Tuner filament supply question

[rdf]

This one's a bit out there. Shown below is the AC filament supply for a Scott LT-110B tuner, labeled by functional block. The MPX/Audio section has been modified the typical way, disconnecting the junction point of R307 and R308 from ground and biasing the filaments up fifty odd volts DC.

Scott went to a great deal more trouble than a pair of balancing resistors on the front end and IF sections. Each filament's return is grounded and each supply filtered with uH chokes and ceramic caps. Ignoring the dial bulbs, would the same biasing trick work here? Part of the intent appear to be isolation between tubes, however DC biasing, by effectively eliminating leakage between heater and cathode, promises to get rid of the requirement. Given the component density this isn't a 'try it and see' thing. Anyone courageous enough to hazard a guess?

[jnb]

If the chokes are, say, 3mH and the caps are 5nF then they'd work above 40kHz. My first thoughts were that they'd add some filtering to RF in the RF sections.

Firstly WRT DC, RF can still appear on the heater lines and I would still regard the filtering as useful.

WRT to floating the heaters to reduce emission to the cathodes, RF can still get through via capacitance. Hum, however may be reduced as with whatever AF gets radiated onto the heater wiring.

Is there any such filtering for the audio section? I am curious as to whether lifting the grounds and common mode filtering the heaters would be an improvement.

[rdf]

Thx jnb. You're right about it targeting RF (the chokes are 1 uH!) From the topology another intent seems to be RF isolation between heaters. The audio section has no filament filtering, following the normal practice of decent gear of the time by balancing to ground via resistors. From my reading the coupling mechanism between cathode and filament isn't capacitive at audio frequencies. While unintuitive, from one perspective it makes perfect sense. DC bias has no effect on a regular capacitor's ability to pass AC. If the coupling mechanism between cathode and filament was simply capacitive, DC bias would have no effect on hum here either. I recall it being ascribed to emmission without any further explanation. Not very useful.

So the big questions are, does capacitance become dominant at RF frequencies, is the local reference to ground important and was Scott doing more than just filtering with these networks? You guessed what I want to try, common mode filtering along with DC bias. On further thought, I think you may be right here. The very small value of C unimportant at audio frequencies would probably become dominant at RF, making simple DC bias ineffective. That's a downer, it gets complicated. One of the reasons for wanting a change is to eliminate the filament current ground return through the chassis shared with the audio ground.

[jnb]

IIRC Chk is often in the order of one to a few pF so this reactance can approach a typical plate load a 1MHz.

WRT the emission, if the heater element is to emit electrons to the cathode it must be more negative than it (for at least part of the time).

WRT the reference, any low impedance reference will do. Disconnecting the ground reference altogether may introduce noise. Using a simply decoupled voltage divider will work (ie two resistors and a cap as a voltage reference), but you can also make a low impedance divider using an emitter follower between the rails.