It's been a while since I last worked on this little amp, but motivated by a friend of mine who saw my build gathering dust in my closet and asked to try it out I finally decided to implement a modification that's been on my mind for over a year.
Inspired by the design of the Millett Hybrid, Millett [Mini]Max, SOHA II, a re-reading of Morgan Jones' "Valve amplifiers" and some posts by Dsavitsk where he mentioned that the best mod you could do to the SSMH was substituting the plate loads for constant current sources, I thought I might try my hand at it.
The benefits of using a CCS as plate load are many and described in detail in many pages (Google is your friend). I'll just summarize a quick list here:
Improved distortion figures
Fantastic power supply ripple rejection (PSRR)
Reduced crosstalk
More gain from the tube
I decided to use the ring of two transistors design, just as in the MiniMAX, since the necessary parts (a couple PNP transistor and a couple resistors) are easily available most anywhere in the world. I settled for the BC327 PNP transistor since that's what was easily available to me where I live.
Based on some current measurements I did in the 12A_7 version, I went for a plate current of ~0.7 mA per tube which based on the details posted here (http://www.diyforums.org/MiniMAX/MiniMAXccs.php), and after rounding to common resistor values, resulted in one 680 ohm resistor and one 10K resistor for the CCS.
I decided to also substitute the cathode bias resistors (R5, R11) for 2K trimpots, so as to be able to set the plate voltage; again, just like in the MiniMAX. I did not use bypass caps around the trimpots.
I put the components in a little perfboard and, after removing the plate load resistors (R1, R7), I ran wires from the junction of R13 and C6 (V+), each tube plate and ground to my CCS "module".
I must say that it didn't work at first, which was terribly frustrating. After much testing, debugging, rewiring, even swapping out one of the output MOSFET's, I found out that one or both transistors in one of the CCSs were blown. After substituting them... Ah! voilá! Sweet music!
I proceeded to measure the voltage at the junction of R13 and C6 (V+) and found 42.6 volts. Then I hooked up the meter to each tube plate and, using the trimpot, I set each plate voltage to 21.3 volts. I measured the cathode voltage to ground and found out it was 0.9 volts. Finally, and just to make sure everything was well before pluggin in my beloved Grado 'phones, I measured DC at the output: 0.0 volts.
So all looks good.
Finally, I got set for a listening session using my collection of 320Kpbs MP3 files, Foobar200 and Bantam DAC. What I'm going to say is of course highly subjective without any detailed measurements to back it up, but, given that disclaimer I'll go ahead: I was absolutely enthralled by the sound coming out from the amplifier. It seemed to have more clarity and definition. It also seemed to me that it had a 'blacker background'. Drums were a delight: listening to Tony Williams' drumming in the second great Miles Davis quintet recordings was a real joy: the transient response, the "air" around the drum sound. Piano was a treat too, whether it was jazz, classical, blues. I just kept listening to album after album and falling in love with my little amp as the added clarity, definition, quieter background and transient response all built up to enhance the 'you are there' sensation.
Again, I'm perfeclty aware that all the above is highly subjective, but for the price of 4 transistors, 4 resistors, a couple trimpots and 2 or 3 hours of your time (just for the build, don't come to me complaining that you spent your whole weekend sitting by the amp with your headphones on!) it seems to me that the bang for the buck factor of this mod is stratosferical.
A couple final things. In the first place, the amp is harder to control now with the additional gain; not a problem with 300 ohm but certainly hard with 32 ohm Grados. It's probably time for me to add those input resistors to attenuate the input signals.
Second, taking into consideration the great PSRR that the CCS adds to the tube stage, I think those who built your own linear power supply should give it a try.
I'll post some pictures later today and will try to add the CCS to the 12A_7 schematic.
cheers!