Hubble Space Telescope

Sometime after launch on 4/24/1990, engineers testing images from the Hubble Space Telescope (HST) noticed that the images were blurry.  They were unable to focus the image through normal equipment adjustments.  This was the result of spherical aberration that occurred after the primary mirror was ground too flatly at the edges.  The mirror is ground based on data from a measuring device called a “reflective null corrector,” or RNC.  The RNC was out of position due to an error in setup.The error in the primary mirror went undiscovered.

This is primarily due to the fact that both NASA and Perkin Elmer relied solely on the RNC for measurements.  There were large gaps in Quality Assurance and Quality Control at Perkin Elmer.  The project was wrought with challenges and threatened with cancellation by NASA.  NASA also had gaps in their QA/QC process. 

Corrective Actions:

Note:  These corrective actions were originally suggested in the Allen Report, section 10.

1. Identify and mitigate risk.
2. Maintain good communication within the project.
3. Understand the accuracy of critical measurements.
4. Ensure clear assignment of responsibility.
5. Remember the mission during crisis.
6. Maintain rigorous documentation.

Background:

Sometime after launch on 4/24/1990, engineers testing images from the Hubble Space Telescope (HST) noticed that the images were blurry.  They were unable to focus the image through normal equipment adjustments.  The blurry images were determined through testing to be a form of spherical aberration.  Spherical aberration occurs when light reflected from the center of a mirror focuses at a different point than light reflected from the edges.  In order to understand the causes of the spherical aberration in the HST, a brief description of how the telescope works is in order.

The HST is a 2.4 m Ritchey-Chretien telescope with a focal ratio of f/24 and an optical range from 1,100 - 11,000 angstroms.  This type of telescope has two mirrors.  In the HST, the primary mirror is concave and 2.4 meters in diameter.  The secondary mirror is convex and much smaller at 0.3 meters in diameter.  It works similar to a complex billiards shot.  The light comes in from wherever the telescope is pointed, reflects off the primary mirror, then reflects again off the secondary mirror back through a hole in the center of the primary mirror to the focal plane where various instruments can process the image.  It may be easier to understand how this works by viewing this diagram:

Click for picture of Ritchey-Chretien Telescope

This style is very common and used in many terrestrial telescopes.  But in order to produce the image clarity required, each mirror must be produced to exact specifications.  These mirrors are ground in way similar to creating lenses for eyeglasses.  The main difference (besides scale) is the addition of a very thin layer of aluminum, which creates the reflective surface of the mirror.  Material is removed with a polishing system until the surface has the desired curvature. If too much or too little is removed during the process, the mirror will not focus properly.

When the test engineers discovered the spherical aberration, they needed to identify what was causing it.  After running several diagnostic tests, they determined that the problem was with the primary mirror.  This report examines both how the error in the primary mirror occurred along with why this error was not caught at any point until on-orbit testing.

Error in Primary Mirror:

In short, the primary mirror was ground too flatly at the edges.  Perkin Elmer Co. manufactured the primary mirror.  Perkin Elmer won the bid in large part due to their successful completion of a smaller proof-of-concept mirror.  The smaller mirror was constructed perfectly.  The idea was to then simply scale up the process to make the larger mirror.

In order to achieve the proper lens specifications, Perkin Elmer required a measuring device that could mimic the desired final image perfectly.  This device is called a Reflective Null Corrector (RNC).  The RNC uses mirrors to determine how much material needed to be removed from the primary mirror.  The process was to grind, test, and repeat until perfection was achieved.  See this image:

Click for picture of Hubble Primary Mirror Polishing

The man in this image does not realize it, but the parameters controlling the massive grinding apparatus are incorrect.  This is because the RNC measuring device was positioned 1.3 mm to far away from the mirror.  This may not seem like a lot, but given the extremely tight tolerances required, it was enough to jeopardize the usability of the HST.

The RNC was out of position due to an error in setup.  In order for the RNC to function properly, it must be perfectly positioned above the mirror.  When Perkin Elmer built their prototype smaller mirror, they achieved this positioning by using a measuring rod cut to an exact length.  This rod was fitted with cap on the end (called a “field cap”) that allowed engineers to achieve perfect alignment between the RNC and the primary mirror.

The end of the rod is reflective.  Engineers reflect light off of the end of the rod to set the distance of the RNC.  The problem was that they bounced the light off of the surface of the field cap instead of the end of the rod.  A small amount of anti-reflective paint had chipped away from the end cap, exposing a reflective surface.  The width of the field cap was 1.3mm – hence the reason the RNC was positioned 1.3mm too high off of the mirror.

Error Went Undiscovered:

The error in the primary mirror went undiscovered.  This is primarily due to the fact that both NASA and Perkin Elmer relied solely on the RNC for measurements.  They had complete trust in the RNC to properly and accurately direct the grinding process.  This was because they supposedly “certified” the accuracy of the RNC (there is actually no documentation supporting how this was completed, or even that the certification was done).  Other measuring devices were available and used.  These were the Refractive Null Corrector (RvNC) and the Inverse Null Corrector (INC).  Neither of these devices was as accurate as the RNC.  Yet, the error was so great that they both actually were able to measure the spherical aberration in the primary mirror.  However their measurements were discounted in deference to the supposedly “certified” RNC.

There were large gaps in Quality Assurance and Quality Control at Perkin Elmer.  Quality was not allowed access to the metrology lab.  And Quality reported directly to Project Management.  This set up a conflict of interest between those trying to employ quality management systems and those trying to build the mirror. 

The working environment in which this all took place was toxic.  The project was wrought with challenges and threatened with cancellation by NASA.  This created a stressful production environment in which QA/QC was considered an additional and unnecessary cost in light of the fact that the proof-of-concept project went so smoothly. 

NASA also had gaps in their QA/QC process.  The Defense Contractor Administration Service (now the Defense Contractor Management Agency) did not exist when the primary mirror was being built.  Prior to that, NASA personnel accepted the Perkin Elmer process as satisfactory.

Final Testing:

There was no end-to-end final testing completed on the fully assembled OTA.  This is because such a test would actually require two mirrors larger than the primary mirror, built to the same exacting standards.  This would cost more than the telescope itself.  Therefore, each of the components was tested individually.  Final measurements on the primary mirror were actually extremely exact – even tighter than the specification required.  The problem was that these final measurements were being compared to a standard that came from the flawed setup of the RNC.

Learn more about Sologic RCA