Raquel - I've never heard either amp that I mentioned but specified S/N=120dB was confirmed by Stereophile measurement and audition of Coda S5. It is dead quiet and very resolving. As for Ralph amps - reviews praise them just for being quiet and very resolving. Putting two tubes in parallel doesn't make it less resolving or noisier than one tube. Part of SixMoons review of Atmasphere MA-1:
"Compared to my BAT VK-75, the Atma-Spheres just offered more of everything: More detail, more extension, more transparency, better soundstaging, greater micro and macro dynamics, more extended and luscious high frequencies"
You assessment of matching output devices is not accurate because it is easier to get average characteristic by connecting many output transistors in parallel. Linearity also improves since each transistor works at smaller current range.
Noise voltage that limits resolution of bipolar transistor is defined by resistance of the Base and resistance of the Emitter that can be reduced by reduction of emitter current (inversely proportional) while Base resistance become dominating reason for the noise. Placing transistors in parallel not only reduces emitter current but also reduces resistance of the Base (in parallel). Placing many bipolar transistors in parallel is well known noise reduction technique in microphone preamps.
As for negative feedback - It is not the complexity that screws up the sound but lack of bandwidth. Limited bandwidth amp delays the signal (shifts phase). When input signal changes fast, feedback loop is momentarily open creating overshoots (or odd harmonics in frequency domain). Our ears are very sensitive to odd harmonics (especially higher order) because loudness clues lie in them. It is called TIM (transient intermodulation) and is known since 70s.
One way is to compromise THD, IMD, DF and make an amp without global feedback but the other is to use just enough feedback to reduce THD to about 0.2-0.5% and then limit input bandwidth (before feedback summing stage) to one that amp had without feedback applied. There are many great amps designed with or without feedback - there is no simple answer. Some of these great amps are very simple some are very complex.
I just realized that Ralph Karsten has the same last name as first name of the Icepower (that I use) designer Karsten Nielsen. Karsten people design good things.
"Compared to my BAT VK-75, the Atma-Spheres just offered more of everything: More detail, more extension, more transparency, better soundstaging, greater micro and macro dynamics, more extended and luscious high frequencies"
You assessment of matching output devices is not accurate because it is easier to get average characteristic by connecting many output transistors in parallel. Linearity also improves since each transistor works at smaller current range.
Noise voltage that limits resolution of bipolar transistor is defined by resistance of the Base and resistance of the Emitter that can be reduced by reduction of emitter current (inversely proportional) while Base resistance become dominating reason for the noise. Placing transistors in parallel not only reduces emitter current but also reduces resistance of the Base (in parallel). Placing many bipolar transistors in parallel is well known noise reduction technique in microphone preamps.
As for negative feedback - It is not the complexity that screws up the sound but lack of bandwidth. Limited bandwidth amp delays the signal (shifts phase). When input signal changes fast, feedback loop is momentarily open creating overshoots (or odd harmonics in frequency domain). Our ears are very sensitive to odd harmonics (especially higher order) because loudness clues lie in them. It is called TIM (transient intermodulation) and is known since 70s.
One way is to compromise THD, IMD, DF and make an amp without global feedback but the other is to use just enough feedback to reduce THD to about 0.2-0.5% and then limit input bandwidth (before feedback summing stage) to one that amp had without feedback applied. There are many great amps designed with or without feedback - there is no simple answer. Some of these great amps are very simple some are very complex.
I just realized that Ralph Karsten has the same last name as first name of the Icepower (that I use) designer Karsten Nielsen. Karsten people design good things.