To learn more about the SA-90, Jos from Magna Audio forwarded my questions to his contact at the factory.

1/ how many gain stages does this circuit consist of; and how do they couple (direct, capacitors, others)?
2/ what are the gain devices in each stage (Jfet, Mosfet, bipolar, other). How many are there in the output stage – two per phase?
3/ how much feedback is there, local or global?
4/ how many watts operate in class A?
5/ the filter capacitance seems quite low for such a device. Could we explain why?
6/ is the entire circuit discrete so without opamps?
7/ the residual noise seems exceptionally low. Could we explain how this was achieved?

"This power amplifier circuit is very simple, with only an input stage, buffer stage, voltage amplifier stage, current driver stage and current output stage. They are all DC coupled, with film capacitor coupling only in the input section. All stages use bipolar junction transistors. We use a single pair of very high-power 220W/ea. BJT in the output stage. To achieve ultra-low distortion one can only use global not local negative feedback."

"This also requires a relatively large amount of NFB but compared to competitors, ours is far lower. This is due to the ultra-linear design. Our class A power envelope is very small, just about 0.5 watts. Our filter capacitance of about 22'000µF isn't large but excellent PCB layout played a big part in not needing more. The power path is fully discrete, only the balanced-to-single-ended input conversion uses two op amps. To achieve ultra-low noise, the first requirement is excellent power-supply design followed by very good PCB layout, low-noise transistors and a low-noise transformer."

Now the purist understands that the RCA input avoids the op-amp stage of the XLR input. The armchair engineer appreciates how very low class A bias decommissions massive heatsinking, a large chassis and how a push/pull circuit without paralleled devices limits the achievable power within this context. The relative budgeteer appreciates how all of these decisions controlled the final spend. That returns us to my opening paragraph to wrap the prelims: the SA-90 moves the classic monaural class AB amplifier concept downmarket to propose itself to a shopper otherwise eyeing an integrated Marantz-class amplifier.

Having selected regular ground and high-gain mode on the underbellies then set the input switches to RCA, the SA-90 went into the upstairs system to clock some wakeup hours. Electrically and mechanically, both amps were dead quiet as promised. This is far from a given with power amplifiers though you'd wish it were. If the noise is in the signal, we call it distortion. Why don't we call external noise created by the playback equipment distortion as well? All too often there's some power-supply surf through the tweeter and/or midrange, hum of transformer or minor ground loop. As one ratchets up speaker sensitivities toward then beyond 100dB, self noise becomes ever more critical as a thief of dynamic range. On my 85dB loads, I was at a cemetery, no ghosts of the recently departed present who in Ireland might partake of whiskey spilled for them at the grave. Just perfect. Another nicety not always guaranteed were the frontal standby switches. They avoid fumbling behind the rack to turn things on/off. The paper stickers for the ground/lift and hi/lo gain switches admittedly felt just a bit declassé but are out of sight so out of mind. The Singxers and I were off to a good start. I didn't miss a trigger port since in 25 years of doing hifi, I've never once used one. As a dealer, Jos would prefer it had Singxer added them. If such feedback arrives back at the factory, perhaps the next batch of amps might have triggers? Personally it's immaterial either way.

In my upstairs use the SA-90 never got warmer than Cool Hand Luke but retained that mild action in standby to not return to full-frosty off. Likely to shorten warm-up from dead cold, some bias current remains live. Pushing the stand-by button caused no switching noises or delays. The safety-sleeved speaker terminals have their spade slot face up and the wide fat spades of my Kinki cables entered and secured without issue. Even on tracks recorded very cool to hover around -20dB, the standard 2V RCA outputs of my Sonnet Pasithea DAC required no extra preamp gain to get loud enough on these 85dB speakers whilst leaving gas in the tank. The DAC-direct connection with variable reference voltage on the R2R ladders for loss-less attenuation was perfectly sufficient. Being power amps, this covers all functionality with flying colours. Now we're free to consider sonics. Published measurements show its 2nd, 4th and 6th harmonic higher than the odd-order 3rd and 5th. All higher-order products reduce linearly toward -120dB, with only the 2nd slightly above -100dB. That THD profile and a very low noise floor make for a uniformly lit up very detail dense sound. It seems fair to call it thoroughly modern because it subscribes to common expectations of high resolution. It fell right into my favored Kinki/Goldmund niche by avoiding anything thick, gelatinous or dark. Compared to my usual EX-M7, it was a very close stand-in. The only repeatable difference I keyed into was the Kinki's top end. That felt even more teased out. It visualized still more of the airy, fizzy or shiny aspects of certain tracks. Perhaps it was even less phase shift in the non-audible bandwidth? This made the Singxer no darker per se, just not as finely filigreed down to about 2kHz. Demonstrably high S/NR meant strong contrast. That naturally led to potent image pop as the subjective intensity, focus or presence sounds obtain versus the silence of surrounding space. Our understanding of ghosts is that we see right through them. Ever since the first Star Trek transporter beam, movies have shown the gradual materialization of humans from ghost-like flicker to full-blown physicality. Playback sound too differs in how much fleshy materialism spectral transparency attains. Our eyes never actually see our sounds in the flesh. Any extra pop or presence in the audible domain undermines otherwise more ghostly encounters between/behind the speakers. The Singxer sound achieved its 'thereness' with high relief against the proverbial light-absorbent blackdrop, not by manipulating the amplitude domain for a bassier balance. This presentation was quick, concise, well lit and direct. It was a bit like an amplifier version of a high-efficidency widebander. Its materializer action lived in the focus and contrast domain. It didn't work on tonal balance, dynamic excess or harmonic thickening.