Dear Paul,
Great thanks for your reply!
Also found your old post from 2007 on audioasylum on this issue. I am posting it below to keep the info in one place. It may be interesting for other members, too.
"The DC gradient will affect the emission life along the filament. If there were no mechanical turn-on/turn-off stresses, no cathode stripping, no filament or grid warping, no mechanical stresses from moving, tube swapping, or shipping - i.e. no other causes of tube demise except cathode exhaustion, this would be a concern. But even then, the difference can be estimated. Cathode current is proportional to (Eb-mu*Ec) to the 3/2 power. Assuming 70 volts bias and 350v plate to cathode (fairly typical), a DC filament would be 67.5v one end and 72.5v at the other end. The currents would be about 43% different by my quick calculation. In other words, one end might fail 21.5% early, while the other end woudl last 21.5% longer compared with AC heating.
Other effects, for example variation in the uniformity of the initial cathode coating, are probably larger than that - certainly tube lifetime is highly variable between tubes, with a standard deviation usually larger than 21.5% even in controlled tests."
Best regards!
