by: John M. Hill
As we approached room temperature over the last few days we were able
to peek in under the lid and see the glass surface with our own eyes.
Over the weekend I thought perhaps I saw an image of Jesus in the
reflections from the ripples on the surface. However looking again on
Tuesday morning revealed that this was actually an image of coach Lute
Olson! (OK --- this is an April Fool story, but the University of
Arizona men's basketball team really did win its first ever national
championship on Monday evening March 31 after defeating U. Kansas,
U. North Carolina and U. Kentucky.) Go Cats!
The faceplate thickness in the center of the mirror is about 32 mm,
which is what we expected from looking at the video images during the
casting. This is a useable thickness since the blank gets ground and
polished down to a faceplate thickness of 27 -- 28 mm during optical
finishing. (The faceplate is the continuous layer of glass formed on
top of the honeycomb structure.) Over most of the mirror surface, the
thickness of the faceplate ranges between 20 and 32 mm. This reduced
thickness could still be successfully polished into a good mirror.
The problem is that about 4% of the mirror surface (at the outer edge
near azimuth 60 degrees) has a faceplate thickness of only a few
millimeters which makes that piece of the mirror unpolishable. The
largest leak was near psuedo-azimuth 60 degrees, although other
similar leaks occurred at several places around the perimeter of the
mold.
Based on half a day of inspection, we believe that the glass leak was
at the joint between the vertical tub walls and the outer edge of the
tiles that make up the base of the mold. Apparently the hydrostatic
forces from the liquid glass caused several of the tub wall segments
to shift and open this joint to allow a glass leak. The tub works
like a barrel with 48 silicon carbide tub wall sections constrained by
a set of Inconel 601 bands. We pull on the bands from outside the
furnace with a predetermined force to hold the tub walls together
against the pressure of the liquid glass on the inside. Determination
of the exact cause of the leaks will require considerably more
measurements and engineering analysis. We speculate that friction
forces between the restraining Inconel bands and the tub walls or
friction forces between different sets of bands caused these gaps to
open during the heating phase of the casting. The gaps have closed
during the cooling phase, presumably because the bands contract more
rapidly than the silicon carbide tub walls and the friction forces are
effectively reversed.
While this remelting of additional glass onto the surface was not
previously part of our Mirror Lab repertoire, laboratory tests and
calculations give us confidence that the remelting onto the faceplate
will be successful. (We and others have repaired smaller blanks by
reheating or remelting, but this is our first time for adding
additional glass to a blank.)
The basic idea is that you heat the whole mold structure slowly back
up to 650 or 700 degrees with chunks of glass piled on top. This
heating process takes several weeks so as not to damage the intact
honeycomb structure. Then we flash heat the upper surface up to 1180
degrees with 400 kW of power from the lid heaters to melt the new
chunks of glass onto the surface. Because only the upper portion of
the mirror is heated in this process, residual leaks at the bottom of
the mold should not be a problem. Also the viscosity is high at the
bottom of the mirror so there is not a significant buoyant force
trying to float the cores during the remelt. This melting only takes
a few hours and is followed by the standard three month cooling and
annealing process.
The additional four months in the furnace does not delay the
completion of the mirror as we have some months of slack between the
casting and the polishing schedules. This 8.4 m mirror for LBT should
have the mold cleaned out before the polishing of Magellan is
completed in Spring 1998.
8.4 meter mirror blank revealed!
After nearly three months of melting and annealing and cooling, the
first 8.4 meter casting has finally reached room temperature.The good news
We removed the lid of the furnace on Wednesday evening April 2.
The mirror blank looks very pretty. There's nothing like the sight of
an F/1.14 parabola to cheer your day. No cores floated out of position
and the glass honeycomb structure appears to be sound. There are very
few bubbles on the surface, and the ribs look like the highest quality
that we have ever cast.The bad news
The leak we noticed during the casting process was larger than we
originally thought. After a preliminary inspection, we estimate that
almost 3 tons of glass leaked out of the mold and onto the floor of the
furnace. This is not such a big problem except that we only put in
two tons of "extra glass" to account for such leaks. The result is
that the faceplate is thinner than the casting target value of 36 mm.More good news
Because we've been worried about this possibility of a leak since the
high temperature part of the casting, we have developed a procedure to
remelt some additional glass onto the surface of the blank. We are
"topping up" the pot so to speak by adding a couple tons of extra
glass to the mirror. This should restore the faceplate to its desired
36 mm thickness without disturbing the honeycomb structure underneath.
Back to LBT Casting Updates