This looks interesting so I printed out the whole thread for perusal during lunch breaks and we'll see if I can get my head around it. The original schematic has so many other fancy things in it for the PI etc that I had trouble grasping the basic idea. George's drawing in post 30 however I think clears up the basic gist of the thing. It appears that the idea is to drive both G1 and G2 with a larger AC signal going to G2 and also using fixed bias to bias G1 lower than G2.
Is that it basically? Is self biasing a possibility on something like this?
mike
That's pretty much it. It's probably reasonable to start with a ratio if G2/G1 voltage approximately equal to the G2/G1 mu factor. Sone of the reasoning behind that is in the thread I linked to at the beginning.
It's going to take a little experimentation to see if there's a preferred offset between G2 and G1 at idle. I don't think the SPICE models are going to be much help in this mode with low G2 voltage and current on both grids.
G1 current will be substantial (~100mA peak), so my strawman design has a MOSFET follower off the voltage divider to drive G1. The SRPP-like gm-followers in the driver anode circuit are for driving the G2 with low impedance, but any low impedance follower arrangement would work as well.
I don't consider this a candidate for self-bias due to the low current idle point and near class B operation. On the other hand, the gm is sufficiently low when driving G2 and G1 together that fixed voltage bias should be stable enough. Also I only need a small positive bias voltage wrt the cathode (~20V) so can generate it with adequate stability using depletion mode MOSFET and a resistor.
The total peak grid current on each side could be 150mA or more summed into the cathode, so I am using a separate stacked supply returned to the output common cathodes to drive the grids and create a small local current loop. The supply stacked under the output cathodes only sees the fixed current through the LTP set by the tail CCS.
There is a lot of stuff for sure, but I'm trying to carry the design forward into a more or less practical amplifier. I could replace the MOSFET in the LTP with another pentode, simplify the driver screen circuit and maybe even go with partial feedback and anode resistors in the drivers. I could also cap couple the driver, but I think I would still need 4 followers per channel and a negative supply for their sources.
I'm not even sure about the advantage over tetrode mode, assuming both are using plate-grid feedback. Combination drive is more linear but there is a lower gm to start with. One may be able to get higher peak current but some of the cathode margin gets used by G1 current. I do see an advantage over G2-only drive, in that approximately 1/2 the drive swing is needed.
Tubes with higher gm and lower G2/G1 mu are more suitable, just as wiith G2 drive. Don't need a regulated screen supply, but there is the extra drive difficulty.
As always a tradeoff but it looks like a fun way to build a higher power amp.
Cheers,
Michael
