Jump to content


* * * * -

My Other Telescope is an 8.4 Meter - Part III: Polishing

Discuss this article in our forums

My Other Telescope is an 8.4 Meter

By:  Patrick Stevenson (Gork)

Part III - Polishing


Steward Observatory Mirror Lab, Polishing.  That's where I spent eight of the best working years of my life.  I've started writing about this aspect of fabricating the world's largest monolithic mirrors a number of times.  Each time I get to about ten pages before I realize how truly complex this phase of fabrication is ( I know you aren't supposed to end a sentence with a preposition, but I was a Psych major, not English).  It always begins with moving a mirror into the Polishing Lab and finding just that operation being so fraught with danger and excitement that I get buried in details.  So, I am going to assume, for the sake of this account, that everything works as planned and there are no side tracks.  That won't be true, but if I tell the real story, I'll never get done.

After the giant mirror has been cast, it is moved into the Polishing Lab on a twenty ton air-cushion  apparatus that looks like a giant “U”.  Jack screws driven by air pressure lift the mirror off stationary jack-stands.  The mirror has been mounted in a huge “cell” that supports the weight of the mirror evenly because the mirror would shatter under its own weight if it were not evenly supported.  Internal air pressure is carefully regulated to keep the mirror from “flexing” during the next step in processing.  Once the mirror and cell are clear of the stationary jack stands, the huge air cart is lifted off the floor on a thin cushion of air.  There is a small tractor attached to the bottom of the “U” that is piloted by a Technician who guides the lifted sixty ton mass of air cart, mirror cell, and mirror into the Polishing Lab.


(mirror in polishing cell)
(floor view of mirror in polishing cell with air cart underneath)


The door between the Casting area and Polishing Lab is barely large enough for the mirror to pass through.  With only inches between the huge mass and the edges of the door, frequent impacts occur.  Even with a number of ground guides the mass will “bump” the edges of the doorway or other stationary building structures.  It's amazing that a mass like that can gently bump something and not destroy something in the process.  The mirror is backed into the Polishing Lab and then placed on a six foot wide rotating table at the south end of the room.  This operation is incredibly complex when you think about it, but it is routine for the Techs that work the room.

The mirror, in its cell, is settled onto the table where six mating tapered bronze blocks support the weight.  The air cart stays parked under the mirror cell because it will be used to lift and lower the mirror a number of times while the mirror is centered and leveled on the table to within a thousandths of an inch on each axis.  The adjustments to the bronze blocks that will move the mirror only a few thousandths of an inch are the result of Trigonometry calculations that predict the movement based on tiny changes in the locations of the bronze blocks.  It can take up to a week to get the mirror stabilized for the first operation in the Polishing Lab.


(top view of table on LPM)


The mirror is brought in to the Polishing Lab mounted face down for processing of the back plate.  Internal stresses in the mirror as a result of the Casting operation will be relieved by milling a few thousandths of an inch of glass off the back plate.  The back plate is then ground to a flat surface over four hundred square feet.  Upward of a thousand pounds of glass are milled off the back plate using a diamond studded wheel that spins at about three thousand rpm.  In essence, this is simply a huge milling machine.  Since there is a “Hard” contact between the milling wheel and the mirror any malfunction has the potential to destroy the mirror.  The thought of twenty tons of glass shattering under pressure is not something you allow yourself to think about or you simply wouldn't proceed.


(grinding lap on back plate)


After the back plate has been processed the mirror is moved, via the air cart, back to the Casting Lab where a giant steel ring is employed to turn the mirror over so the face plate is facing up.  The movement of the mirror back to the Polishing Lab is pretty much the same as before.  Again, the mirror and cell are settled on the rotating table, leveled and centered to within a thousandths of an inch.  The parabolic curve resulting from the rotating Casting Furnace has brought the face plate about 90% of the way to the desired shape.  About three thousand pounds of glass will be milled from the surface to reach the desired parabolic shape.


(steel ring used to flip mirror)


The face plate will be milled, ground, and then polished to a nearly perfectly smooth surface shaped precisely as specified.  The final grinding and polishing take place on the LPM (Large Polishing Machine) located at the north end of the Lab.  Although previous mirrors have been completed on the LOG (Large Optical Generator), the LPM has more precise controls that make the final polishing of the mirror surface easier (if that is possible) than the LOG.


(side view of Stressed Lap in action)


The precise surface shape of the mirror is achieved by utilizing the Stressed Lap.  The Stressed Lap is a tool comprised of a six foot diameter aluminum plate about two inches thick.  There are a series of steel “towers” mounted around the edge of the aluminum plate.  The towers are connected via steel bands to another tower on the opposite side of the Lap.  Tightening and loosening tension on the bands actually bend the aluminum plate angstroms to match the exact location of the Lap on the mirror surface.  Computer tracking of the exact location of the Lap on the mirror surface dictate the exact curvature of the Lap to match the prescription of the finished surface.


(note steel towers on Stressed Lap red, white, blue)


Throughout processing, the mirror is tested over and over to ensure that the desired shape is being achieved.  This is done by moving the mirror onto the Test Tower located in the center of the Polishing Lab.  The tower can be seen from outside the building as an extension of the top story up to about ninety feet from the floor.  The entire tower is supported by giant air bags beneath the floor.  Controlling air pressure in the air bags will lift the mirror and insulate it from vibrations present in the floor.  Testing is so precise that, were the mirror not supported by the air bags, a truck passing the Lab out on sixth street would produce enough vibration so as to nullify test results.


(side view of mirror on Test Tower)
(top view of mirror on Test Tower)


The surface of the mirror is reflective enough to return laser light projected onto the surface from the top of the test tower.  The reflective image is compared to a miniature “perfectly” shaped version of the desired finished product.  Adjustments are made to the grinding or polishing to keep the shape on the right track to hopefully finish at a perfect shape.  If the surface of the mirror were as large as the United States, the worst error in the surface would be only one half an inch in height.

Again, if everything works as planned (never) we end up with a perfectly polished surface perfectly shaped as specified.  What can happen?  We make a mistake during a test and we grind or polish the surface to the wrong shape.  Correcting an error like this can add months to processing.  A mechanical failure can occur that scratches or scrapes the surface.  Whatever the anomaly, the surface must be corrected via grinding or polishing.  Of note, if the anomaly results in a “low” spot the entire mirror must be ground or polished down to that level.  The closer we get to the desired shape, the more often we test.  Getting to the final polishing step using the finest polishing medium (cerium) brings to bear constant vigilance, testing, retesting, verifying data, double and triple checking corrective measures.  One of our mirrors had an anomaly that caused us to go back to coarse grinding to correct the error.  It took three months to fix that one.


(Tech horsing around getting IR photo on Test Tower)


Mechanical malfunctions that result in crazing, scratches, and fractures in the surface must be repaired to relieve surface stresses introduced by the damage.  The damaged area is removed using Dremel tools that cause the surface to look like glass-eating termites got loose on the mirror.  The fractures are, hopefully, no more that a few mils deep.  Once the fractures are removed to the point that there are no visual remnants, the groove is then etched using  a 20% solution of Hydroflouric Acid.  This is the most dangerous and caustic acid in existence.  There is an incident well known about an Australian Technician that spilled 20% HF on his thigh in a spot no larger than a quarter.  He immediately flooded the area with DI water.  In spite of his immediate actions, he was dead in three days.  The HF does not cause a surface burn, but instead cruises through the skin and attacks deeper tissues at an alarming rate.  The typical medical response is to immediately amputate the limb hopefully above the HF damage.


(Fractures caused by broken glass between mirror and grinding tool)


Only a very few of us Techs are qualified to use the HF process.  The fatal dose of exposure is measured in PPM (Parts per million).  In 2004, while repairing the surface of a mirror I received a dosage via inhalation that was barely strong enough to catch the faintest smell of the acid.  The result, obviously was not fatal, but I still have a chronic sinus headache to this day.  Generally, if you smell the acid, you have been exposed to a fatal dosage.

During the final stages of “figuring” the surface we will locate exact positions needing attention.  An Engineer will calculate the exact action to be taken to fix the anomaly.  This operation is so precise that a Tech will gingerly move onto the mirror surface to the spot identified and using a tiny polishing pad, take as little as two or three strokes only inches long, and then a retest.  This may go on for weeks before the Engineers are satisfied with the perfection of the mirror surface.  Only when the surface can be certified as “perfect” will the mirror be prepared to move to the last operation, Integration, where the mirror will be mated to the final supportive cell that will be home for the mirror's lifetime.


(final exact surface correction on LPM)


(Compliments to Howard Lester who edited this draft)

Stand by for Part IV, Integration

  • gordtulloch, SometimesKen, Rollo and 22 others like this


I just wanted to say how much I've enjoyed this series of articles. I've been visiting the site nearly daily looking for the next installment. Many thanks.

    • Patrik Iver, Don Taylor, TSSClay and 1 other like this

Thanks for your message.  I've really enjoyed writing these articles.  It not only brings back old memories but it makes me realize how complex the process really was and how much fun we had.  I envy the folks working there now and I hope they have some sense about what they are accomplishing.

    • Don Taylor likes this

This is fantastic reading... hanging on every sentence.  Super High-Tech but where the human element is so important.  No place here for bureaucratic bunglers!




    • XB-36 likes this

Thanks Rick.  It was an exciting job that I miss greatly.  It was a highly technical job, and politics just would not survive.

    • XB-36 likes this

You've captured your years at the lab quite well Pat.  Maybe there's a writer in you after all!  I see multiple mirrors in some of the images.  I may have missed you referencing it that but how many were going at the same time?

    • XB-36 likes this

Remember those old plastic puzzles that had little squares that you would move around that had one empty space?  The Mirror Lab was just like that.  There were times when we had only one mirror in the que, and there were others when we had up to four or five.  Moving them around to make room for processing was a real challenge.  There was no way to store a mirror outside the Lab, so we had to make room inside.  Not all of our mirrors were obtained by buyers as finished mirrors.  For instance, we had one 6.5 meter mirror that was purchased by a foreign government as a cast-only mirror.  They apparently thought they could process the mirror blank.  Since we had contracts for finished mirrors, that one tended to get pushed back in the Casting que.  We were working on finishing the LBT#2 mirror when we got an order from an American corporation for a 6.5 meter mirror that had specifications beyond what we had ever done for an astronomical mirror.  At about the same time we got an order for the GMT center mirror and the LSST primary and secondary mirrors.  For awhile, we were spending almost as much time moving mirrors around as we were processing!  My memory is a bit hazy, but I think the most we ever had processing at one time was five.  That included the Integration Lab which was about the same size as both Casting and Polishing.  That would put one mirror being integrated to its cell, three in the Polishing Lab, and one being cast.  We also had to cast and polish a 5 meter mirror that was ultimately hung face-down up in the test tower that would be used for testing the off-axis mirrors for GMT.  We were all a bit nervous about working ninety feet below a suspended 5 meter mirror!  They finally installed a steel roll-out safety door beneath the mirror that remained closed except during testing.  These mirrors were massive enough that the structural integrity of them could only be calculated.  No actual testing was ever done for obvious reasons.  The safety shield was as much for genuine safety as it was to provide a work environment that we Techs could live with.  It really would take a book to fully describe how that Mirror Lab functions.  Busy days!

    • PirateMike, happylimpet and Digipainter like this

I made a mirror like that in my garage once.  But it took several years to get it right.  The Stewart process is better, I think.

Oh, wait, unit error.  My mirror was 8.4 in in diameter, not 8.4 m.  Oops.


Hydrofluoric acid is nasty stuff.  It is also used in one process in oil refineries (alkylation, a process to convert low-weight hydrocarbons into higher-weight hydrocarbons for high octane gasoline).  I didn't work in the alkylation unit often, but when I did I had to wear a full PVC suit, gloves and boots, goggles and face shield.  And that was in an area where the bad stuff is supposed to stay in the pipes, we never handled the acid directly.  First aid is now calcium gluconate, but I understand even that can't help if somebody is exposed to too much HF.

    • tdeclue and Sporocyte like this

Great installment, keep them coming!

Thanks for all the comments and suggestions.  I'm working on the last installment now.  I need to get back over to the Lab for some more pictures of the Integration process.


    • Patrik Iver likes this

Howinhell do ya grab an 8.4m hunk of glass w a steel band around it to flip it? What kept it from slipping out of the band and pray tell how was the assy. grabbed to flip it??   Scary

Cloudy Nights’ version of a hit series, I have been thoroughly enjoying reading each one multiple times. 


Hydrofluoric acid is nasty stuff.  It is also used in one process in oil refineries (alkylation, a process to convert low-weight hydrocarbons into higher-weight hydrocarbons for high octane gasoline). ...

Placid in BR?

    • XB-36 likes this

I've walked on one of your mirrors.  All over it, stomping around, waving my arms...


I was with Steward Observatory Mountain Ops, and we were washing LBT #2 before aluminization.  The only way to do it is to climb on up there...


I also helped build the LBT thin shell adaptive secondary, I have a good size chunk of glass hung on my office wall from one of the failed attempts.

    • XB-36 likes this

Looking at Wikipedia it looks like there have been nine mirrors made in this class (8+ meter mirrors). Are there more "on order" (planned)?

Hydrofluoric acid is nasty stuff. 



People tend to think that HF is dangerous because it is an acid. It is dangerous because it is a poison. Though it is corrosive its toxicity is not closely related to that. Being a small soluble molecule HF penetrates all tissues rapidly, and removes calcium from cellular availability, which is crucial for nerve and muscle function.


HF poisoning killed (directly or indirectly) 25% of Iceland's population in 1783-1784 when the Laki volcano emitted millions of tons of the stuff.

    • Starsareus likes this


Sorry I took so long to answer your question.  The process is very complicated and would take a book to cover the details so I just kinda skipped over some stuff.  Flipping mirrors.  Good question.  There were two ways to pick up a mirror and move it to new fixtures.  During the early stages there is a large steel "spider" with big pads fastened to the bottom.  Those pads were bonded to the mirror surface using a very strong silicone based adhesive.  The spider was then lifted, along with the mirror, using an overhead crane.  The whole thing was lower into the Polishing Cell.  Once secured the spider was unhooked from the pads and moved back to storage.  All of the pads were removed using piano wire saws.  That was a **** to put it mildly.  We spent a couple of days see-sawing each pad.  We then removed the residual adhesive and began grinding the back surface.  When the back was done we brought the spider back out, only this time we used vacuum pads.  Each pad had its own dedicated power supply in case there was a failure.  Then we would only lose one pad and not the whole mirror.  The crane lifted the mirror and set it down inside the steel ring where the spider was attached to the ring (which was horizontal).  Huge hydraulic rams were used to tilt the structure.  Once it was rotated 180 degrees the mirror was again lifted via a spider and pads.  The mirror was lifted and set back into the Polishing Cell so we could start on the front surface.  Any time a mirror was being lifted all personnel not involved with the process were evacuated.  Dropping a mirror that size would result in an explosion that would likely kill anyone in the room.  Great question!

Haha, that photo of lying down on the mirror cell.


I'm looking forward to the next part.

    • ericthemantis likes this
Jun 23 2019 12:40 PM


Thanks for your write-ups.  They have brought back a lot of memories.  As a professor at OSC, I was in the mirror lab a number of times and I spent a fair amount of time over there installing interferometers when I was with 4D Technology.  Are you guys still using all of those PhaseCams that we supplied for testing those big mirrors?



Wile E. Quixote
Jun 27 2019 06:48 PM

This is a very cool series. Thank you for writing it. I'm looking forward to the next part. I was wondering if anyone has ever cast a large telescope mirror using metal. A large metal casting would be heavier than a glass casting but it would also be stronger. The biggest disadvantage I can think of is that metal is a better conductor of heat than glass is. Would this make it impossible to create a mirror that met the tolerances you need for a precision astronomical instrument?

Patrick, thoroughly enjoying reading this series,  can’t wait for the next installment.  Reading through this installment got me to thinking, and it may sound like a dum question but...  When you’ve reached polishing stage, was there ever a case where introduced errors in the surface, ever resulted in you having to discard the entire mirror and start from scratch?



Using stuff like Imron paint requires full hazmat suits and respirators.  Why would someone be using acid dressed in normal clothes?

Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics