a question of tube behaviour

[g-banjo]

As the owner of a Vox Tonelab amp modeler I've been experimenting with various valves as substitutes for the stock EH 12AX7.
The Tonelab uses the 12AX7 in it's "valve reactor circuit", I'll quote the Vox website:

" The Valve Reactor circuit uses a 12AX7/ECC83 dual triode vacuum tube, a virtual output transformer and a dummy speaker circuit that simulates impedance changes of a real speaker. In other words though the output power (level) is low, it has the same power amp circuit as an actual full-tube amp, and faithfully reproduces the response of the power amp circuits of the original amp that is being modeled. It reconfigures itself so that its characteristics are the same as the amps it's modeling (class A, class AB, negative feedback circuit, etc.) This means that all of the nuances of the original amp model including sound, feel, distortion and presence are reproduced."

I have been comparing various 12AX7s and 12AT7s both in this unit and in a DBX preamp. The differences between the valves seem to be greater in the Vox than in the DBX. My questions are;
why do the differences between individual valves seem more pronounced when run at lower power? Or am I just imagining this?

[Tom Bavis]

I suspect that the Vox runs the tube at at the edge of linear operation - near saturation or cutoff to produce a particular tone - distortion, really. So different tubes may react quite differently. In a preamp (especially if negative feedback is used) there may be little difference if the tube stages remain in their linear region.

[g-banjo]

But it's actually on clean sounds, with no distortion, that the differences in dynamic response is most obvious to me. Does that fit?