hubble space telescope - home | stsci · – high%frequency.sampling@full ......

34
Hubble Space Telescope STUC Meeting –5 November 2015 John W. MacKenty

Upload: vutuong

Post on 28-Aug-2018

217 views

Category:

Documents


0 download

TRANSCRIPT

Hubble Space Telescope STUC Meeting –5 November 2015

John W. MacKenty

Current  Hubble  Status  

•  All  science  instruments  are  performing  well  and  HST  observa:ons  con:nue  at  high  efficiency  

•  STScI  5-­‐year  HST  opera:ons  proposal  -­‐  Helmut  Jenkner  •  COS  wavelength  calibra:on  –  Cris:na  Oliveira  •  Cycle  23  and  Plans  for  Cycle  24-­‐  Claus  Leitherer    

STUC  –  November  2015  

Cycle  24  Schedule  (2016)    CP  Released  –  Jan  13  Mid-­‐Cycle  #2  –  Jan  31  Proposals  Due  –  Apr  8  TAC  /  Panels  –  Jun  5-­‐10  Phase  2  /  Budgets  –  Jul  21/8  Observa:ons  Start  –  Oct  1  

2  

•  Cycle 23 Long Range Plan released in early August, 2015. •  Observing started Oct 1, 2015.

•  Early progress: 86 orbits/week through six weeks. •  Consistent with previous post-SM4 cycles:

•  Cycle 17: 84.0 orbits/week •  Cycle 18: 83.4 orbits/week •  Cycle 19: 83.2 orbits/week •  Cycle 20: 85.5 orbits/week •  Cycle 21: 84.0 orbits/week •  Cycle 22: 84.0 orbits/week

•  Previous Cycle Completeness: •  Cycle 21: completed in October 2015. •  Cycle 20: 1 orbit remains, planned for late Dec 2015.

•  Cycle 23 Frontier Fields: •  The first epoch (70 orbits) of Abell S1063 is scheduled to be completed

by mid-November.

STUC  –  November  2015  

LRP:  Highlights  

3  

STUC  –  November  2015  

Current  state  of  the  opera:onal  LRP  •   complete  through  calendar  ending  11/15/15.  

Instrument Orbits

WFC3 1703

COS 886

ACS 622

STIS 574

FGS 3

Total 3788(3)

Cycle Orbits

20 1(1)

21 0

22 303(2)

23 3465

Total 3769

(1)  1 orbit planned Dec 2015. (2)  Plan Windows into January 2016.

C22 snaps 529

C23 snaps 1105

Total snaps 1634

Visits not in current plan orbits

unschedulable 45

no plan windows 0

C22 misc 113

C23 misc 306

Total not in plan 464

(3) Some programs have more than one prime SI.

4  

STUC  –  November  2015  

Fron:er  Field  programs  •  First  four  fields  complete  •  First  epoch  of  first  cycle  23  field  has  been  observed/scheduled.  

Frontier Field Cycle Total alloc

Exec/sched by 11/15/15

Planned before 9/30/16

Abell 2744 21 140 140 0

MACSJ0416.1-2403 21 140 140 0

MACSJ0717.5+3745 22 140 140 0

MACSJ1149.5+2223 22 140 140 0

Abell S1063 23 140 70 70

Abell 370 23 140 0 140

5  

STUC  –  November  2015  

Cycle  22  Large/Treasury  programs  

Program Total alloc

Exec/sched by 11/15/15

Planned before 9/30/16

Planned after 10/1/16

comment

Benneke 124 73 34 4 13 not in LRP

France 125 125 0 0 completed

Freedman 132 120 12 0 done 12/15

Malhotra 160 144 16 0 done 12/15

Oesch 132 128 4 0 done 11/15

Perlmutter(ToO) 86 86 0 0 completed

Robberto 52 52 0 0 completed

Skillman 81 81 0 0 completed

Spencer 194 194 0 0 completed

Tripp 99 99 0 0 completed

6  

STUC  –  November  2015  

Cycle  23  Large/Treasury  programs  

Program Total alloc

Exec/sched by 11/15/15

Planned before 9/30/16

Planned after 10/1/16

comment

Apai 112 18 22 0 72 not in LRP

Bedin 66 22 44 0

Borthakur 100 0 74 26

Coe (ToO) 190 5 78 82 25 not in LRP

Deming 111 0 74 37

Kirshner (ToO) 100 6 1 1 92 not in plan

Lehner 93 24 22 47

Papovich 130 6 106 18

Peterson 74 0 74 0

Siana 48 0 48 0

7  

ACS  Calibra:ons  Update  (N.  Grogin  and  the  ACS  Team)  

•  The  ACS/WFC  and  ACS/SBC  con:nue  to  perform  well:  –  Over  6.5  years  with  the  revived  WFC;  no  ACS  anomalies  in  >2.5  years  –  Strong  Cyc23  usage:  WFC  =  700o.  prime  +  505o.  parallel;  SBC  =  105o.  

•  The  recently  approved  Cyc23  ACS  Cal  plan  newly  includes:    –  Refined  photometric  &  astrometric  calibra:on  of  WFC  polarizers  –  Measurement  of  the  SBC  extended  PSF  (out  to  ~5”  radius)  

 

•  Long-­‐term  monitoring  of  WFC  low-­‐voltage  power  supply  (LVPS):  –  CY15  :  “Jump”  in  daily  average  of  8-­‐bit-­‐downsampled  LVPS  current  –  High-­‐frequency  sampling@  full  16-­‐bits  suggests  steady  +1  mA/yr  trend  –  Voltages  unchanging;  all  

 exps.  remain  nominal  

STUC  –  November  2015  

CCD  rest-­‐state  begins  crossing  next  8-­‐bit  threshold  

CCD  rest-­‐state  ~always  above  threshold  

8  

•  Subarray  Redesign  –  Calibra:on  headaches  for  post-­‐SM4  ACS/WFC  subarrays:  

•  De-­‐biasing  post-­‐SM4  subarray  images  •  Readout-­‐:ming  Δ  makes  pixel-­‐based  CTE  correc:on  inapplicable  to  non-­‐2K  subarrays  

•  Readout  overheads  longer  than  full-­‐frame;  <2K  columns  prevents  bias-­‐ship  correc:on  

–  Solu:on:  Re-­‐define  WFC  subarray  readouts  to  match  full-­‐frame  :ming  •  Twelve  new  subarray  modes,  all  with  2K  columns:  (512,1K,2K)  rows  on  all  quadrants  (A,B,C,D)  •  Subarray  biases  no  longer  needed  (excerpt  from  full-­‐frame);  iden:cal  CALACS  steps  as  full-­‐

frame  

•  “Spare  the  Pixels”  –  Elimina:ng  “bad-­‐column”  DQ  flagging    

 from  WFC  superbiases  –  Stable  hotpix  à  stable  “bad  cols.”    –  Reducing  warm-­‐  &  hot-­‐pixel  DQ  flagging  from  “stable”  dark  current  –  Only  DQ-­‐flag  the  “unstable”  pixels  (variance  significantly  g.t.  Poisson)  

ACS/WFC  Subarrays  Re-­‐design  and  “Spare  the  Pixels”  Ini:a:ve  

Actual  superbias  (includes  gradient)  

Simulated  readout  dark  (no  gradient)  

STUC  –  November  2015   9  

HST  Focus  Maintenance  

HST  experiences  temperature-­‐induced  focus  varia:ons  due  to  varying  poin:ng  avtude  and  LEO  geometries.  On  top  of  this  there  is  a  long-­‐term  shrinkage  of  the  HST  metering  truss  which  brings  the  SM  closer  to  the  PM.  We  periodically  correct  for  this  by  backing  the  SM  away  from  the  PM  to  maintain  focus.  This  shrinkage,  and  the  compensa:ng  SM  adjustments  are  clear  in  the  plot.    Our  three  SM  moves  since  January  2013  have  reacted  to  recent  shrinkage  rates  and  have  brought  and  kept  us  close  to  best  focus  over  the  past  two  years.    In  May  2014  we  re-­‐fit  focus  data  to  regenerate  our  temperature-­‐based  focus  model,  which  contains  a  func:on  describing  “nominal”  secular  focus  change  (from  shrinkage)  along  with  temperature  terms  expressing  temperature  driven  excursions  about  nominal.  We  made  this  updated  model  available  to  the  community.  

See  hxp://www.stsci.edu/hst/observatory/focus  for  related  informa:on  and  access  to  the  model  data.    We  con:nued  inves:ga:ons  this  year  into  tracking  focus  using  the  complete  library  of  WFC3  observed  science  PSFs.  This  holds  the  promise  of  producing  a  poten:ally  viable  focus  indicator  that  could  supplement  or  possibly  even  supplant  the  dedicated  focus  monitor  program.    

Plot  shows  the  mean  loca:on  of  the  HST  Secondary  Mirror  (SM)  since  SMOV4,  expressed  in  microns  of  axial  displacement  wrt  WFC3  best  focus  (WFC3  focus  is  within  0.5  microns  of  ACS  focus  and  similarly  confocal  to  STIS  &  COS)    We  obtain  this  secondary  mirror  displacement  from  the  observed  focus  aberra:on  as  determined  by  our  rou:ne  phase  retrieval  code  from  star  images  taken  in  ACS/WF  and  WFC3/UVIS.  1  micron  of  Secondary  Mirror  axial  displacement  produces  6  nanometers  rms  wavefront  error  (via  focus  aberra:on),  or  lambda/100  at  600  nm.  

STUC  –  November  2015   10  

HST  Focal  Plane  Evolu:on  

50% within 0.23 arcsec (monthly means)

50% within 0.26 arcsec (all points)

A B C

E

Observed minus Predicted Guidestar Separations

D

To  provide  insight  into  the  evolu:on  of  the  FGS  Fields  of  View  on  the  HST  focal  plane,  we  produced  a  plot  of  the  observed  vs.  expected  guidestar  pair  separa:ons,  serving  as  a  coarse  indicator  of  FGS  misalignments.  Approx.  200,000  guidestar  pair  measurements  spanning  10  years  were  obtained  and  ploxed.  The  result  corroborates  our  understanding  that  focal  plane  calibra:ons  in  the  form  of  FGS  alignment  updates  (when  performed)  result  in  very  low  errors  of  a  few  tens  of  milliarcseconds,  but  that  evolu:on  of  the  FGS  FOVs  con:nues  to  steadily  build  posi:on  errors  that  can  grow  to  a  few  tenths  of  an  arcsecond.  Currently  we  are  working  on  an  update  to  address  the  accumula:ng  error  (mainly  from  FGS2).  

A.)  HST’s  opera:onal  guidestar  catalog  is  changed  over  from  GSC1  to  GSC2.  Note  the  scaxer  of  the  individual  (grey)  points  decreases  at  this  :me.  B.)  Freshly  characterized  FGS  alignments  were  updated  in  HST  opera:ons.  Note  the  improvement  in  the  mean  error  to  <  0.1  arcsec  for  the  next  year.  C.)  During  SM4  Observatory  Verifica:on,  the  newly  installed  FGS2  was  characterized.  (Disregard  servicing-­‐mission-­‐related  points  in  first  half  of  2009.)  D.)  Freshly  characterized  FGS  alignments  were  updated  in  HST  opera:ons.  Improvement  is  clear,  though  FGS2’s  dynamics  are  primarily  responsible  for  con:nued  trending.  E.)  Freshly  characterized  FGS  alignments  and  distor:on  mappings  produce  ~<0.05  arcsecond  mean  errors  for  a  number  of  months  before  con:nued  FGS  trending  causes  error  to  steadily  accumulate.  

STUC  –  November  2015   11  

COS  +  STIS  –  Generic  Updates  (C.  Oliveira  and  the  COS/STIS  Team)  

•  Flux  and  flat-­‐field  reference  files  to  calibrate  COS  LP3  data  were  released  in  July    2015  

•  New  version  of  the  COS  Data  Handbook  published  in  Oct  19  2015  (v3)  –  Contains  many  updates  from  v2,  including  a  descrip:on  of  the  new  TWOZONE  extrac:on  

algorithm  for  FUV  data  taken  at  Life:me  Posi:on  3  

•  New  CalCOS  pipeline  to  be  installed  by  mid-­‐November  –  Contains  code  and  a  new  reference  file  to  exclude  events  affected  by  hotspot  from  final  co-­‐

added  products  and  minor  tweaks  related  to  TWOZONE  extrac:on  

•  Monitoring  of  gain  sag  on  COS/FUV  detector  –  High  voltage  in  FUVB  will  need  to  be  increased  in  mid-­‐Dec  2015  to  maintain  modal  gain  

above  PHA  of  3  in  the  whole  detector  

•  Have  started  high  level  planning  for  COS/FUV  Life:me  Posi:on  4  –  At  current  usage  rate  will  need  to  move  to  LP4  in  ~Summer  2017  –  LP4  will  likely  be  below  LP3,  impact  on  resolu:on  will  be  evaluated  using  LSF  models  and  

data  

•  Pixel-­‐based  stand-­‐alone  automated  script  to  correct  STIS  CCD  CTI  was  released  to  the  community  in  Sep  2015  

STUC  –  November  2015   12  

COS  –  FUV  TDS  Trending  (C.  Oliveira  and  the  COS/STIS  Team)  

STUC  –  November  2015  

LP1   LP2   LP3  

Data up to Sep 2015 included

Breakpoints in slope HV Increase

•  TDS slopes in 2014 steeper than 2013, up to ~8% per year, stable up to now

•  Updated TDS ref. file delivered Fall 2014 •  Overall decrease since COS installation varies

~ 10 – 35%

•  Rate of decline of COS FUV Time Dependent Sensitivity (TDS) varies with time, detector, and λ

•  Steeper TDS slopes in periods of increased solar activity – likely due to atomic O at HST’s orbit reacting with CsI photocatode of open-faced COS FUV detector

•  Increase in sensitivity with HV and TDS dependence on cenwave are under study

TDS slope vs. wavelength t > 2013.8

13  

COS+STIS  Cy  23  Calibra:on  Plans  (C.  Oliveira  and  the  COS/STIS  Team)  

•  COS  calibra:on  program  –  Nine  programs  monitor  the  performance  of  the  FUV  and  NUV  detectors  (dark  rate,  sensi:vity,  TA,  gain,  and  

wavelength  scale)  as  well  as  a  special  program  to  obtain  Lyα  airglow  spectra  

•  STIS  calibra:on  program    –  Twenty  programs  monitor  the  performance  of  the  CCD  and  FUV  +  NUV  MAMAs  (dark  rate,  read  noise,  flat  field,  

sensi:vity,  etc.)  as  well  3  special  programs    •  Measure  the  gain  of    Amps  A,  C,  and  D    •  Monitor  STIS  focus  in  rela:on  to  WFC3,  ACS  •  Push  the  limits  of  coronographic  BAR5  (see  discussion  below)  

STUC  –  November  2015  

•  STIS has the sole coronograph still operating in space –  Current STIS performance is 10-4 contrast at 0.25”

•  NIR ground-based extreme-AO has ~10-6 contrast at 0.18” •  WFIRST/AFTA CGI will have ~10-9 at 0.18”

–  Goal of special BAR5 program (9 orbits) is to demonstrate contrast levels of 10-7 at 0.15” •  Will enable new discovery science with increase in

contrast by 1-3 more orders of magnitude compared to current capability

•  Processed data and new capability will be advertised to users before Cycle 24 proposal deadline

•  Data obtained recently from first 3 visits (3 orbits) indicates a contrast of 10-6 at 0.5” (~10 pix)

Pushing the Limit of BAR5 (cont.)!

•  Current STIS performance: ~10-4 contrast at 0.25”!

•  NIR ground-based extreme-AO: ~10-6 @ 0.18”!

• WFIRST/AFTA CGI: 10-9 @ 0.18”!

=> Push STIS performance to 10-7 contrast at 0.15”!

F1V! K6V!

Classical PSF Subtraction!

24!

BAR5 contrast curves from !commissioning data (P12923)!

14  

Smart archives organized by target and science use: all relevant co-

added data in one click!

Full database starting with COS FUV (10000 datasets) available in

time for Cy24 Proposals

Preview release to be shown at AAS in Kissimmee in January -

come to the STScI booth to see a demo and provide feedback!

Contact for more information: Jason Tumlinson (tumlinson)

Molly Peeples (molly) Andrew Fox (afox)

² WFC3  is  operaHng  nominally  

²  New  features  available  to  the  observers:  ²  IR  SPARS5  Sample  Sequence  –  First  GO  observa:ons  successfully  

acquired  on  September  29,  2015  ²  2  new  full  array  UVIS  aperture  defini:ons  (place  the  target  near    the  C  

amplifier)    ²  New  version  of  CALWF3  delivered  to  TEST  –  will  be  sent  to  OPS  (and  release  

to  the  community)  in  December  2015  –  all  WFC3  data  (sta:c  archive)  will  have  to  be  reprocessed.  

²  Ongoing  ac:vi:es:  ²  IR  background  model  for  GRISMs  and  and  imaging  ²  Extend  astrometric  solu:on  to  all  the  UVIS  filters  ²  GRISM  tools  ²  CMS  monitoring  +  CMS  movement/week  <30  

Wide  Field  Camera  3  

STUC  –  November  2015   16  

UVIS  Channel  Performance  

²  UVIS  read  noise  over  a  period  of  6  years  (from  installa:on  to  May  2015)  increased  by  1.4-­‐1.9%,  due  likely  to  CTE  losses,  the  ever-­‐growing  hot  pixel  popula:on,  and  the  general  aging  of  the  instrument.    

²  Over  6  years  of  observa:ons  the  varia:on  in  exposure  :me  across  the  detector  due  to  shuxer  shading  remain  less  than  0.1%.    

²  There  is  no  no:ceable  difference  in  the  present  performance  of  the  shuxer  mechanism  compared  to  its  performance  during  SMOV    

²  Aper  6  years  UVIS  gain  measurements  remain  within  1-­‐2%  of  the  values  derived  in  TV3  and  SMOV.    

STUC  –  November  2015   17  

UVIS  Charge  Transfer  Efficiency  (CTE)    

Radia:on   damage   effects   in   low-­‐earth  orbit  CCDs  are  responsible   for  degrading    CTE  

Flux  loss  due  to  CTE  degrada:on  is  a  func:on  of  the  source’s  distance  from  the  amplifier,  the  source  signal  level,  the  background  within  the  image,  and  the  epoch  of  the  observa:ons.  In  the  worst  cases  in  early  2015  losses  for  faint  sources  can  be  as  high  as  50%.  Small  amount  of  pos}lash  reduces  the  losses  for  faint  sources  to  15%  Pos}lash  +  pixel-­‐based  CTE    correc:on  reduce  losses  to  3%    

CTE  close  to  the  Amp   CTE  far  from  he  Amp  

CTE  as  a  func:on  of  background  STUC  –  November  2015   18  

IR  Channel  Performance  

²  Aper  6  years  of  opera:ons  IR  gain  measurements  remain  within  2%  from  the  values  derived  during  SMOV    

²  Since  Cycle  18  we  observed  a  decrease  of  0.3%  yr-­‐1  in  the  count  rate  of  the  tungsten  lamp  (aging)  

²  Par:culate  maxer  on  the  CMS  mirror  is  imprin:ng  small  roughly  circular  regions  of  moderate  axenua:on  on  IR  images  (a.k.a.  blobs).    ²  Blob-­‐corrected  flats  are  improving  stellar  photometry  ²  Blob-­‐corrected  flats  are  now  available  to  GOs.      ²  Very  few  new  blobs  in  recent  years  

STUC  –  November  2015   19  

Improved  Model  for  Persistence  in  IR  

²  Faint  aperglows  of  earlier  exposures  are  some:mes  seen  in  WFC3  IR  data  ²  Persistence  is  due  to  traps  in  the  pixels  of  an  IR  detector    

²  The   amount   of   persistence   is   a   func:on   of  exposure  history  of  a  pixel  in  the  detector    

²  We   improved   our   model   by   including   the  exposure   :me   of   the   earlier   exposure   as   an  addi:onal  factor  in  the  persistence  predic:on    

²  Persistence   clearly   varies   across   the   WFC3   IR  detector  

²  Varia:ons   appear   to   primarily   on   large   spa:al  scales,  with  pixel  to  pixel  varia:ons  being  small.  

²  A   correc:on   “flat”   has   been   incorporated   into  the   persistence   predic:on   sopware   used   to  es:mate  persistence  in  HST  images.  

Persistence  decay  as  a  func:on  of  exposure  :me    

STUC  –  November  2015   20  

GRISM  tools  

In   progress   advanced   GRISM   data  reduc:on  algorithms/sopware  

o  Tool  to  handle  observa:ons  at  mul:ple  roll  angles  

o  Forward  Modeling  methods  to  extract  fainter  sources  and  understand  errors  

o  Contamina:on  predic:on  code:  use  a  direct  image  of  field  to  predict  the  contamina:on  for  each  target  of  interest    

Recalibrated   1st   order   using   250/350   G102/G141   datasets,   800+   uniformly   distributed  posi:ons  on  the  field,  bexer  field  coverage  and  solu:on.    

IR   wavelength   solu:on   using   the   Zero   order  image  as  reference  point.    

STUC  –  November  2015   21  

Gyro  4  Bias  Drips  

STUC  –  November  2015   22  

Gyros  –  Take  away  message  

•  Highly  likely  that  HST  Gyro’s  will  support  science  opera:ons  un:l  (and  probably  beyond)  end  of  the  decade  

•  Hubble  has  a  very  complex  poin:ng  control  system  with  considerable  resilience  to  failures  of  its  various  components  –  6  (now  5)  Gyros,  3  FGS,  3  FHST  (plus  addi:on  elements  to  support  safe  modes)  

•  Most  likely  impacts  are  to  scheduling  efficiency  and  field  of  regard    

STUC  –  November  2015   23  

Rate Gyro Assemblies (RGA)"Consists of Electronics Control Unit (ECU) and Rate Sensor Unit (RSU) (manufactured by L-3 Communications (then Bendix))"

–  Each RSU contains 2 Rate Integrating Gyros; HST has a total of 6 Gyros"

–  3 Gyros are routinely in control loop"

STUC  –  November  2015   24  

HST  Poin:ng  &  Avtude  Control  System  

•  HST  uses  two  types  of  sensors  to  maintain  vehicle  avtude  –  Rate  Gyro  Assemblies  (RGAs)  –  Fine  Guidance  Sensors  (FGSs)  

•  RGAs  are  used  to  provide  the  rate  input  to  the  40  Hz  vehicle  control  law  (VCL)  during  all  science  mission  phases  

•  FGSs  provide  posi:on  and  rate  correc:on  in  vehicle  frame  to  the  VCL  through  the  1  Hz  Avtude  Observer  during  fine  guiding  intervals  

•  During  FGS  guiding  intervals,  the  feedback  from  the  FGSs  is  used  to  provide  real  :me  measure  and  compensa:on  of  the  gyro  rate  bias  error  produced  by  all  of  the  gyros  in  the  control  loop  

 

STUC  –  November  2015   25  

Impacts  to  Opera:ons  

•  Gyros  are  needed  to  acquire  guide  stars  (i.e.  knowledge  of  vehicle  avtude  during  slew  and  occulta:on  periods)  –  Poorly  performing  gyros  lead  to  guide  star  acquisi:on  failures  –  One  or  two  gyro  modes  developed  

•  Lower  observing  efficiency  •  Dependent  upon  FHST  avtude  updates  

•  Avtude  error  accumulates  due  to  the  varying  bias  signal  between  FGS  guiding  intervals  (AOA  =  Avtude  Observer  Anomaly)  

•  There  are  two  types  of  GS  Acquisi:ons  –  Primary  Acquisi:ons  (Acqs)  

•  Occur  aper  large  vehicle  slews  &  are  preceded  by  an  FHST  OBAD  avtude  correc:on  which  corrects  any  avtude  errors  from  AOA  bias  errors  

•  These  also  typically  have  search  radii  of  55  arcsec  or  bexer  –  Reacquisi:ons  (Reacqs)  

•  Occur  aper  primary  acquisi:ons  and  other  Reacqs  with  no  FHST  OBAD  avtude  correc:ons  

•  These  always  have  search  radii  of  30  arcsec    

STUC  –  November  2015   26  

Impacts  to  Opera:ons  

•  During  the  FGS  guide  star  acquisi:on  process  the  FGSs  measure  the  residual  avtude  error  aper  gross  avtude  correc:ons  (>0.3  arcsec)  have  been  completed    

•  The  observer  convergence  period  is  then  executed  to  drive  the  residual  error  to  zero  within  40  seconds  

–  This  convergence  is  affected  by  any  residual  or  uncompensated  rate  error  –  If  the  error  is  not  within  the  Loss  of  Lock  threshold,  the  acquisi:on  will  be  re-­‐axempted  up  to  

3  :mes  –  Generally  uncompensated  bias  errors  (due  to  the  AOA)  will  cause  all  4  axempts  to  fail  and  

the  acquisi:on  will  fail  –  It  can  also  delay  the  sevng  of  the  science  ini:aliza:on  flag  and  impact  science  opera:ons  to  

varying  degrees    •  Two  complementary  approaches  have  been  iden:fied  to  reduce  the  vehicle  errors  

induced  by  the  AOA  –  A  historical  observer  that  will  supplement  the  exis:ng  gyro  bias  compensa:on,  providing  

varying  levels  of  gyro  rate  bias  compensa:on  depending  on  the  vehicles  orbital  posi:on  rela:ve  to  EOD.  

–  An  ini:aliza:on  of  the  exis:ng  avtude  observer  to  rates  observed  by  the  Dominant  FGS,  that  will  allow  the  observer  to  more  rapidly  reduce  rate  error  residuals  and  posi:on  error  during  the  acquisi:on  sequence.  

•  Cascading  failures  are  more  likely  to  occur  in  automated  opera:ons  

STUC  –  November  2015   27  

HST Gyro Run Times through 2015/Sep-30"

Key"" = Failed Gyro""SFL = Standard Flex Lead "EFL = Enhanced Flex Lead (silver plated)""RR = Rotor Restriction "AOA = Attitude Observer " Anomaly"

1  Flight Times are powered-on times and do not account for increased degradation when off due to thermal effects from a companion gyro that is powered on"

Launch Gyros 1 – 108 – 1005 SFL Failure 3.9343 0.4573 4.3916 38470

(4/1990) 2 – 113 – 1005 healthy 4.1884 0.4219 4.6103 403863 – 110 – 1002 healthy 3.6142 1.1620 4.7762 418404 – 138 - 1002 healthy 2.4964 1.0763 3.5727 31297

5 – 104 – 1003 healthy 3.6142 1.2574 4.8716 426756 – 127 - 1003 SFL Failure 2.4570 1.5185 3.9755 34825

SM1 Gyros 3 – 112 – 1007 SFL Failure 5.3731 0.4776 5.8507 51252

(12/1993) 4 – 158 – 1007 SFL Failure 3.3434 0.2553 3.5987 315255 – 118 – 1006 healthy 6.0478 0.4943 6.5421 573096 – 151 - 1006 SFL Failure 4.8808 0.4019 5.2827 46276

SM3A Gyros 1 – 155 – 1003 healthy 6.2003 0.6997 6.9000 60444

(12/1999) 2 – 110 – 1003 SFL Failure 6.4126 0.2128 6.6254 580393 – 104 – 1002 RR, stall 3.3482 0.4369 3.7851 331574 – 138 – 1002 AOA 5.6929 0.4046 6.0975 534145 – 156 – 1008 RR, stall 1.3458 0.2361 1.5819 138576 – 159 - 1008 healthy 2.8769 0.2217 3.0986 27144

SM4 Gyros 1 – 127 – 1005 healthy 1.6583 0.7339 2.3922 20956

(5/2009) 2 – 160 – 1005 healthy 1.6583 0.7535 2.4119 21128

EFL 3 – 148 – 1004 AOA (powered off) 1.8833 0.6684 2.5517 22353

EFL 4 – 150 – 1004 RR, high motor current 6.3847 1.1981 7.5828 66425

5 – 161 – 1006 SFL Failure 4.8174 1.0613 5.8787 51497

EFL 6 – 152 - 1006 Healthy (powered off) 3.1089 0.9944 4.1033 35945

Total(Hrs)Gyro-ID/RSU Status at End Flight time 1

(Yrs)Test Time

(Yrs)Total (Yrs)

ß Gyro 4 à longest on-orbit run time"

Current On-Orbit Configuration:"G1 – G2 – G4"

STUC  –  November  2015   28  

Gyro  1  Gyro  2  Gyro  4  

Post  SM4  Gyro  4  has  been  exhibi:ng  rela:vely  significant  bias  ships  perhaps  indica:ve  of  flex  lead  degrada:on  

4/30                                                    5/28                                                        6/25                                                    7/23                                                    8/20                                                    9/17                                                10/15                    10/29                                            

STUC  –  November  2015   29  

" " "Recent Gyro 4 Trends"

-  Over the last several months, Gyro 4 has been exhibiting large bias rate shifts [see plot]"-  On September 19, Gyro 4 experienced a large ~150 arcsec/hr rate bias shift "

•  Attitude errors accumulated as a result of the large drift rate and caused the failure of several Guide Star Acquisitions over the course of ~ 4 orbits"

•  The effect was exacerbated by the unusually long delay between the FGS Guide Star Acquisition and the preceding On-Board Attitude Determination update"

-  Another large bias shift of ~80 as/hr occurred on September 23, however the onboard control system was able to compensate "

-  A third large bias shift of ~ 86 arcsec/hr occurred on October 16 and subsequent bias updates of the same general level indicated a rapidly changing bias"

•  The large bias rate accumulated error over an 8 hour period because onboard bias updates could not be be scheduled during the 8 hour science observations "

•  The rate of the changing bias slowed and returned to nominal values with no loss of science"-  The Gyro 4 bias trend has re-stabilized after the large bias shift on October 16 and it appears to be

following the trend-line established prior to May 2015, albeit noisier [now up again!]"-  Mitigations"

-  The GS Acquisition maximum search radius was increased to support larger potential attitude errors from uncompensated biases"

-  The minimum data threshold for performing an onboard gyro bias update was reduced from 20 minutes to 5 minutes (to accommodate rapidly changing bias rates)"

-  Other mitigations are being considered; the effectiveness of the mitigations have not been determined because of the quieting down of Gyro 4"

-  Other than the loss of science on October 19, no further impacts have occurred"STUC  –  November  2015   30  

RGA Anomalies – Flex Lead Failure"

Standard Flex Lead (SFL) Failure"–  Lead Composition 85% Ag, 15% Cu"–  Corrosion along the flex lead(s) creates a hot spot"

•  Spot is sensitive to nominal temperature variation, causing mechanical motion (large Gyro bias rates)"

–  Lead eventually breaks causing loss of 1 phase of motor"•  Large bias shift in gyro frame (hundreds of arcsec/hour)"•  Motor current ~doubles "•  Increased heater duty cycle due to less current power

dissipation in the motor"–  Rate of SFL degradation increases exponentially with temperature

and continues even if gyro is off when companion gyro is on"–  Successful spin-up after powering down is not possible as both

leads are required for full phasor to be applied to solid core rotor"

Enhanced Flex Leads (EFL)"-  Coat the flex lead with a protective barrier to stop

degradation caused by the BFTE float fluid"-  Silver plating will double the life expectancy of the gyros"

•  Copper dissolves 100x more quickly than silver"

STUC  –  November  2015   31  

RGA Anomalies – Rotor Restriction"

Gyro  5  Rotor  Restric:on  Jump  in  Motor  Current  

~50.8 mA"5 min"

–  A build-up of particles and lubricant between the shaft and the rotor results in a transient load torque and a momentary disruption in spin motor phase, causing motor to draw more current to maintain speed"

–  Gyro operates at higher current until a power cycle"

•  Jumps of 30-50 mA not uncommon"•  Moderate bias shifts in gyro frame

(Tens of arcsec/hour)"–  Successive jumps may cause gyro to stall (stall

current ~ 330 mA)"–  Power cycle may bring gyro back to pre-rotor

restriction current levels, however vendor recommends that gyros that continue to run following rotor restrictions  not  be power cycled due to increased risk of non-start"

–  It was determined that gyros that failed on the ground and on orbit due to rotor restriction tended to have shaft to rotor spacing on the low end of tolerance. Tolerance was modified to ensure greater spacing. Gyros already built were evaluated with new tests to ensure rotor freedom of movement about the shaft (Tumble Test)."

STUC  –  November  2015   32  

RGA Anomalies - Attitude Observer Anomaly (AOA)""

•  Trending showed that the day/night cycles were causing increasingly larger variations in the estimated bias in the V2 axis (in current on-orbit configuration, V2 axis bias error is driven by Gyro 4, although Gyro 4’s AOA signature continues to be small) "

–  These variations caused uncompensated rate errors sufficiently large enough to cause the subsequent GS Acquisition to declare Loss of Lock as insufficient time is allowed for bias estimate to be established"

•  The cause of the AOA signature is not completely established; a possible cause is asymmetric corrosion-induced degradation between the 7 gyro flex leads "

–  The orbit terminator heat pulse creates mechanical stresses as it travels through the asymmetrically corroded flex leads, these impose a force on the float."

-  Not all of the Gyros with AOA signatures impacted science operations"-  Mitigation:"

–  The Historical Observer (HO) algorithm developed to mitigate AOA bias and activated in November, 2012 was designed to measure/compensate the orbital variation per minute (vs. an orbital average applied previously) in the gyro rate bias "

–  Currently, only a nominal orbital variation is evident in the gyro bias compensation; Gyro bias is well monitored."

•  Additionally, the vehicle control law’s Attitude Observer, which determines real time estimates of uncompensated gyro bias from the FGS while fine guiding, was modified to initialize with a coarse estimate of the uncompensated bias from FGS data early in the acquisition sequence, allowing for rapid convergence on large bias error during allotted time during guide star acquisition (Attitude Observer Initialization (AOI))."

•  For Gyro 4, the AOI and HO compensation have been successful in minimizing the effect of the bias variations. To date, there has been relatively minor impact to science operations due to the Gyro 4 bias"

"

"

STUC  –  November  2015   33  

"Though Gyro 4 is an EFL gyro, there is the possibility that the recent bias signature indicates Gyro 4 may fail in the near term (hopefully not). ""-  In the event of Gyro 4 failure (or any of the current complement), the vehicle will enter Kalman Filter Sun

Point (KFSP) safemode interrupting science "

-  The failed gyro will be replaced with Gyro 6 to maintain the 3-Gyro Science Mode configuration "•  3-Gyro Mode provides maximum science performance and scheduling flexibility while HST and

Science Instruments are operating at near peak performance "

-  The KFSP Safemode recovery procedure was recently updated and tested"•  A redlined procedure is in development specifically for a Gyro 4 failure

"As the failure described above would be the second gyro failure (since SM4), an assessment would need to be made as to the gyro configuration for subsequent failures"

Note:!Though Gyro 3 has some performance problems, it can support 3-Gyro Mode (with some impacts).!Also, when down to only two viable gyros, the option of going to One Gyro Science Mode rather than Two Gyro Science mode is very advantageous for preserving the second gyro. Modeling and on-orbit testing has demonstrated that One Gyro Science is as effective (assessed over all metrics) as Two Gyro Science Mode.!

Forward Plan (to date, and regularly revisited)

 

STUC  –  November  2015   34