Avo Valve Tester

[Mike Watterson]

"If there is no separate g2 HT, then it's only testing pentodes as triodes (g2 connected to anode)."

Your statement above, indicates you are getting confused, not me!     I never said there was no G2 (Screen grid) supply, because there is, and I set it correctly!      There is no G3 (Suppressor grid) provision in the tester.     The valve-holder connection pins are:
Filament, Filament, Control grid (G1), Screen grid (G2) and Anode (A).

As I said, I am as aware as anyone that G3 should be connected to the cathode, and that is what I did initially, but the resultant readings were way out.   As soon as I removed the link, they showed correct.    Not just on one valve, but a number of them!

No, I just haven't looked at the manual. There are many old testers, even from 1950s, that only triodise the valve, no separate g2 HT.

I'm not confused at all.
I've not ever seen G3 supply on a tester, because it's ALWAYS connected to Cathode, Anode, G2 or Filament. It's never floating and on types not intended for RF it's usually internally connected to the cathode. Always the cathode or if a direct type, the filament, if it's a Pentode mode test. The entire point of a Pentode vs a Tetrode is that is has a third suppressor electrode (need not be a grid, can be a pair of pins or plates) that must be below anode lowest voltage or you get secondary emission, a current flow from anode to Screen (g2) which causes the  Tetrode Kink. Beam Tetrodes or Kinkless Tetrodes are really Pentodes too (named differently for patent reasons) when they realised an electrostatic lenses does same thing as the third very open screen.
   

G3 (suppressor) has to be connected, not floating, wrong readings means connections are wrong, or something else is wrong.

   

Shown innards are from an EF89. You can see that the electrons are really two opposing beams from centre and thus the outer most open "grid" can be replaced by two pairs of plates, one pair between the each of the two anodes and the g2 (sceeen) grid.

So if you are only getting what seems like reasonable readings on EF39 with the cathode floating, then something is very very wrong. Don't shrug and ignore it. Double check connections etc.