Intermediate recalibration of position
  sensor will give an idea of reproducibility of
  mechanism (not of the sensor) along the years.

74
Conclusions

• Measurement of motor/jaw response time.
• Risetime is L/R long cables are benefic (they
  add more R than L)
• For our case L30 mHenry 1 km cable 30 Ohm 1
  msec (compatible with preliminary measurements).

• In practice it depends on the chopping voltage
  (higher voltages boost the speed by charging
  quickly the inductance of the circuit).
  Eventually a series resistance could be added
  (but we would waste a lot of power)

75
Conclusions

• Evaluation of electro-magnetic compatibility with
  sensitive equipment close to the collimator (e.g.
  specify noise levels at some distance to the
  stepping motor, to the cable, ). BLMs must work
  unperturbed during jaw movement.
• The problem is fully understood
• A company exists that can sell equipment having
  no disturbance at all
• We are trying to motivate other companies to do
  that (A. Masi will visit SPS/IPC/Drives expo in
  Nurnberg end of November
• We will try ourselves to find an optimum filter
  to reach MIDI performances (or close to that)
• We might have a test at point 3 where all the
  cables have already been pulled.

76
Conclusions

• Motors, LVDT, Electronics problems are understood
  and in most cases the solution is already
  available or at least under study
• The real problem is the schedule, We have defined
  all the component interfaces so that in theory,
  we can mount all the motorization components
  directly in the tunnel not to stop the production
• LVDTs cannot be mounted in the tunnel because it
  has to be calibrated in metrology.
• Yearly calibrations have been conceived to avoid
  any intervention from human beings in the tunnel
  (apart maybe for the first). Their effectiveness
  has still to be demonstrated.

77
Stepping motors characteristics and test

• FULL step driving
• Half step driving