Here's the Vacuum Fluorescent Display salvaged from the Canon P170-DH calculator driven by a Lattice iCE40-HX1K FPGA through my driver test board.
I had to make a last-minute change in the Verilog to get it to synthesize (Synplify Pro doesn't support initialization of multi-dimensional arrays) which resulted in displaying the high end of the 16-digit "Working Register" rather than the low, and in the opposite order of what I intended. But that's irrelevant to the test of the VFD driver board.
This picture was taken using the MAX253 and the Halo 2:1 transformer to drive the filament, and a color filter Noritake sent me along with the VFD modules I bought from them. Carefully holding the original calculator shell over the VFD demonstrates that the digits are easily readable with the lower filament voltage, even if I lower the 3.3V supply by 10%, so I'll probably use the Halo 2:1 transformer rather than the 4:3 to lower the current demands on the power supply. I'll probably go with the SN6505B though.
I also noticed that driving the filament with the 4:3 transformer causes the filament to glow faintly (hard to see unless you turn out the lights and the display is otherwise blank), which really isn't necessary for a VFD to operate and is a good indication the filament is being overdriven.
What's next? I may experiment further with a 100K pull-down resistor on a couple of grid drivers, just to see if it's really a problem. And I still have some work to do on the Verilog to allow me to fully test the display. But it's ever so satisfying to see something working!