Model S:
Making a serious listening easy

 

Model S is state of the art D/A converter, and leading edge design with venerable TDA1541A D/A chip.

Originally released in 2008, to supersede and improve upon The Model, it remains the highest level real multibit (ladder, a.k.a. “R-2R”) D/A conversion performer, and reference in natural music reproduction achieved by the digital audio source.

And Model S MkIII, available since July 2015, makes another step ahead. Beside the S/PDIF input, MkIII also has a direct PCM input, which can accept either I2S signal, for use with increasing number of digital sources supporting I2S as external interface, or simultaneous data signal, for ultimate TDA1541A jitter performance – a TDA1541A property discovered during the Model S USB development.

At its output side, Model S MkIII has two pairs of outputs, any of which can be either single end (RCA) or balanced (XLR).

 

 

● S/PDIF and PCM direct (I2S or simultaneous data mode) input

● Advanced DEM clocking scheme for TDA1541A converter chip

● Audial unique hybrid output stage without feedback yet with top level linearity

● Transformers, capacitors or directly coupled output

● Isolating mains transformer with split secondary output (balanced AC power)

● Nine independent supply lines, each with three stage passive common mode filtering and zero feedback low noise regulators

● A highest quality parts

● Engraved solid aluminum chassis

● Each sample of the Model S is shipped with its own lab report.

 

Inputs

The Model S has true 75 Ohm BNC S/PDIF input. This input is transformer coupled, with advanced termination scheme for proposed 75 Ohm impedance over required bandwidth. The ground of S/PDIF line can be capacitively coupled or left completely floating with respect to the DAC, by the switch at the back plate. With S/PDIF ground left floating, a differential input network effectively forms symmetrical input, bringing a highest possible common mode rejection of unbalanced line.

The other, PCM direct input, passes the signal directly to the Model S MkIII D/A section. This input can accept either I2S or Philips simultaneous data mode protocol, and uses the set of four BNC connectors (I2S requires only three of them), each carrying particular clock or data signal, as proposed by the given protocol. Typically, this input is terminated to 75 Ohm, but on request it can be terminated to 50 Ohm, or left non-terminated. This input is DC coupled, and works with both 3.3 V and 5 V sources.

 

D/A chip classic

Just as its predecessor, the Model S employs Philips TDA1541A D/A converter chip, introduced more than two decades ago, but still with ultimate performance among multibit D/A chips.

In addition to that, the Model S employs improved clocking scheme for TDA1541A internal Dynamic Element Matching circuit.

 

Non oversampling

The Model S doesn’t use an oversampling, and it especially doesn’t use an “upsampling” (sample rate conversion) of any kind. The things with such a signal processing indeed have improved over the years, however it is still not possible to equal naturalness, ability to convey musical content and ability to make a long term listening enjoyable, of well implemented non oversampling device.

 

No feedback

The output circuits of the Model S run in healthy class A and don’t employ a global feedback. A current to voltage conversion is performed by the hybrid circuit which uses both integrated circuit and discrete parts.

So, no opamps here, but also no discrete transistors just to claim “discrete output stage”, essentially using the same old opamp topology. Frankly, as long as one settles on the opamp architecture, a monolithic opamp is not bad choice at all, simply because there are excellent monolithic opamps nowadays.

A high signal slopes, such as those coming out of the D/A chips, however push the opamps to work beyond the limits of proper feedback operation. That is why important sonic improvements can be achieved once the feedback is abandoned in favor of adequate no feedback circuit; natural scale and dynamics, and exceptional transparency and soundstage are the first descriptions that come to mind. Abandoning feedback, of course, additional effort was needed to achieve static performance specifications required by the modern criteria. With 0.002% THD, the Model S sets new standard in this regard. Please note that this figure includes multibit DAC, zero feedback output stage, and transformer at signal path.

Of course, the active parts used must be able to follow and not to limit such a high signal slopes – if and when necessary, the signal slope must be slowed down only by the passive, and not by the active parts. The Model S hence uses a high bandwidth active parts, and current to voltage converter is a common base circuit, which brings a highest possible bandwidth in itself. The output buffer, with bandwidth well above 100 MHz and slew rate higher than 1000V/us also do not set the limits of its own.

The output buffer is capable of driving any line load one can think about, and its internal output impedance is 3 Ohm.

 

click to see full size picture
 

 

Power supply & layout

The Model S employs probably the most serious power supply on the audio market.

First, inside the Model S chassis there is isolating mains transformer that feeds the main power transformer by two half-voltage secondaries (“balanced AC power”). Both isolating and main mains transformer employ internal electrostatic shield between the primary and secondary windings.

The Model S is supplied from nine secondary windings. (And there is also a tenth winding that feeds the input switching relay.) This way each stage of the unit has its own supply so, among the other things, the Model S output stage operates as “dual mono” device. Each supply section includes rectifier made of Schottky diodes – for their lack of recovery overshoot and superior sonics – followed by three stage common mode RC filter, and dedicated regulator. All the regulators are fully discrete, low noise, wideband circuits, that do not employ feedback. As a result, the power supply lines are quiet with practically eliminated crosstalk between particular sections.

Such a supply scheme made possible to design layout without compromises, where not only supply sections are kept separated, but different signal paths are kept separated too, and they are also separated from supply paths. In addition to that, all the signals and their return paths are made short, making loops physically as small as possible. No star ground approach is applied, since it can’t address requirements set by high speed circuits. The PCB is double sided, with ground fill on top side used mostly for shielding and not for routing.

 

High quality parts

The Model S uses high quality parts like Schottky diodes, Rubycon, Panasonic and Elna electrolytic capacitors, Panasonic PPS (polyphenylene sulphide) capacitors for TDA1541A active divider decoupling, and polystyrene capacitors for HF filtering, Allen Bradley resistors, Toshiba low noise audio grade transistors, Jensen signal transformers, Neutrik connectors etc. Everything is housed within solid aluminum chassis, with all the markings engraved. The net weight of the unit is about 9kg.

 

Transformer or directly coupled RCA output?

Balanced (XLR) outputs of the Model S are always transformers coupled. Audial uses high quality signal transformers made by Jensen, and these are the Jensen’s flagship units JT-11-BMCF, a vanishingly low distortion, high bandwidthThis means that they preserve linear phase high above the audio band. They also have excellent transient response, with no overshoot. units.

On the other side, unbalanced (RCA) outputs can be either transformers, or capacitors, or even directly coupled.

With S/PDIF input, directly coupled output has negligible DC offset (up to several milli Volts). TDA1541A however can exhibit significant offset if it is not fed by proper clock and data signal, so caution is required with PCM direct input. For further information in this regard, it is best to contact Audial directly.

 

Inputs:

S/PDIF electrical, 75 Ohm BNC connector (*); transformer coupled, with capacitively coupled or floating ground

PCM direct: I2S or Philips simultaneous data protocol; direct coupled

 

Sampling Frequency:

S/PDIF: up to 96kHz

PCM direct: up to 96 kHz (I2S mode with 64 bit frame), or up to 192 kHz (I2S mode with 32 bit frame, or simultaneous data mode)

 

Outputs:

Unbalanced (RCA) – transformer, directly or capacitively coupled, and balanced (XLR) – transformer coupled;

2V RMS both

 

Output Impedance (20Hz-20kHz):

Transformer coupled output: ≤ 90 Ohm

Directly coupled output: ≤ 4 Ohm

 

Frequency Response:

Sin(x)/x equivalent:

@ fS=44.1kHz: 0dB @ 20Hz, -3.2dB @ 20kHz

@ fS=88.2kHz: 0dB @ 20Hz, -0.8dB @ 20kHz

@ fS=192kHz: 0dB @ 20Hz, -0.2dB @ 20kHz

 

Absolute Phase:

Correct

 

Transient Response:

Transient response is clean and damped, with no overshoot.

The pictures show 1kHz and 20kHz square waves, transformer coupled output.

 

 

 

Harmonic Distortion:

0.002% @ 1kHz, 50% full scale (-6dBFS)

0.012% @ 1kHz, 10% full scale (-20dBFS)

0.87% @ 1kHz, 0.1% full scale (-60dBFS)

 

 

 

 

Intermodulation Distortion (CCIR):

0.009%

 

 

Mains voltage:

220-240VAC, 50-60Hz, IEC (C14) connector

110-120VAC, 50-60Hz is also available

 

Dimensions (W x D x H):

431.5 x 320 x 102 mm (including knobs, connectors and legs)

431.5 x 278 x 78 mm (excluding knobs, connectors and legs)

 

Net Weight:

Approx. 9 Kg

 

* – Most of the S/PDIF cables are still equipped with RCA plugs, and Model S is hence for compatibility reasons shipped with RCA (female) to BNC (male) adapter. This adapter is however by no means a long term solution, but only a help for initial use, or possibly for some cables testing. For regular listening, it is highly recommended not only to use real 75 Ohm BNC plug, but also to put in order the whole S/PDIF line, which includes both cable with connectors, and source.

Please note the difference between 75 Ohm and 50 Ohm BNC connector. Model S S/PDIF input, as well as Model S MkIII PCM direct input specified as 75 Ohm, employ real 75 Ohm BNC connectors, and not 50 Ohm BNC, which you can in fact usually find at audio devices. Please also note that an RCA connector, due to its given geometry and so long as it preserves required co-axial construction, can be 35 Ohm at most.

 

MkII (2012):

Black Gate capacitors are replaced with Rubycon and polyphenylene sulphide / polystyrene / Elna Silmic capacitors.

Mains transformers do not use outer shields (cans) anymore.

 

MkIII (2015):

Instead of two S/PDIF inputs, this version has one S/PDIF and one direct PCM input, which supports I2S and Philips simultaneous data protocol.

Mains transformers are improved.

 

 

 

 

Recommended topics:

TDA1541A today

TDA1541A simultaneous data mode

Transformer coupling pros and cons

 

Model S users’ reviews at:

audio.codexwilkes.com/audial-model-s.html

www.audioasylum.com/audio/digital/messages/14/149651.html

theaudiostandard.net/thread/429/audial-model-dac

www.changstar.com/www.changstar.com/index.php/topic,2452.0.html