Hubble Space Telescope · 1990

Diagnosing Hubble's spherical aberration

First to fit Hubble's first-light image with phase retrieval and find half a wave of spherical aberration.

Method

Phase retrieval: least-squares fit of a Fourier-optics model of the telescope to star images

Result

About 0.5 wave rms of spherical aberration, confirmed by every later analysis

Recognition

NASA Public Service Medal, 1992

Oral history

Interviewed by Chris Gainor, 15 February 2017, HST Operational History Project

Before launch Chris had written the Optical Telescope Assembly Handbook and, with Hashima Hasan, the Telescope Image Model: an end-to-end simulator built on optical prescriptions of every instrument, including diffraction, dust and mirror microroughness. That preparation is why he could read the first images when nobody else could.

The first-light image was expected to show a misaligned secondary mirror. Chris wrote a new fitting procedure in IDL that solved for the aberrations directly from the brightest star in the field, including spherical aberration only for completeness. Within about a day the fit converged on roughly half a wave of it. The next day he arrived at a Goddard meeting where Jim Westphal was presenting the image and told the room he understood it, with simulations that reproduced what the telescope was seeing. Adding the secondary-mirror spiders to the model then explained the image's puzzling tendrils.

For about a month the result was resisted, in particular by the contractor, whose own fits had found nothing wrong. The turning point was a meeting at which Bob Basedow said that if Chris was right, the primary-mirror actuators were a factor of seven too weak to correct it. That was when he knew the problem could not be fixed in orbit. The project ordered silence while out-of-focus images that Chris had planned were taken; they showed the through-focus behaviour of spherical aberration. Jon Holtzman independently confirmed the result by running Chris’s Telescope Image Model, which simulates images but does not fit them, to simulate the out-of-focus images with Chris’s aberration values; the simulations closely matched the observed out-of-focus images, showing the error was smooth. Robert Brown, in his NASA oral history, recalls Chris as “the first person to know that it was a spherical aberration problem”.

Chris then presented to the Science Working Group and sat on the technical panel at NASA's announcement. Faint Object Camera images, far enough off axis that a secondary-mirror error would have shown as coma, placed the fault on the primary. He served on the HST Strategy Panel that chose the recovery path, and on the HARP panel that pinned down the prescription from flight images and the “fossil” ground-test data; it came back to the half wave he had first announced. He then used the same software to verify, on the bench, that the WFPC2 relay optics and the COSTAR mirrors carried exactly the opposite error.

In recognition of your achievements in understanding and characterizing the optical performance of the Hubble Space Telescope. — NASA Public Service Medal citation, 1992

Timeline

• 1984, 1 April
  Arrives at STScI as one of ESA's fifteen staff; joins the Research Support Branch with Pete Stockman, Jim Crocker and Pierre Bely.
• 1984–89
  Writes the OTA Handbook and, with Hashima Hasan, builds the Telescope Image Model (TIM).
• 1990, 24–25 April
  Hubble launched and deployed. The optical control system's wavefront sensors fail to give usable fringes, later understood to be because of the aberration.
• 1990, 20 May
  First-light image with the Wide Field/Planetary Camera. Chris identifies the field from a star catalogue and starts fitting the star images.
• 1990, 22–25 May
  Dated IDL fitting code in Chris’s STScI files: simplex optimiser (22 May) and PSF-fitting driver (25 May).
• Within about a day
  IDL phase-retrieval fit finds about half a wave of spherical aberration.
• Next day
  Interrupts a Goddard meeting where Jim Westphal is presenting the image: “I understand this image.”
• Following weeks
  Second- and third-light images refine the model; spider obscurations explain the tendrils. Frank Cepollina has a VAX and a viewgraph printer delivered to his office. Two-day loops: design an observation, take it, fit it, present it.
• 1990, 30 May
  Nineteen-page memorandum making the quantitative case for spherical aberration from a “symmetrically deviant circular mirror”, addressed to Rodger Doxsey and Chris Blades; shown to Eric Chaisson ten days after first light and then sent to NASA (Chaisson, The Hubble Wars, pp. 190–191).
• 1990, 5 June
  At the weekly performance meeting at Goddard, Hughes Danbury reports its phase retrieval of a 3 June image: spherical ~0.2 wave, coma ~4 waves, attributed to secondary-mirror misalignment (meeting minutes, in the Archive; also quoted by Chaisson, p. 192).
• About a month
  Contractor (Hughes Danbury Optical Systems) disputes the result.
• June 1990
  Key meeting with John Campbell, Bob Basedow, Jean Olivier, Jim Westphal, Sandra Faber, Jon Holtzman: actuators a factor of seven short. Instructed to keep quiet.
• June 1990
  Iota Carinae test images: Chris’s simulations reproduce the observed images (Chaisson, p. 194). By 22 June the Institute’s position, based on his simulations, is that the primary mirror is the likely culprit (p. 196).
• 1990, 21 and 23 June
  Out-of-focus Planetary Camera exposures of a star near Iota Carinae, with the secondary mirror moved 312 µm toward the primary and then the maximum distance away (exposure log IMAGES.LOG in the original files).
• June 1990
  Out-of-focus images planned by Chris confirm spherical aberration. Jon Holtzman runs Chris’s Telescope Image Model (a simulator, not a fitting code) with the half-wave value to simulate the out-of-focus images, and independently reproduces them. The phase-retrieval analysis is documented in Krist & Burrows, Applied Optics 34, 4951 (1995).
• June 1990
  Warns STScI news director Ray Villard the day before the announcement.
• 1990, 27 June
  Chris presents the diagnosis to the HST Science Working Group, then sits on the second, technical panel at the NASA press conference that announces the flaw in the primary mirror to the world.
• Summer 1990
  Faint Object Camera first light shows no coma: the error is on the primary. Visits HDOS with John Wood; the backup primary is uncrated.
• 1990, July
  Writes the STScI Newsletter assessment of imaging performance: “approximately 0.5 wavelengths RMS of spherical aberration” (NEWSLETTER.TEX, 27 July 1990, in the original files).
• 1990, Aug–Oct
  HST Strategy Panel, chaired by Robert Brown and Holland Ford, including a meeting at ESO in Garching. Chris argues for a full-aperture corrector; COSTAR is chosen.
• 1990–91
  HARP panel reconciles flight and ground data; prescription confirms half a wave rms.
• 1991, 10 March
  Burrows et al., The imaging performance of the Hubble Space Telescope, ApJ 369, L21.
• 1991–93
  Verifies WFPC2 relays and COSTAR mirrors on the bench. Weeks before shipping, resolves a false out-of-focus result in Goddard's WFPC2 test.
• 1993, December
  Servicing Mission 1 installs WFPC2 and COSTAR; Chris aligns WFPC2's pupil on the 30 Doradus field, chosen so alignment images would also yield science.
• 1994, January
  Presents WFPC2 first-light images to the AAS: “we got the sign right.”

Original figures, 1990

Original files, 1990

Copied without change from Chris’s STScI VAX directories (later archived to his own files), with their original dates. IDL code (.PRO), GEIS image headers (.HHH, .D0H) and data (.HHD, VAX floating point), logs and PostScript. The original exposures are public in the MAST archive; the last two groups below link the calibrated and the raw files directly. The 1990 analysis began on raw data that Chris and Glenn Schneider took directly, bypassing the routine ground system to save time (the B suffix in the 1990 local file names corresponds to the archive’s T suffix).

Record of the discovery

• Eric J. Chaisson, The Hubble Wars (HarperCollins, 1994), pp. 190–196, 205, 221. Chaisson, then STScI’s director of public affairs, records that on 30 May 1990, ten days after first light, Chris showed him a nineteen-page memorandum with a thorough numerical analysis of the WF/PC first-light images making the case for spherical aberration, and that it was then sent to NASA. He calls Chris the person “who had originally identified the problem” (p. 205) and “who had uncovered the mirror’s flaw” (p. 221), and records that a week later the contractor’s own analysis still found negligible spherical aberration. Source →
• Minutes of the HST Performance Monitoring and Assessment meeting #5, Goddard, 5 June 1990. Contemporary record of the contractor’s position: Hughes Danbury’s phase retrieval on a 3 June Planetary Camera image gave spherical aberration of only ~0.2 wave peak-to-valley and ~4 waves of coma, blamed on secondary-mirror misalignment. This was six days after Chris’s memorandum. Excerpt in the Archive. Source →
• Chris’s IDL code, May–June 1990 (STScI files). The surviving fitting programs carry their original dates: a simplex optimiser and a PSF-fitting driver of 22–25 May 1990, two to five days after first light; a routine of 30 May converting the fitted coefficients, including a spherical-aberration term, to normalised aberrations; and a plotting script of 31 May holding the fitted coefficient sets. The dates match the memorandum of 30 May.
• NASA, The Hubble Space Telescope Optical Systems Failure Report (Allen Report), November 1990. States that computer simulation of the first-light images indicated 0.5 wave rms of spherical aberration at 547 nm in the telescope and not in the instruments: the value from Chris’s analysis, which the report does not attribute to anyone by name. Source →
• Burrows, Holtzman, Faber, Bely, Hasan, Lynds and Schroeder, ApJ 369, L21 (1991). The first refereed account of the aberration and its effect on imaging performance, with Chris as first author. The author list records how those involved allocated credit at the time: WF/PC team members who worked on the problem, including Sandra Faber, Jon Holtzman and Roger Lynds, appear as co-authors. Source →
• Krist and Burrows, Applied Optics 34, 4951 (1995). Phase-retrieval analysis of the pre- and post-repair images, including the aberrated first-generation cameras and the out-of-focus images Chris planned to confirm the spherical aberration. Gives the primary-mirror conic constant as −1.0144. Source →
• Robert A. Brown, NASA Hubble oral history (2016). Brown, co-chair of the HST Strategy Panel, says Chris was the first to know the problem was spherical aberration, and credits his diagnosis as the basis for the confidence in COSTAR. Source →
• Ray Villard, NASA Hubble oral history (2015). STScI’s news director recalls Chris recognising spherical aberration in the first released image and warning him the day before NASA’s announcement. Source →
• NASA press conference, 27 June 1990. NASA’s archival video is a shortened edit showing the first panel. Chris sat on the second, technical panel; the full recording has been requested from NASA. Source →
• Chris Burrows, oral history with Chris Gainor (2017). Chris’s own account of the fitting procedure, the Goddard meetings and the month of resistance; drawn on in Gainor’s NASA history Not Yet Imagined (2020).
• Other accounts. The WF/PC team’s recollections (Westphal’s 1998 Caltech oral history; Astronomy, June 2020) credit Roger Lynds with remarking on first-light day that the images looked spherically aberrated. Chris recalls that when he met Lynds at the WF/PC team’s facility at Bowie State, Lynds told him he did not know what it was. Lynds is a co-author of the 1991 paper, on which Chris is first author. The distinction the record supports is between a qualitative impression and the quantitative diagnosis: Chris’s phase-retrieval fit, the half-wave value, the memorandum of 30 May, and the defocus test he designed to confirm it.

From the archive

• HST Performance Monitoring and Assessment meeting #5 minutes, Goddard, 5 June 1990 (excerpt) →
• All original 1990 files as one zip (5 MB) →
• STScI Newsletter article, July 1990: HST’s imaging performance (TeX source) →

Papers

• 1991
  The imaging performance of the Hubble Space Telescope

  Burrows, C. J.; Holtzman, J. A.; Faber, S. M.; Bely, P. Y.; Hasan, H.; Lynds, C. R.; Schroeder, D.

  ApJ 369, L21

  ADSPDF
• 1995
  Phase-retrieval analysis of pre- and post-repair Hubble Space Telescope images

  Krist, J. E.; Burrows, C. J.

  Applied Optics 34(22), 4951

  DOI
• 1990
  Hubble Space Telescope Optical Telescope Assembly Handbook, v1.0

  Burrows, C.

  STScI; NASA-CR-189751

  NTRSPDF