Imagine a single-ended Class A power amplifier
comprising a single transistor, that is devoid of any feedback, and which
behaves similarly to a power triode? Does this scenario sound too good to be
true? Well, that's exactly what the First Watt SIT-1 is all about. It deploys a
unique type of device, a variant of the JFET, referred to as a static induction
transistor (SIT). It was invented in the early 1970s by Jun-ichi Nishizawa, a
true genius who is considered the "father of Japanese
microelectronics." The SIT enjoyed a brief audio career in power amplifiers
by Yamaha and Sony, but has continued to evolve since then mainly for UHF and
S-band applications in radar and communication satellite systems. It is a
voltage controlled device whereby the signal modulates the electrostatic
potential barrier produced by the gate – the grid like element of the SIT.
Unlike an ordinary FET whose current-voltage characteristics are pentode like,
those of the SIT are triode like. One important consequence of this is the low
output impedance even in single-ended operation. And because of its inherent
low-voltage and high-current characteristics, no output transformer is required
for matching the output stage to a loudspeaker.
In the past 10 years SIT research and development has been focused on using silicon carbide (SiC) instead of a traditional silicon matrix, mainly due to SiC's significant advantage in critical field strength. While in the United States, SemiSouth had become a leader in SiC based SIT power devices, offering customized application-specific solutions. The time was right to invest in this technology despite the prohibitive cost of a custom run of SiCSITs optimized for audio applications. Nelson Pass decided to step up to the plate and take the financial plunge. The devices SemiSouth delivered made their way into two First Watt power amplifiers, the SIT-1 and SIT-2, thus making audio history in the process.
A Few Technical Details
According to Nelson, a practical reason to
explore simple single-ended triode-like circuits is that they generally generate
an harmonic distortion spectrum that is predominantly second order in nature
with the residual being essentially third order: "If you are going to have
distortion, you will want it in a low order harmonic form, kept to only second
and third harmonic if possible." Low-order distortion products are consonant
with the music and are certainly imperceptible at total levels below about 1%.
The SIT-1's THD is specified at 0.7% at 1W for both 4 and 8 Ohm loads. Measured
distortion does rise to over 2% above 5W, though the character of the amp does
not appear to change subjectively. Most SET amplifiers measure at 2% to 3%
harmonic distortion at low power levels but the distortion spectrum is
characterized by the near absence of higher order distortion products whose
annoyance factor detracts from long-term listening enjoyment. It should be noted
that while feedback is useful for the reduction of THD it does not generally
alter the relative distribution of distortion products, which means that you can't
make a pentode sound like a triode by merely cranking up feedback levels.
The SIT-1 is passively biased using several 7.5 and 5 Ohm 100W film type power resistors. The front panel features a bias meter and a bias adjustment pot. This feature serves a couple of useful purposes. First of all, it allows for compensation of minor variations in line voltage. There are times in my neck of the woods, when under heavy demand conditions the line voltage will drift low, and then return to nominal value late at night. And secondly, it enables the user to tweak the operating point for optimal coupling to the loudspeaker load. Any bias setting within the green zone is said to be safe. However, distortion levels, which are mainly second order, do vary slightly with bias setting for a given load. For example, Nelson's measurements show that biasing to the left of the green zone center line works best for a 4 Ohm resistive load, while for an 8 Ohm load total harmonic distortion (THD) is reduced for bias settings in the right side of the green zone. However, due to the complex nature of a real-world speaker load, there is no correct setting, rather (as Nelson puts it) only the one that sounds best to your ears. Incidentally, the SIT-2, the stereo version of the SIT-1, is biased at a fixed value corresponding to the center of the SIT-1's green zone. And being biased by a FET based constant current source, the SIT-2 does not really need compensation for AC line variations.
The input is AC coupled via a 4.7 uF Clarity
polypropylene cap. And since there is no output transformer, a large DC blocking
cap is used in the output, made up of a parallel combination of two 4,700
uF/100V electrolytics and a 4.7 uF Clarity polypropylene cap. Efficiency is
quite low at 5%, much lower than the 20% of a typical SET. The heat sinks get
quite warm to the touch, as the SIT-1 dissipates 200W at idle, but is only able
to deliver 10W into a load. That's a consequence of being a passively-biased
single-ended design with triode-like source-gate characteristics. As with other
single-ended Class A designs, the output stage works less when it is actually
playing music. This may seem counter intuitive, but as power is delivered to the
load the output stage runs a bit cooler.
My review samples were production in every way
except for the omission of the JFET based input buffer stage. Without the buffer
stage, the SIT-1's inherent input impedance is 10 kOhm, which should not be an
issue with any line preamp sporting a source impedance under 1 kOhm. But in
order to ensure compatibility with any preamp or source component out there,
production units incorporate a buffer that provides a 100 kOhm input impedance
option. Input impedance selection (10 or 100 kOhm) is performed by means of a
gold jumper which plugs into an XLR connector near the RCA input jack. The 10K
setting is of course more purist in nature, as it bypasses the buffer stage. I
was fortunate to have both the Pass Labs XP-30 line preamp and XP-25 phono stage
on hand for the duration of the listening tests. These front end components are
an excellent match sonically for the SIT-1, while the XP-30's 50 Ohm source
impedance was more than satisfied with a 10 kOhm input impedance. Consisting of
a single gain stage, the SIT-1 inverts signal polarity. However connections to
the speaker binding posts are already reversed internally, the Red post being
grounded, while the Black post is the positive or live connection. That means
that you will automatically obtain the correct polarity by simply following the
color coding of the output connections.
The Sound Of One Transistor
My first and lasting sonic impressions were of startling immediacy, soundstage transparency, and 3-D dimensionality. The spatial impression expanded to flesh out a superbly delineated depth perspective populated with tightly focused image outlines. In the immortal words of the late Harvey Rosenberg, there was plenty of wholosity to behold! After the initial shock of it all, it took me a few seconds to gravitate toward a critical aural sampling of the harmonic textures. And what a luscious treat that was: purity of tone coupled with naturally hued timbre. In particular, female vocals were amplified with sensational tonal color fidelity. Violin overtones were fleshed sweetly and with the delicacy of the real thing. All-tube analog recordings came through with their warmth and analog flavor fully intact.
An important point to make, and one I shall endeavor to make perfectly clear, is that the SIT-1's presentation was far removed from that of a traditional SET amplifier. A case in point is Pete Millett's single-ended DIY amplifier that squeezes perhaps 2.5 wpc from a French R120 triode, a 6.3V indirectly-heated version of the venerable 2A3. While it got clobbered by the SIT-1 at the frequency extremes, the R120's midrange sounded texturally richer, more relaxed, while tonal colors were a bit more vivid and saturated. Subjectively speaking, and with program material that did not stress its limited power delivery, I found the R120's siren song to be more persuasive. You could argue that the R120 is more euphonic due to the blooming of second order distortion with program level, or that its textures are overly liquid and you would be probably right. But at the end of the day, note that if you're in the mood for passionate SET sound, the SIT-1 would most likely not satisfy that need.
In my experience, a huge advantage of no-feedback
SET over traditional push-pull amplification is in bringing to life microdynamic
nuances. It may sound surprising to recognize that even a 2 wpc SET can outdo a
100 wpc push-pull amplifier in terms of dynamic contrasts, but that is the power
of the first watt in the context of a high-sensitivity speaker load. The SIT-1
was able to dig deeply into the fabric of the music and communicate the music's
soul, drama, and passion. In this it was aided by a phenomenal grasp of rhythmic
nuances that helped propel the music forward with plenty of kinetic energy.
Needless to say, my late night listening sessions were nothing short of magical.
overall presentation was exquisitely detailed without overtly calling attention
to specifics. Yet, the SIT-1 was able to place a mix under the microscope and
readily resolve reverb decay and vocal overdub harmonies that often get fudged
over by lesser amps. That was a direct consequence of a super low noise floor
combined with superb transient speed and control. Whereas most SET amplifiers
suffer from a constrained power bandwidth which leads to a soft and overly
liquid presentation, the SIT-1 was unhindered by bandwidth limitations. The
treble range was open, airy and always in control. Bass lines were reproduced
tightly, with plenty of punch, and with exceptional pitch definition. In these
respects, the SIT-1 far transcended the performance of transformer-coupled SET
Nelson Pass Responds
United States Dealer