It has a drive level on the front which points to a grid driven situation but the finals are grounded grid. I have never seen one set up like this one.
Inspection of the schematic reveals the drive control is a mislabeled, screen voltage adjustment. Lowering the positive screen voltage on the tubes has a similar effect as increasing the negative bias voltage on the control grid and pushes the tubes closer to class C, rather than linear operation. Use on SSB with the "drive" control turned down, would show the nasty class C distortion, where AM would hide it and make the knob appear to function as a carrier control.
Inspection of that screen voltage supply shows some really bad design flaws. The load that a screen element places on the supply has some very unique characteristics. Did you know under hard drive conditions the screen can actually switch from placing any load on the supply and act like a source, by reversing the current flow and forcing it to flow backwards into the power supply? This condition is known as "Negative Screen Current" and is caused by secondary screen emissions, within the tube.
To stabilize the screen grid voltage in an amplifier like this, requires that the power supply have a very low internal impedance, in comparison to any load placed on it AND be able to "sink" any current the tube may be able to develop, if it's pushed into a condition where negative screen current occurs. Alternately, a shunt regulator can be added to the output of the screen supply to insure the most stable voltage under all operating conditions.
The Eagle tried everything from adding shunt resistance, to neon bulbs that ran completely over rated for the job. There is evidence the builder knew this flaw existed with two neon bulbs being placed in parallel. One burned up within the first 30 second high power key and two got it down the road, "just long enough".
Because the neon bulb has such a small internal element surface area, it is unable to sustain more than a few milliamps of current. It was never intended for use as a voltage constant, like a VR tube with much larger cylindrical elements. It was designed as an indicator bulb. Although, the neon bulb could be pressed into VR service if the required current was miniscule and FAR less than these screens place on them.
Many sweep tube amps that use "Micky Mouse" methods to manipulate the screen voltage, suffer from unstable output that tends to increase, the longer the amp is keyed. Most often noticed in the AM carrier, that keeps creeping upwards until the tubes cool down and secondary emissions stop flowing.
When this happens, the screen voltage will actually be larger than the output of the supply they are connected to! That causes the DC bias current to rise and contributes to premature tube life. In most cases all of that "power drift", disappears as soon as you ground the screen, suppressor (beam deflection plates) and control grid, treat it like a triode and use simple cathode bias.
Reliable sweep tube circuits can be found in old "Amateur Radio Handbooks". One article in the 1980 handbook, outlines a "Quarter kilowatt from 6KD6 tubes".