Because most amplifiers particularly in the CIA-1T's price range don't pursue equivalent circuit bandwidth hence speed, I asked Roland Krammer to explain some of the inherent HF design challenges and audible benefits; and how a switch-mode not linear power supply plus current not voltage feedback factor into his equation. Whilst awaiting his reply, I studied Benjamin Baum's review of the CIA-1 over at our German syndication partners fairaudio.de. He opens up with "Grand space. Generous colours. Seductive vocals. Realism which more closely approaches the live experience. It's not just hopeless nostalgia or owners of high-efficiency speakers which swear by tube gain despite well-known risks and side effects: glass bottles mean maintenance and extra care, shed plenty of heat, limit speaker options with notoriously low output power and often pay for the desired midrange magic and musical flow with flaws in bass control, transparency or tonal balance." Then he sets up his expectation bias by admitting to hardcore valve love which, mostly, has curbed his enjoyment of solid state. For exceptions he singles out his Graham Slee Ultralinear headfi amp or a Lavardin ISx which, at the time, approached but not equalled his tubed Jadis DA50.

CIA-1 [photo property of fairaudio.de] showing the lower-cost RCA and binding posts, different mother board and power supply.

After putting his Crayon through its paces, Benjamin concludes that "the Austrian 'flavour enhancer' plays the tunes with a rare combination of potent and intensely rendered tone colours plus arrow-quick/exact accuracy of rhythm, timing and groove"..., calling it "...a dream amp for 'romantic pragmatists' who want to be as close as possible to the tonal essence of tubes without their deficits; or risk financial misfortune by falling for an exotic high-power triode." In a nutshell, tone and speed; one aspect of tubes, one of transistors to become a quasi hybrid. This condensed my sentiments about our Crayon CFA-1.2 rather well to presage happy times for my latest Austrian go-around. And it's one thing for transistor lovers to invoke common ground with tubes, quite another for diehard owners of actual valve gear like Herr Baum. Having immediately prior to the CIA-1T written up the 300-watt push/pull AudioValve Baldur monos with 10 power triodes per channel, I can unequivocally state that into our speakers, the Crayon sounded categorically different. It was far superior on resolution, bandwidth, clarity and reflexes.


Of course tube amps come in sundry topologies: with/without output transformers, with capacitor or interstage-transformer coupling, as push/pull or single-ended, triode, ultralinear or pentode, for simple low power or complex high power. It's important to be specific whenever referencing their kind. Here it's instructive to remember that to go beyond their Jazz tube linestage with its outboard linear PSU, Nagra of Switzerland transitioned to an integral switch-mode power supply for their costlier one-box Classic Preamp successor. It's how they pushed their SN/R to 125dB. This illustrates how switching supplies can be quieter than their traditional equivalents. And to resolve the very smallest of recorded signals, isn't any circuit's 'sound of silence' a - well, grave prerequisite? So much for widespread snobbery concerning SMPS as though their lightweight status from doing without a big power trafo has to be synonymous with a somehow lesser cheaper option by default. It can but needn't be.


In use, the Crayon's power ramp-up is visualized by running its red input LEDs ticker-tape style. Once those settle on the chosen input, your sleeper agent is ready for action. Now one discovers that its remote was coded inverted. Its up/down buttons are for sequential source select, the left/right buttons volume down/up. If at first you have no sound, it's because you're still driving on the wrong side of the street. You're changing inputs, not volume. Exorcize your contrary habit—Brits, Kiwis, Cypriots and other right-hand drivers excepted—and you'll be fine. Ditto for an attenuation taper that comes on nicely slow to not jump from too quiet to too loud in one click; not that the continuous Alps pot has any clicks. Being an integrated, there's nothing else to operating it.

In a secondary system with Qln Signature 3 monitors and Soundaware SD-card DAC to clock some break-in hours. Our usual April Music Stello 100MkII separates were benched for the duration.

Designing it of course wasn't so easy. "Our unique current feedback shuns conventional opamps and is a unique low-impedance circuit of its own. Its advantages are extended bandwidth—1.5MHz in the CIA's power stage—and speed. We achieve this with extremely short feedback loops and associated supplies of low impedance


"The disadvantage is open-loop gain in excess of 100dB, well beyond conventional opamps. Yet for audio-signal amplification, the bandwidth/speed of current feedback is advantageous because its error correction is ultra fast to create no typical timing issues. Plus, whatever errors are introduced by the loop fall well outside the audible band. We use SMPS because they offer regulated output voltages and—in our experience—higher resolution. Linear supplies could be used of course but their price and weight would be drastically beyond the CIA range, never mind the far bigger chassis. Why 1GHz? Our power supplies are designed so the wave impedance of audio frequencies remains below 1Ω all the way up to 1GHz. We talk of wave impedance because even the power supply of an electronic circuit is in fact a signal conductor. Unfortunately multi-layer power planes don't create virtual plate capacitance - an impossible ideal due to output/input impedances. So the power supply should be terminated with ultra-low resistance to avoid high-frequency reflections. That's an unrealistic demand when you consider what such a supply would have to generate to remain <1Ω. It's just not sensible. Instead we design our power planes for <1Ω. This converts ultrasonic energies into heat, thus eliminates their disturbances on the signal path. We take this effect very serious because no transistor, not even relatively low-frequency transistors for audio, are immune to ultrasonic contamination. That makes them behave as diodes to demodulate. Once in the signal path, those demodulation products cannot be removed. In this case, prevention really is the best cure to secure a pure hifi signal. To reiterate, wave resistance [above left] is different from standard impedance [right where it shows as a function of inductivity, ohmic resistance and capacitance]. Any circuit, transformer or other electronic part has a typical impedance (not a pure ohmic resistance). A circuit trace too has an impedance. However its qualities are created by characteristic wave resistance where, in English, 'resistance' becomes 'impedance' to cause widespread confusion."

On cosmetic sibling rivalry, the 1T has the same footprint, a lower chassis but nicer connectors and terminals.

The implementation of the above parameters relied on PCB layout and construction of "innovative methods having layer stacks with prepreg thickness of 50μm (the prepreg material is a thin sheet of fibreglass impregnated with uncured epoxy resin which hardens when heated and pressed during the PCB fabrication process)" according to design rules by Nils Dirks as published in the Elektronik Praxis journal under "The circuit board 2010" [chapters 12+13]. Studying that paper reveals that having one's cake and eating it—a wide bandwidth fast circuit that's immune to ultrasonic interference—is no walk in the park. Back to current feedback whose 'c' becomes the capital 'C' of Crayon, "like in vacuum tube circuits, this topology exhibits gentle clipping behaviour. We call it 'smooth clipping'. Small short overdrives are hardly noticeable because the harsh sounds usually created due to recovery times of output transistors aren't present." To avoid pops without speaker relays on even highly sensitive speakers, the CIA-1T uses a software-steered soft-start protocol. This runs during the ticker-tape phase of the source LEDs for both the turn-on and power-down cycles.


But Roland's CV isn't all tech, time-sequence controllers, program logic, microprocessors, FPGA and PCB wizardry. He apprenticed as a tool maker when still young to later handle Crayon's CNC chassis fabrication. He also attended the Graz conservatory of music and the performing arts with a major in drums and vibraphone to be very keyed into musical timing and transient accuracy. Whilst working for industrial electronics firms, he moonlit creating mixing boards and doing live sound around Austria to know what the real thing sounds like on stage. With music very much in his genes, we now segue into power-up for the comparative listening impressions followed by driving the CIA-1T in 'PA' mode with a Nagra Classic Preamp to bypass its own attenuator.