Since davygrvy has started the ball rolling here on "heretical" tube designing (I wonder who influenced that? ), I might as well follow suit
Been cooking up a driver for the ST-70 (well, good for many PP amps, I just have the ST-70 to experiment with) and wanted to use the CMFB phase inverter first played with here: http://geek.scorpiorising.ca/GeeK_ZonE/ ... 7#msg31977
Tube choice for development was a 6CG/FQ7 phase inverter and a 5751 driver. These being minature versions of the 6SN7 and 6SL7 being renowned for great sonics (yes, the circuit works for the octal versions too). I also tried a 12AX7 and 6N2P with almost identical results, just a tad more gain.
First, a look at the circuit:
The specified voltages are not all that critical... performance was maintained on the pre with a B+ of 240-290V and on the PI of 380-420V.
Gain selection resistors on the cathode of the PI were selected to have roughly the same as (maybe a wee more) overall gain to the KTA-HiFi driver I am using for a reference: http://www.kta-hifi.net/projects/amp_pa ... /st70.html
Common Mode Feedback simply samples the difference between the phases and adds it back to the cathode in a little local NFB loop that provides phase correction. It has some spectacular results as shown below:
Using a 6CG7 with a known wonky grid on one section (~40% left on it) and CMFB, the output on a dual-trace scope is:
(scope settings are 20V/div. vertical, 200uS/div. horizontal for all images)
Using a 6CG7 with REALLY mismatched mu + CMFB results:
Notice there is roughly a 5V total difference between the phases at 80V P-P output.
Here again is the same stage as a LTP with the tail resistor chosen for exact same total current:
(scope settings not touched from the previous photograph)
The use of equal anode resistors allows the tube to have DC balance. CMFB allows an automatic AC balance without resorting to expensive "Matched Sections" tube sales :)
The CCS shown in the circuit is one I have a PCB for, so can even be tacked on to an existing compatible driver board. CMFB using this method is a SS adaption from the one developed by Fred Nachbuar here: http://dogstar.dantimax.dk/tubestuf/driver02.htm
With the assortment of 6CG/FQ7 I have on hand, I was able to get anywhere from a 4:1 minimum AC balance improvement, to over 20:1 improvement on a tube that already had matched sections.
Cheers!