S&M + SNR = whollotta noise? Abbreviations get cryptic fast. Here S&M isn’t sado/maso. It’s sales and marketing. And the other thing isn’t snot ‘n’ rasp but ordinary signal-to-noise ratio. As ex national sales manager for Mesa Boogie’s hifi division, Meadowlark Audio and Soliloquy Loudspeaker Company, I knew a thing or two about life on the road, speakers or amps in vans (not white ones), sleeping in cheap hotels and criss-crossing territories to visit existing dealers and make new ones. I also knew about getting slaughtered or gypped by reviews and how to kill sales. But that’s a story from a previous life.
For today let’s simply say that keeping it honest whilst being in sales and marketing can be interesting. Now we’ll look at the number’s wars being played out for once not with sample rates, bit lengths or power ratings but S/N ratios. For them a basic definition is the ratio of desired signal to background noise expressed in decibels. A ratio greater than 1:1 indicates more signal than noise. It can also be applied outside audio, say in biochemistry and nerve signal on the cellular level.
Most hifi noise about SNR is made with D/A converters. That’s because signal-downward from such modern sources things become inglourious basterds in a snap. For DACs today’s spec hunter expects and routinely gets120dB. As a logarithmic function the value of decibels gets easily confusing. A 2 x power increase nets a 3dB gain, a 10 x change in power represents a 10dB level change. A voltage change by a factor of 10 meanwhile clocks a 100 x power change. That’s a gain of 20dB. Our graph shows how power and amplitude correlate.
Back to our super DAC. We’re starting off very high on the totem pole. If you watch Bones, king of the lab. Now suppose we leashed to our DAC a tube preamp because we like what it does for the sound. Take my very expensive Nagra Jazz. My personal unit’s bench test sheet shows a S/N ratio of 118dB. For a tube pre that’s unbelievably wickedly awesome. It’s flatly one reason for my choice. If we went with a far less extreme valve pre however, say Eastern Electric’s very solid but only $1.100 MiniMax, we’d look at an SNR of 93dB. Hello deafer Dolly. That’s a loss of nearly 30dB against our source.
Like a drunkard ready for another, let’s continue on our downward spiral to really hit rock bottom and the next AA meet. We buy ourselves a shiny pair of Trafomatic Audio top 300B monaural amps. Because their very gifted maker is also a very honest chap to even list noise (many in this particular game do not!), we learn of their 80dB S/N figure. Cough. Versus our 120dB source we’ve thrown away a staggering 40dB. That’s an amplitude differential of 100!
Now we appreciate why Mola-Mola’s Bruno Putzeys of Hypex and Ncore fame makes such exuberant noise over their very powerful Kaluga monos (400/700/1200w into 8/4/2Ω). Despite high voltage gain of 28dB, their unweighted S/N ratio clocks in at a whopping 128dB. Katchinnnnnnng! Finally we have a real-world amplifier which relative to the very best current DACs isn’t a bottle neck on noise performance but actually exceeds most of ’em. Pretty much everything else lags behind our best sources and often to an alarming degree.
Viewed from this perspective one appreciates why many engineers look at single-ended valve amps of the zero feedback persuasion like my prior Yamamotos were with undisguised disdain. They literally represent the counter culture or flip side to current efforts on improving our source material with high-resolution ‘studio master’ files. Noise after all is the enemy of micro-level signal. The more noise we burden our systems with, the less of the tiny stuff makes it through. That’s not me lobbying against valve amps. Home audio is about personal pleasures. It’s not about winning the spec wars. Whatever makes you happy wins. Period. But …what if vastly lower noise did equal greater listening pleasure?
Viewed from this perspective one also appreciates why switch-mode power supplies ought to get their day in the sun. Getting rid of 50/60Hz power line noise from conventional power transformers is essential to advancing this aspect of the art. To design a proper audiophile-grade SMPS whose ultrasonic switching noise is perfectly contained simply seems a lot harder. Linn have worked on theirs for decades. So have Chord. My €4.250 Crayon Audio CFA-1.2 from Austria instead uses one designed and engineered out of house, in Taiwan by a specialist giant in the sector. It’s superbly quiet and the amp sounds terrific. As amplifier power ratings go up, the transformers supporting them with linear supplies get larger and larger and as such more and more challenging to keep quiet.
SMPS still have a bad rap. In general they tend to be nothing more than cheap laptop-type bricks which dump plenty of HF noise into our power grids to infect other gear. But that needn’t be the case. Class D amps are slowly overcoming their bad rep though it’s taken a good decade. The reason is probably similar. Far fewer boutique hifi makers are sufficiently competent at class D and SMPS than they are at the traditional topologies we’re all used to. Let’s face it, hifi in general isn’t an industry which attracts the brightest lights from engineering school. Proper engineers pursue far bigger money and fame in the IT, medical, automotive or military sectors.
To wrap up our 4th keep-it-honest chapter, we acknowledge that S/N performance remains an oft-overlooked aspect of the game. Here the poorest component in our chain acts as the literal stranglehold or filter on how much micro nuance survives translation from source to speakers and finally our ears. It really doesn’t take being a stone-faced Bill Murray to get Lost in Translation. Even us D-list actors can get a part in this movie without any effort. The real question is, why should we want to?