I had a 1626 laying on my desk, the eTracer within reach and a boring movie playing on the PC. I had to do something while watching. I tried lowering the filament voltage on the 1626 to see if there is any pattern to emission. Below are the results.
NOTE: This is a new tube right out of the box. It would be interesting to see the same test with an older tube.
The table shows the different rp/gm values at each heater voltage. I adjusted the grid voltage to keep the 250V/20mA bias point and checked the rp/gm each time. The right columns shows the percentage in regards to the stock 12.6V.
As you can see the rp and gm stay quite stable right down to 8.0V and starts to fall off down to 7.0V after which is totally collapses. Yeah, 50% of the rated voltage!
It might be interesting to use 10V to light the filament and see if life can be extended, asthere is plenty of emission down to 8V. Although at that voltage, older tubes might already be affected.
I've added the curves for each 1.0V step.
This is the specified 12.6V as a reference.
At 7.0V everything still seems to be fine
At 6.5V we can see a slight rolling off starting to occur at higher currents
At 6.0V the lines start to collapse at 25mA or over. The area below that is still quite stable though.