Happy New Year - A New Amp to Argue About...

[roline]

Kinda looks like the input of a bipolar IC, but the current mirror is only one sided, one side of the mirror is a constant DC voltage, the other swings between the collectors, errr collector and plate.
The second design sets the VBE of both output current sources, the 1k desensitizes the design to VBE/IC variation, so now you loose the mirror to just current sources and improved linearity and differential output. You could make one of the 1k's a pot to balance the stage. In IC's we would use a NPN mirror on the bottom to run the diff pair off of the same bias resister. Again you can use a mismatch in the emitter resisters to set the current ratio. A pot would make it adjustable.. Quite a hybrid circuit.
The MPS trans are only good to 300V with a beta of 20-40

[tubelab.com]

How did you determine the load impedance of the tubes would be well matched for your transformers?

The load impedance applied to pentodes can vary over quite a wide range especially when feedback is applied. When experimenting with a tube that you haven't used before and no data is given (like using a sweep tube for audio) it is best to either experiment, or look for a tube of similar size where the plate curves at least look similar and start there. In this case the pentode section of a 6LU8 / 6LR8 is somewhat like a 6V6 with the cathode from a 6W6. The 6V6 data sheet shows an 8K to 10K load. The fatter cathode in the 6LU8 will work just fine with a 6.2K load. It should crank out 20 WPC or more.

[wrenchone]

Roline - the typical gain of the MPSW92 is much higher - on the order of 100 or so, and of course I can select high gain units. The tubes at the bottom will help keep the top transistors from seeing their max voltage. 10-20k in series with each collector would also help.

[smoking-amp]

Something like this for a complementary mirror load. Q1 and Q2 are P channel Mosfets. R1 and R2 around a Meg-Ohm. Although this is the more obvious way of making one. J.L. Hood's version looked more clever as I recall. To the point that I couldn't figure it out. I don't think it had any control resistors down to the drains.

Maybe it had extra P-chan source followers driving R1 and R2, so as to not load down the outputs. Just guessing.

Hmmm, or how about a floating NPN cascode set below Q1 and Q2. Current into their bases from Q1,Q2 drains would determine the output currents (xBeta) and the control voltages for R1,R2 could come from the base voltages. Wild speculation.

[smoking-amp]

Well, I don't know how J.L.H. did it, but I've figured out how to do it without any loading on the outputs now. Call it the Cas-complement mirror. Q1 thru Q4 are P channel Mosfets. Q3 and Q4 can be low voltage types. R1 and R2 around a 100K Ohm say still. Q3 and Q4 act as floating cascodes in the downward direction, and as source followers in the upward direction. A class act.

edit: Hmmm,, Errr... except one could also call Q3 and Q4 low voltage zeners. Guess they aren't doing much useful. Nevermind.

[Michael Koster]

How about the differerntial gyrator?

Another kind of hybrid

I have an improved version but the basic concept is the same.

Cheers,

Michael

[roline]

Just for fun, spud re-hashed.... Tried to keep as much of the original design with current biasing.

[smoking-amp]

Yeah, Michaels got the differential load in fit form. I still would like to find that J.L.H. version someday. (was an old W. W. article for a SS amp)

Spud Re-Hashed!
Nice scaleability of current thru-out! Laser trimmed? Was it Zilog that did a CPU chip once that had an IC socket on the back? Two compactron sockets topside could be tight.

Brings to mind an interesting chassis design. Put all the parts in a long rectangular case, similar to an IC DIP shape, with a bunch of shiny "lead wires" sticking out along the bottom edges for springy support, and the compactrons plugged into the topside. Toroid OT inside out of sight. Power supply from an umbilical. A small notch in one end of the chassis for orientation. Epoxy black finish. A new meaning to "integrated amp".

[smoking-amp]

Looking at Michael's Differential Gyrator schematic(s) using caps in the cross coupling controls gives me an idea on how to eliminate the residual V sense resistor loading (for AC) on the tube outputs. Since the current drawn by those V sense resistors is predictable, one could use Hawksford's cascode compensation technique to remove the same currents from the Mosfet outputs.

Need some resistance in series with the D1,D2 Zeners, and a small cap (like pF) from R9 and R12 top ends (improved version posted) (or could just use another set of sense resistors) to the same side Mosfet sources. Idea is to subtract out from the Mosfet current (at it's source) the same tiny AC current as drawn by the sense resistors, making them transparent effectively.

[smoking-amp]

In simplified form, something like this: