How does the K2-W (vacuum tube op-amp) work?

[wdbarker3]

Yes, I read "NEWBIES PLEASE READ BEFORE POSTING" which says not to ask "How does this work". Well, I've certainly done some work on my own but I still need some help.

The K2-W came out about when I was born, but I only "discovered" it about 2 years ago. It is a really cool device but its method of functioning is driving me nuts. The schematic is confusing and the device is encapsulated in plastic and not easily probeable (is that a word? probe-able), so, I built another from the schematic with all the component leads long enough to get scope probes on.

I still can't figure it out. The first 12AX7 looks like a long-tailed pair (differential amplifier) but acts like an error amplifier. I imagine those two tasks are not inconsistent. But, I have not figured how that makes the second 12AX7 output enough to do the op-amp thing (whatever it takes to make - = + on the other side)!

I put a square wave into the thing with two 10K resistors in voltage-doubler mode, and then switched in/out a 1ma load. I figured something the second tube was doing would wiggle to cause the current to ramp way up for the load and back for open, but you could barely see anything on the scope except the "correct" voltages on any of the components - it was just working as if by magic!

The schematic has what looks like an NE-2 bulb annotated with +125V. Well, an NE-2 is about +65V so I put two in series. shorting out one of them changes NOTHING! Very weird. The notes say to expect about a 1.5V bias to be needed (see the AA cell?). My "rebuild" also needs the offset. I'm not sure where that comes from either, but it sure centers up the vertical display on the output!

I crossed up something and blew my HV power supply (+300V and -300V required for this circuit) so I built another: more robust, 3 channel, HV power supply to continue. Works well but I'm still clueless wrt the K2-W.

Even if you don't know how this thing works, if you have a good idea for a test to SHOW it working (or disprove an idea), please let me know.

Attached: The original, Mine, the Bench, Philbrick schematic, my schematic and board.

[Jwillis]

It will work just the same as a chip Operational amp. The dual triode 12AX7  are the same as two transistors.
http://www.philbrickarchive.org/k2-w_refurbished.pdf
The first stage is  a differential pair.
I'm probably wrong but the second stage looks like an inverter?

[Jwillis]

No not an inverter but a Common cathode amplifier and a  cathode follower with low output impedance.

[wdbarker3]

Well, three things (plus sleeping on it) helped me to figure it out!
 a)   The rubber duck principle.
 b)   Jwillis saying essentially “D-oh!”.
 c)   Me invoking magic (the only magic in electronics is the smoke that sometimes escapes).

Of course it was not magic; it was the op-amp doing the op-amp thing; keeping the voltages correct regardless of the load.

I had spent a lot of time measuring voltages whereas it is tiny currents that make it go. I was always using negative feedback, so the output was always “correct” for any network I tried. Checking for anything meaningful in a balanced circuit was not going to work – I had to look around with it unbalanced.

So, with 1.6V on the non-inverting input (inverting grounded, no feedback), it would drive the output to the positive rail (65V) and the R2/R3 junction went to -3V. Put 1.5V in and the output goes to the negative rail (-135V). R2/R3 at -1V).

The common cathode amplifier buffers the output and the cathode follower assures the low output impedance. This device is hard to understand when negative feedback is forcing it into balance, but far simpler when driven to a rail.
 

[TimFox]

I remember a well-intentioned article in EDN a while back about the K2-W history.
The commercial K2-W amplifier followed an earlier WW II design by Loebe Julie that used octal 6SL7s instead of miniature 12AX7s.
The article referred to those triodes as "high-micron" devices.