You could probably increase the FB resistor to as much as 680K with little effect on the tone control action. This would effectively double the gain of the first half of V2. Since the plate load resistor for this tube (33K) is rather small, it also is helping to reduce the output impedance of this stage as well.
The basic power amplifier section of the HF-20 (which starts at the wiper of the loudness control) requires a level of 1.0 vac to achieve full rated power output. The second half of V2 actually operates at a slight loss when the tone controls are accurately centered, so all the gain required to raise a high level input source to the needed 1.0 vac to drive the power amp must be produced in the first half of V2. Accounting for a slight loss through the second half of V2, and based on the original sensitivity specification, the first half of V3 (from a high level input jack to the input of the tone control network) must be providing an overall gain of about 3.
Modern day input sensitivity levels have been reduced since the days the HF-20 was designed (1953), so that now, most equipment requires about 1/2 the signal the HF-20 does to produce full rated power output. Therefore, I have no doubt you have to advance the controls significantly for a reasonable level. But there is another component you can address to wring out a little more gain.
The 56K input isolation resistor connected to the top of the level control is also part of the FB network as well. Its primary purpose is to prevent the source impedance of the input signal from shorting out the FB around the first half of V2 when the level control is well advanced. Reducing it to about 22K will also effectively reduce the FB level by as much as one half as well, since it is the dominate element in the bottom ladder of the FB circuit.
If you raise the 330K FB resistor, and lower the 56K isolation resistor, you would effectively increase the gain for the first half of V2 by a factor of as much as 4 with the level control at maximum. This would bring the overall gain of the stage (as defined above) to about 12, which would increase the overall sensitivity of the unit to requiring just less than .10 vac for full power output. This may be too much compensation, so all you can do is see what values work best for your application. Even at this level however, rough calculations indicate that the input stage would still be operating with at least about 10 db NFB around it. This along with the low value plate load resistor should keep the driving impedance of this stage low enough to minimize any impact on the tone control networks. The fact that the level control still has an influence on the amount of FB around the input stage however, means it will still throw much of the gain in the upper level of the control's rotation. These changes may make enough impact however to make the settings more reasonable for you. Realize too that any increase in gain will also make a commencerate increase in noise as well, so be mindful of that as you make your adjustments.
The AC ground cap always presents problems in older equipment: Without it, hum will likely increase. With it, a shock hazard exists if the polarity of the plug is such that the chassis becomes hot if the cap fails. Even if it doesn't fail, it can cause lower AC potentials to exist on the chassis. Many will say to rip it out and install a modern green wire ground line cord to maximize safety, yet that often produces ground loops with other grounded equipment (cable boxes, etc.) to make the amplifier all but unusable. The best compromise I have found by far is to replace the cap with a modern safety rated cap, and install a polarized AC plug so that the cap is always connected to the neutral side of the line. A regular three wire line cord could accomplish this as well with the green wire simply disconnected.
Students of modern electronic technology will likely cringe at this thought, citing what if the power transformer shorts internally to the chassis. Likely they have never known of such an incident, always citing the theoretical possibility. It is there to be sure. But perils in life exist everyday. History shows you likely have more danger of being taken out by a runaway bus than this theoretical short ever happening. Enjoy your amps!
Dave