Enourmous emitor resistors
0.47 ohms is not "enormous", it's more like a typical value. In addition, in the NPN part the resistor is not even tracking emitter current.
absent Vbe mutiplier that doesn't needs to sense any temperature (and can fail at that)
One important function function of the bias circuit IS to track temperature and a VBE multiplier can do it much more accurately than a diode string. It's a sole reason why such more complex circuit is used pretty much as a standard. It has been proven to be a better and more reliable method than a simple diode string.
nor needs to be set by a potmeter (which can fail)
I see proper bias adjustment as a virtue, not flaw. Safe potentiometer failure mode can be noted in the design process.
the low overal gain before feedback, the low bandwidth (good stability)
The current feedback topology actually has an enormous bandwidth. Granted it is more stabile than the (nowadays) more usual long-tailed pair. Ignoring HF local feedback might work with this scheme but overall it might also be wishful thinking that only works on paper. I have hardly ever seen this topology used in practice without requiring some local HF NFB - just like any other generic SS scheme.
the quasi complen output stage that is not sensitive to output trannie overheating (good Iq stability)
Where did you get the idea it's non-sensitive to thermal effects? It's actually a combination of two kinds of thermal coefficiens, those of darlingtons and those of sziklais, both requiring their unique thermal tracking scheme.
Overall the quasicomplementary stage also has much weaker stability than a fully complementary output. The crossover region is very asymmetric and coupled to weak biasing arrangement this can create terrible stability issues where the amp begins to oscillate near crossover regions.
the single rail
...Makes it way more prone to power supply introduced flaws, such as hum, DC offset drifts... And you need a large output cap, which should be able to handle all the load current and huge AC fluctuations....reliably.
zobel network
Has been a standard since late 1960's in just about any decent design to begin with. A good design should really have an inductor zobel too.
and indded, the output Cap prevents any DC current.
Perhaps it wil, but in this particular design the wacky bootstrapping arrangement might make even that doubtful. Since the speak was about "bullet proof" things I'd rather trust generic DC coupling with relay switching coupled to a DC detector, at least it doesn't rely on blind faith that cap protects the thing from everything and it also doesn't need a high capacitance, high voltage capacitor with stupidly high ripple current ratings.
Everything about that design is about as simple as it can get, keeping it robust and low-fi.
I can agree about simple and lo-fi but not about robust.
Do enlighten me....
- What tracks temperature accurately when coefficients of the output pair vs. diode do not match?
- What protects it from overheating?
- What protects it from short circuited output, and how well?
- What protects the voltage amplifier from failing if the output stage starts to hog current above "safe" levels?
- What protects the speaker from DC at output, assuming the amp fails in a manner that there will be DC in the output (e.g. the bootstrap or the output cap fails)?
- What prevents hard saturation?
- What protects the output transistors from reverse biasing during flybacks?
- What protects the speaker from the loud transient "thump" when amp is powered or shut down?
Oh... in closer inspection "bullet proofness" actually seems to be kinda lacking. If we really want to use that adjective then may I kindly guide towards high end PA amplifiers and alike, not ludicrously simple designs dating back to 1960's something.
