This review page is supported in part by the sponsors whose ad banners are displayed below

Reviewer: Srajan Ebaen
Financial Interests: click here
Source: 27" iMac with 3.4GHz quad-core Intel Core i7, 16GB 1.333MHz RAM, 2TB hard disc, 256GB SSD drive, ADM Radeon HD 6970M with 2GB of GDDR5 memory, PureMusic 1.84 in hybrid memory play with pre-allocated RAM and AIFF files up to 24/192; April Music Eximus DP1, Esoteric/APL Hifi UX1/NWO-M
Preamp/Integrated: ModWright LS-100 with Synergy Hifi tubes, Esoteric C-03, Bent Audio Tap-X
Amplifier: First Watt F5, J2, M2, ModWright KWA 100SE
Speakers: Aries Cerat Gladius, Voxativ Ampeggio
Cables: Complete loom of Zu Audio Event, KingRex uCraft USB cable with UPower battery supply
Stands: 2 x ASI HeartSong 3-tier, 2 x ASI HeartSong amp stand
Powerline conditioning: 1 x GigaWatt PF2, 1 x Furutech RTP-6
Sundry accessories: Extensive use of Acoustic System Resonators, noise filters and phase inverters
Room size: 5m x 11.5m W x D, 2.6m ceiling with exposed wooden cross beams every 60cm, plaster over brick walls, suspended wood floor with Tatami-type throw rugs. The listening space opens into the second storey via a staircase and the kitchen/dining room are behind the main listening chair. The latter is thus positioned in the middle of this open floor plan without the usual nearby back wall.
Review Component Retail: $10.000/pr

In August of 2011 I'd reported on a pre-production version of what since has become known as the FirstWatt SIT-2. That nomenclature stands for Static Induction Transistor 2. The extra words channel amp are implied. The output device is Nelson Pass' very own. He custom-commissioned it in costly silicon carbide technology from US supplier SemiSouth.


Why do we need or want a new output transistor? The above might be the answer. Triode curves. This transistor related to the VFets Sony and Yamaha experimented with in the 1970s (today's generation benefits from radar-related government-funded R&D) can be operated like purist triodes*1. The implications won't be lost on our readers. Clearly Pass isn't the only contemporary amp designer to be fascinated by the possibilities either.
_________________________________________________________________________

*1 These Jfets have a structure which keeps the device out of the 'saturation region' where they normally operate and where the current through the device is relatively independent of the voltage across the device (Drain to Source voltage or Vds). Below saturation is the 'linear' or 'ohmic' region where the device behaves like a triode. - np
Kazuhiko Nishi of Digital Do MaiN too has reissued SITs based on the 2SK278B, 2SK78 and 2SJ78 [left, plus basic schematic of his B-1a amp]. He simply packages his parts in rather more complex circuits to obtain fashionably high power. There's also the SD-011 amp from Maxonic. Using the same Nishi-San SITs with an output transformer, it makes only 30wp but its 40-watt power consumption is clear indication of class A/B bias. Nelson's amp is a SIT as SET(ransistor).

SIT-1 prototype

High power. Or rather not. Therein lies today's tale. In the Pass Labs company Nelson already has a clearing house for mid to brute-power amplifiers. Very much by shameless design, as FirstWatt he can explore the greater simplicity of low-power circuits. It's the somewhat exotic sort which thrives on a single output device per channel (or one per phase if p/p). In the SIT-1 monaural amplifier, this aesthetic is pursued to its practical limit. The singular result? A single-stage zero feedback zero degeneration class A single-ended circuit with a single active device. There's just one transistor. The rest is resistors and caps and a power transformer. If that weren't enough effrontery to how things are usually done in transistor land, there's also user-adjustable bias. And switchable input impedance*2.
First pair of SIT-1 pre-production units in Japan. Owner demonstrates load-line bias adjustment.

*2 The junction capacitance of the SITs won't display their full 500kHz bandwidth with high output impedance sources. I thus replaced the prototype's fixed 47K input impedance with a switchable 10K/100K. Sources with high Z-out like tubes without cathode followers or 'passive' preamps can now be mated to the 100K setting. This buffers out the effect and allows for full bandwidth performance as you can get it with low-impedance solid-state preamps.

Such extreme minimalism comes at an obvious price. Think SETtributes. Think high output impedance of 4 ohms; quickly rising 2nd-order harmonic distortion under more power; and a mere 10 watts of go juice into 8 ohms. As the distortion graph shows, the SIT-1 just about hits 5% THD at full power. Where things get decidely unbottled is frequency response. It's flat from 1.5Hz to 200kHz (-3dB at 500kHz). Here valve amps can only dream. Ditto for the SIT-1's ultra-low noise levels (150uV unweighted full bandwidth). There are clear differences. And why otherwise bother?


Most designers lucky enough to have such a novel part on their hands—Pass would remind us that luck had nothing to do with it, just a risky buy-in commitment in the lower six figures for a first batch of untested parts and a quote of $300.000 for the second batch—would go after multi-stage push/pull circuits with feedback. That would guarantee the much higher power and lower measured distortion which are considered hallmarks of modern transistor amps. After all, who'd pay good money for a micro-power transistor amp with middling distortion measurements?*3 The SIT-1 begs to differ. It doesn't even try to appeal to that crowd. Hence it gets to pursue triodeness to its ultimate conclusion. It actually behaves like a traditional valve SET in more ways than not. It simply arrives with circuitry that's decidely simpler. Go ahead. Build a valve amp with just one tube, one stage and no output transformer. Then drive a loudspeaker with it.
______________________________________________________________________

*3 That's relative of course. This SIT is loaded to only produce low-order THD which remains constant across its bandwidth. Multi-stage valve amps with degeneration—which Nelson calls nothing but another form of feedback—exhibit very similar levels of 2nd-order distortion but aren't linear in this bargain; and suffer phase shifts from bandwidth limits and output transformers.


11 January 2012 00:01: "To keep you up to date, I have silver metal on order but the initial February shipment of SIT-1s will be black. I should have silver for the 2nd shipment and of course your name is on a pair." Being into this for the long haul—living with the loaner SIT-2 proto had sunk in its teeth—I'd decided to forgo instant gratification. I'd stick by my guns and continue to hate black hifi all the way to no.

$5.000 SIT-2 prototype vs. €40.000 Trafomatic Audio Vilobha monos on Voxativ Ampeggio speakers
Now that we've established how the SIT-1 at least on paper is the closest any transistor amp has ever come to a bona fide glow-in-the-dark valve SET, how would sonics confirm or defy expectations? For those not averse to technical reading, here is a link to some preparatory background on static induction devices. As that paper puts it, when the Japanese first introduced the SIT concept, most transistor manufacturers couldn't comprehend how it worked. At the time only Japan managed to ever build such parts. And collectors like Nelson still acquire them. Not NOS valves but NOS transistors. Talk about full circle for solid state!
Another photo of the very first pair of pre-production SIT1s in Japan
"btw, I haven't taken a photo of production but here's the pic I took of the first proto SIT-1. Except for the three-legged SIT, the eight things that look like transistors on the sink are 100-watt Vishay resistors which set the bias (the SIT-2 is biased by a FET current source). The power supply is just shy of 100V and the efficiency comes in a 5% or a 200-watt power draw."
How does the SIT-1 get away with so little? The transistor itself cleverly combines all the necessary characteristics for a low-power amplifier. It has sufficient voltage and current gain—18dB of voltage gain and a current gain of ca. 1:10.000 as the input/output impedance ratio of 47.000:4—high input impedance and usefully low output impedance. "This SIT device has higher gain and lower distortion than tubes and conveniently operates at the voltages and currents which loudspeakers require. That eliminates an output transformer with all its bandwidth and distortion limitations."