Fantastic work Josiah. What was the source of the brown glass? I suspect the darker glass absorbed more solar energy as anything darker tends to, but different types of glass also melt at different temperatures. Heck even different sizes of chunks melt at significantly different temperatures. Sand sized glass bits apparently insulate themselves from the sun and don't melt as well as bigger chunks. The formula I found from folks who do this all the time that worked with the rocket glass melter was bring it up above 1600 degrees for half an hour plus so it will melt uniformly, then let it cool down very slowly (anneal) so the glass cools evenly and doesn't crack along temperature fault lines. For anything over 1/8" thick this process takes at least 6 hours according to what I've read and according to what I got to work with a rocket heater.
I've attached here a firing graph from a glass industry paper showing how long to keep the glass at what temperature in order to have it fire properly. The attached pdf is the entire document in case anyone wants more technical data.
I wish I had found this before I started putting any time into the first version of the solar melter from the ATC that only melted a 1/4" circle of glass. Josiah did a lot better, but it also kept cracking. I think the details in this industry paper point the direction needed to get a better result next time, both in telling us what temps we need for how long and telling us how fine our control of our machine needs to be in order to have it work for thicker and larger panels. Reliably getting a properly annealed tile of glass much bigger than a quarter I think is really a stretch for this approach, but not impossible. Below are my takeaways. I invite comment on this list, especially from Josiah and any other folks following this thread who actually melt glass on a regular basis.
The cracking of the cooling glass that we observed (both in my version and Josiah's version) are consistent with the industry technical statements that the glass needs to be uniformly and consistently heated over its entire surface and down through its entire body throughout the entire process, not spot heated, and not heated by a moving spot.
Our current lens focuses enough energy to produce a 1/4" spot of 2000 degree Fahrenheit according to specs. That is sufficient to provide the 1600 degrees necessary to melt that spot about 1/8" deep.
If we unfocus the beam to uniformly cover a 2"x2" tile we are applying the energy at 1/16th its original concentration, which is not enough to melt the 2x2 tile.
Josiah's insulated chamber improves performance, but not 16 times as much. Maybe 4x based on reliably producing quarter-sized medallions that didn't crack. Josiah did the reflector help?
A glass lid of any sort appears to impede solar gain more than it helps by trapping residual heat. Further conversations with glass experts and Josiah's tests have convinced me that even the cleanest uncoated glass of any sort absorbs or reflects too much energy. Someone else may find a better glass for this, but even common window glass which is 99.99something optically pure provides too much impedance.
A bigger fresnel lens (say a free one with 4x the surface area from a bigger donor TV) I think would take us from iffy melty to consistently melty temps on a 2x2 tile, maybe even bigger.
Adding thermal mass within the insulated envelope could help but only if our lens is big enough to focus enough BTUs to melt the glass uniformly AND heat the thermal mass all at once. Without a lot more BTUs I think additional mass only keeps the glass too cool to uniformly melt. We also don't gain anything from attempting to preheat additional mass (with solar) since the firing takes most of a solar day.
Automated tracking should make a significant difference in giving us the longer exposure to high temperatures we need to melt the glass completely and uniformly and give it the many hours of BTU input it needs to anneal. 6 hours plus at 900 degrees plus throughout its body for anything more than 1/8" thick is what the book says.