Peregrinations about Preamps
A meandering trip through some thoughts on preamp design
If I was inspired to build a new audio preamp from scratch today, I’d call up NASA and ask about their suppliers for the Mars Reconnaissance Orbiter. I figure there are some real interesting challenges when your transmitter is up to 248 million miles away, and can only muster 100 watts of power to get back to Earth. As near as I can figure, the received signals are at around -155dBu (-215 dBW). Certainly getting a usable signal out of that must be a triumph of low-noise amplification.
As I looked into it further, I learned that the space agency is satisfied with a mere 8dB signal to noise ratio, because they can extract the digital data from the bedraggled stream of data bits that fall, exhausted from their long trip, onto the antennae dishes in Gladstone, California. Obviously they’re solving a different problem. The lesson here is “horses for courses”.
I then looked into the outer reaches of the Hi-Fi enthusiasts. I only dare go there when I’m prepared to be surrounded by congenitally dissatisfied people, for whom no product, specification or design strategy is ever ‘good enough’. Therefore, the ever-expanding search for a ‘better’ product keeps the wheels of commerce turning. If you surf The Absolute Sound website, you can find some amazing hardware at amazing prices. I took the time to read one review of a preamp and mono block power amps that have a DC-to-1Mhz frequency response, and an SNR of 108dB. The noise floor on the FFT shows a noise floor around -130dB, but I couldn’t ascertain the reference level … whatever dBr is, that was the reference. By the way, the preamp costs $40K, and the mono block amps are $115K for the pair.
http://www.soulution-audio.com/en/serie7/701/index.php
One of the lines from the review explains that that the preamp/amp combo has “…slightly less dimensionality, but not less density of tone color and texture… " I don’t know what that means to my listening experience, but I’m pretty sure this guy is shopping for a can of dissatisfaction in the happiness aisle. Having learned that it’s impossible to please the gatekeepers of high end audio. I went back to the drawing board to get some bearings.
Another way to go at this is to admit that we're using microphones capturing sounds through the air. The new start focuses on the noise floor of the microphones. After a bit of web surfing, the lowest self-noise figures for microphones that I could find were the 4.5dB SPL and 5dB SPL figures from Rode and Audio-Technica. Sometimes these self-noise figures are ‘A-weighted’ and sometimes not. A more typical value for a high end microphone is 14dB SPL, and small diaphragm condensers usually ring in closer to 19dB SPL. So as long as the noise floor of a gaggle of my proposed preamps doesn’t raise the perceived level of the self-noise of an equally large gaggle of microphones, my design goal has been met.
Then, there’s the bandwidth problem waiting for us. Whenever you amplify or convert something, the wider the frequency response, the higher the noise. It’s just the physics and math of signal theory forcing these realities. The A/D converter people have done a stellar job of managing the wideband noise problem in the process of converting a voltage to a digital value (and back). The preamp people make a box that will have to amplify a microphone signal, but they have to deal with the always-there Johnson (thermal) noise inherent in materials that are at room temperature. There’s no getting around that, although the folks at The Absolute Sound would encourage a cryogenic encapsulation strategy.
Signal to noise ratios and Equivalent Input Noise (EIN) are kind of insidious, earning the badge of perfidy that was once the title of this piece. A good preamp with an EIN of -130dBu, attached to a microphone putting out -50dBu will have an SNR of 80dB. In a great preamp, the noise and the signal will rise with gain by equal amounts. There’s no getting around the noise rising with gain. Rats! If we drop the bandwidth of the preamp in half we can get back a 3dB improvement in EIN but that’s a big sacrifice for a small improvement.
Even if the preamp has only one knob on it for gain, there are likely to be at least two gain stages within the preamp. The first is likely to give 30dB to 40dB of gain, and then the next one is 10dB to 20dB or more. Sometimes the first stage is unity gain just to provide a stable input for subsequent gain stages. Managing the ranges and handoff between the stages is where much cleverness and magic has been expended.
Shall we go fully clean, DC-coupled solid state, or perhaps transformer-coupled, with tubes? How about transformer-coupled transistors, or hybrid tube and transformers with capacitor coupling? You want fries with that? Okay, we’ll include deliberate amounts of 2nd, 3rd, and 4th harmonic distortions via the loading on the transformer. It’ll be sweet because the transformer will do its magic in a dynamic, level-and-frequency sensitive way. You’ll love it! How about jalapeños? Well, now, that’s another matter!
You see, if you want the special jalapeño 2nd, 3rd and 4th harmonic distortion via tubes, there are a couple of ways to do that. The expensive way is to have a set of full-voltage (150v to 300v depending…) tubes driven via an appropriate step-up power supply from the mains. This has the advantage of intellectual purity, historical coherence with guitar amplifiers, and high maintenance; providing excuses to ‘tube roll’ various brands through the preamp and discuss the differences wearing a silk smoking jacket while sipping brandy.
The other way is to use a ‘starved plate’ approach to generating the harmonics. This is reprehensible to the purists who reserve their most derogatory barbs for the horrid companies who conspire to trick us into getting our distortion at lower cost via lower voltages. Yea, verily, I have read the most impassioned commentary about the alleged torture of tubes running at less than 50 volts. These tirades read like they’d discovered an active device trafficking ring, forcing Zener diodes past their current limits, biasing transistors into non-linear regions, and excoriating Hall Effect sensors with neodymium magnets. The horror!
One factor that explains these market dynamics is the problem of the barriers to entry. Whenever technology lowers the barriers to entry for newcomers in a field, the defense shields go up to protect the old guard establishment. We can see it in the challenge that micro-breweries have made to the beer market. We can see it in the way blogging and freelancers have affected conventional print media, and we can see it in audio. As long as ‘good sound’ meant a $3K microphone into a $2K/channel preamp, mixed through a $20K hardware-based DSP and controlled by the fastest-available computer, that kept the poseurs and amateurs away.
When the technology changes to allow a $400 microphone through a $400 preamp connected through a $400 interface to practically any computer, the barrier has been obliterated. Now it’s time for the priests of the preamps to defend their temple! They must intone with great gravitas about the wimpy build quality of the challengers. This is the context in which you'll see the term 'prosumer'. It's only invoked from the priestly altar of the Temple of Expense. Surely, none of these poseur devices have passed NEBS earthquake testing! With that framing of the issue, they go on to prove to us all that distortion must be earned, not just applied. It must be created via complex circuit topologies, discrete components, high labor costs, low volume manufacturing techniques, and invisible marketing. These all work together to maintain the mystique.
I can’t actually denigrate the high-end audio preamp market. (You don't tug on Superman's cape / You don't spit in the wind...) I would love to have a 500-Series lunchbox with a population of Millennia, Daking, API, BAE, Burl, Grace and LaChapell preamps (to name a few). But overall, it’s still performance, composition, microphone, position, and acoustic room environment that carry the day. I can toddle along with my Golden Age, Aphex, ART, and Focusrite preamps, making compelling tunes at great prices. When the chance is there, I'll buy the big-dollar stuff because it appeals to me, not because I think it'll make the difference between a dud and a hit.
In the creative fields (and I include audio engineering as a creative endeavor) the practitioners operate in that nether world between intent and results. Just as artists suffer with the permanence of the gap between their intent and performance, so do we engineers imagine a sound. We are goaded by the hope of a perfect microphone, processed by a perfect preamp that's feeding a perfect A/D converter. That hope is swimming past us, to be speared by yet another circuit topology in a preamp, just maybe this time...
Perhaps those guys at The Absolute Sound are just a mirror of us studio engineers at the other end of the audio chain.
- Ted Gary of TedLand
June 7, 2016