Do I esentially have a class A amp?

Digepix, I understand that you consider class A better sounding, otherwise you wouldn't care. It should be better sounding if it is designed that way. Class AB has to use more of negative feedback to counteract nonlinear operation of output transistors. This negative feedback improves everything: THD, IMD, Bandwidth, Output impedance etc. but unfortunately can produce Transient Intermodulation (TIM) - basically overshoot of fast changing signals (signal fed back is late because amp introduces delay) that in frequency domain shows as addition of odd order harmonics. Since it can make sound unpleasantly bright designers try to keep amount of negative feedback to minimum. 30dB gain of power amp represent voltage gain of 30 but amplifier will most likely have gain of 300 in class A and 3000 in class AB before feedback that brought gain back to 30 was applied. Class AB needs more feedback (and therefore initial gain) not only to linearize non-linear operation of output transistors but also to reduce "gm doubling" effect. Gm doubling (trans-conductance doubling) is simply gain difference between two (class A) vs one (class AB) transistors operating at given moment.

Since your amp has both - nonlinear region of transistors and gm doubling it has to use more gain and "damage" was already done. This whole issue is very complicated and it was even shown that increasing bias in typical SS amp can actually increase THD distortions (overbias). I'm sure Krell engineers know what they're doing and this amp can sound better than class AB amp but it is not as good as equivalent class A amp at any loudness level.

I understand you need to know what you really got for your money but I would pay more attention to system synergy and overall sound and less to classes of operation, specifications etc. Often amp with poor specifications has wonderful sound.

35W won't tear the walls but don't forget that 350W (that tears the wall) is only 2x louder. Also 3dB more sensitive speakers will give you same as doubling the amp power.

So, how your 35W amp compares with good 140W amp driving 3dB less efficient speakers? 140W amp will be only 22% louder.
People often ask if it is better to buy 200W or 100W amp. In spite of amp's power being useless specification IMHO, I would always choose 100W because for the same money I can buy better amp.

Runnin, 250W would be nice if you can get it for the same price as 100W but if it is more money than perhaps getting better 100W (better design, better components) would bring more satisfying sound.

Bifwynne, Local feedbacks are better. Even emitter (or cathode) resistor is a form of NFB. SS amplifier, being voltage source, needs feedback at least for the output stage. Tube amps, being power source, most likely need less of it.

It is possible to design amplifier with negative feedback free of TIM. Since problem of TIM is caused by summing fed back delayed signal, during fast transition, it would be logical to prevent fast transitions and to eliminate delay. It comes to bandwidth limiting input stage before summing point and extremely fast output stage. High quality parts have to be used to achieve good initial linearity and wide bandwidth. NFB should be mostly local but also shallow just to bring THD distortions to about 0.5%. Bandwidth should be limited at the input to one that amp had before feedback was applied (since NFB widened bandwidth).

Such amp will have 0.5% THD and perhaps only 50kHz bandwitdh not to mention relatively high output impedance. THD distortion is not that audible, 50kHz bandwidth is passable and output impedance doesn't matter because inductor in series with the woofer is about 0.1 ohm (limitting DF to 80) while speaker impedance (being source impedance for back EMF) is mostly resistive.

Would you buy 100W amp with THD=0.5%, 20-50kHz, DF=50 when somebody else is selling 250W with THD=0.01%, 20-200kHz, DF=1000 ?
Specifications are useless - only sound matters. All comes to good design, quality (expensive) components and smart compromise.

NFB lowers output impedance. If amp has 10x higher gain before applying of NFB then feedback will lower output impedance tenfold. We call it 20dB feedback. As a result deep NFB some amplifiers end up with output impedance of single milliohms. There is no need for that, since inductor in series with the woofer is about 0.1 ohm, while the fact that speaker's internal impedance is mostly resistive limits even more. When speaker's coil moves in response to signal it also generates response (back EMF) that works like brake that limits coil movement. This brake has damping effect on the membrane.

DF=1000 looks good on paper but it doesn't make big sound difference. Deep NFB used to achieve it, on the other hand does, making sound bright and unpleasant. Highly praised Atmasphere amps have output impedance in high ohms (DF=1).
Some speakers work well with low output impedance amps while others, (overdamped) sound great with tubes, but sane level of DF is perhaps 100.

It is very tempting to improve specifications with NFB because it doesn't cost anything. If you see fantastic spects in cheap amp you know that something has to give - it is sound.

Charles1dad, Thank you. Atmasphere mentioned negative effects of NFB many times. I still believe that shallow NFB can be used but design has to start with good circuit and quality components. The best is perhaps not to even think of it and just to listen.

I don't design audio but it would be quite easy to design an amp that would measure THD=0.001% IMD=0.001% BW=500kHz and DF=1000 with absolutely horrible sound. I suspect that if you find amp like that (plenty of them in 70s) IT WILL have horrible sound (and they had). Again, spects have nothing to do with the sound. Even power spects are so vague that amp listed as one with higher power might actually play softer. Price of the amp might be even better indicator of quality than spects.

Czarivey, 22dB is not a deep feedback. Most of digital amps don't use deep NFB because there is no need for that. PWM is quite linear while output has by definition low impedance to start with (since one Mosfet is always conducting to low impedance point - GND or VCC).