How to do a perfect large print bed?
The “right” way is to use a thick aluminum plate (best heat transfer) with a milled-flat (dead-flat) top surface, and A/C driven massive heater (to temperature quickly), with enough time to allow any thermal settling.
Could be enough by itself. But we want more.
A thin spring-steel plate, magnetically attached, with some sort of optimal coating, both tends to even the heat distribution and allow detachable plates.
The above is certainly the best we of present know. Also a bit expensive.
(Ignoring printer-belts, for the moment.)
Milled-flat surface is expensive. Thick aluminum is expensive.
Glass can give us a good cheap flat surface. Good choice for cheap unheated print beds. Not good for heated beds, as poor thermal transfer. Also regular glass tends to distort (especially if heated fast). Could use borosilicate glass as it does not distort, but costs more, and still slow to heat.
Borosilicate glass on a thick aluminum bed is ideal, if we do not care about cost or time. If we care about cost (or time), then the equation changes.
Someday we will have dead-flat heated print-beds, made with lightweight stamped frames, and tweaked flat by robot jig. That needs a larger market.
Visiting this as I bought a cheap-ass Chinese large Core-XY printer (a TronXY X5S), which comes with a pathetic print-bed. Thin aluminum that sagged under the weight of the bed. Insulated and braced the (thin) aluminum bed, so near-flat, and found the plastic(!) sheet topping the bed had a crown.
All that said, the same printer is laying down a pretty nice Benchy, at present.
At 60mm/s. Child’s play - kind’a.
Might be in future near dead-flat beds are a given. In present, printers must adapt to less perfect beds.
As a software guy, going to invoke Butler Lampson. The end-to-end check matters most. With a 3D printer, the end-to-end check is when the print head touches the bed, at any position. Given present beds fabricated at imperfect level, we need software to sense and make up for the cheap/imperfect hardware.
Recent Marlin has some crude support for a levelling mesh, and test.
Interested in direct physical tests. We care most about when the print nozzle comes into physical contact with the bed. Anything else is indirect.
What is best present practice?