During development of the Sonoma Model One (M1) electrostatic headphone system, particular attention was paid to every aspect of the design to ensure that the system, as a whole, met two critical criteria:
1. to deliver High-Resolution Audio in unparalleled sound quality;
2. to provide the ultimate in listener comfort.
Here we explain which technologies have been used and the benefits each delivers.
The ideal loudspeaker transducer would have zero mass, respond infinitely quickly to any signal, be perfectly damped, have no distortion and be perfectly linear. Until now, the technology that best exemplifies those characteristics is electrostatic. Introduced almost 60 years ago, electrostatic headphones have remained the choice of discerning listeners who demand the highest audio quality. Now, a revolutionary new electrostatic transducer, derived from the world of ultrasonics, has been developed in the UK by Warwick Audio Technologies Ltd. (WAT). The patented High-Precision Electrostatic Laminate (HPEL) audio transducer ushers in a new paradigm in the field of electrostatic drivers, and the Sonoma M1 is the first headphone system in the world to use it.
A conventional electrostatic transducer consists of a thin membrane (coated with a conductive material) between two electrically conducting metal grids. There is a small gap between the membrane and grids. The membrane is kept at a high DC potential relative to the grids, and the audio signal is applied across the grids. This results in the membrane moving in response to the audio signal thus creating sound. Clearly, in order for the sound to propagate through the grid/membrane ‘sandwich’, the grids have to be perforated in some way.
In contrast, the HPEL uses a thin (15 μm – less than the thickness of a human hair), flexible laminated film for the ‘front’ grid. The laminate is affixed to the open (cell) structure of an insulating spacer (made of Formex™), and the film is very accurately machine-tensioned in the x-y plane. In this way, small ‘drum-skins’ are created by the cells. A stainless steel mesh forms the ‘back’ grid. When the audio signal is superimposed on a 1350 V DC bias voltage, the ‘drum-skins’ formed by the flexible ‘front’ grid vibrate, producing sound. Unlike a traditional electrostatic panel, the sound you hear from a HPEL does not pass through a grid! To take full advantage of this feature, everything has been done in the design of the M1 to keep the areas in front of and behind the transducer as clear as possible so as not to impede the sound waves.
Thanks to a proprietary Finite-Element Analysis software package, WAT is able to fine tune the characteristics of the ‘drum-skins’ such that they have different resonant frequencies. Each cell is acoustically independent, but driven in parallel. As a result, the sound from each cell combines in acoustic space, but the independent resonances average out, avoiding any large resonant peak in the audio band (as can happen with a single driver area).
The thin, light laminate material ensures an extended frequency response, with the panel remaining linear to over 60 kHz. The HPEL has unmatched transient performance, and its surface area has been maximized to deliver a full frequency response. Additionally, due to the fact that it is produced with modern automated manufacturing techniques, the HPEL delivers unprecedented consistency and matching between transducers (< ±0.8 dB difference between left and right channels). The simplicity of its design also enables exceptional durability and reliability.
Naturally, such a light, thin transducer is susceptible to bending, etc., so the panel is encased in a special, super-rigid, high glass-fill polycarbonate ‘cassette’ which allows the transducer to perform optimally.
Injected Magnesium Ear-cups
Compared to Aluminum, Magnesium is about ⅓ lighter and has superior acoustical damping. Combining excellent strength-to-weight ratio, high stiffness and outstanding EMI/RFI shielding, Magnesium is the ideal material in which to house the HPEL transducer to ensure optimum performance. The ear-cups of the M1 are formed by precision injection molding at very high pressure. The result is a weight of 10.7 ounces (303 grams) which contributes to outstanding headphone comfort. Only high-grade stainless steel screws and fasteners are used in the headphones for high strength and resistance to corrosion.
Handmade Cabretta Sheepskin Ear & Headband Pads
Comfort in a headphone is of paramount importance. We want you to enjoy the M1 for extended listening sessions. Producing an excellent sounding, lightweight design would all be for naught if the ear and headband pads induced any discomfort. So, to ensure maximum listening pleasure, the pads on the M1 are all handmade from top-grain Cabretta sheepskin leather. Cabretta leather is renowned for its light weight, smoothness, suppleness, and durability.
The leather used in the M1 is sourced from Ethiopia. The hides are then tanned by Pittards in the UK, who have been tanning leathers since 1826. Finally, the pads are hand sewn in Germany.
Nylon 12 Headband
It is crucial that a headband combine sufficient flexibility – to adjust to different head sizes – with sturdiness and a resistance to cracking under stress: just think how many times a headphone gets put on and removed from a head over its lifetime. To meet these criteria, we have opted for a headband made from Nylon 12 (aka Polyamide 12). This allows the headband to flex as required, while remaining strong over prolonged use. In addition, Nylon 12 acts to dampen noise and vibration.
Internally, the headband contains stainless steel parts which have a vapor-deposited titanium coating to ensure strength and smooth operation.
Custom Low-Capacitance Cable
An electrostatic headphone does not present a normal load to any partnering amplifier. As a result, it is impossible to use just any cable to connect amplifier to headphone. Therefore, in collaboration with Straight Wire Inc., a totally new, ground-up design was done for the M1. The goal was to meet the following criteria:
1. capacitance had to be minimized;
2. mechanical noise pick up had to be minimized;
3. signal purity had to be maximized;
4. the cable had to be thin, flexible and light weight.
Getting some of these criteria in a cable was not a problem. Getting them all was more challenging, as obtaining gains in one area could come at the expense of something else, usually the thin, flexible and light-weight aspect.
The cable is composed of very fine strands of silver-plated oxygen-free high conductivity (OFHC) ultra-pure copper. The insulation is a foamed polyethylene chosen because it has a suitably high dielectric constant, and is well damped. There is no shared ground between the left and right channel signal cables, and a fiber filler material in the jacket keeps the conductors as far apart as possible (to reduce capacitance and crosstalk) as well as helping to dampen any cable microphonics. To ensure excellent strength, two Kevlar® fibers are woven into the cable.
The resulting cable capacitance is a low 50 pF/m. At the amplifier end, the jack is electrically isolated from the amplifier chassis, while on the headphones high-precision self-latching connectors are used to ensure a secure connection. A sense loop built into the cable runs to each ear cup, and if the cable is disconnected at either the amplifier or headphone ends then the amplifier automatically shuts itself down.
ENERGIZER & DAC
Discrete Single-Ended Class-A Amplifier
Like all electrostatic transducers, the HPEL requires a high-voltage drive amplifier in order to function. In the case of the M1, the drive comes from a high-performance, single-ended, discrete FET Class-A amplifier with very low distortion and wide bandwidth which is optimally matched to the HPEL. The amplifier was designed and optimized to drive the inherent capacitive load of an electrostatic transducer, and the Class-A output stage is operated at a high bias level, and delivers a very high slew-rate. Operating at such high bias levels results in improved linearity.
The drive signal has a maximum amplitude of 145 V (rms), which is superimposed on the 1350 V DC bias. Despite operating at low current levels, the high voltages in use translate to significant power (for a headphone amplifier) at the output of the amplifier. Consequently, high quality devices are used throughout which are designed to cope with the power levels.
“Regardless of the type of gain device, in systems where the utmost in natural reproduction is the goal, simple single-ended Class-A circuits are the topologies of choice.”
— Nelson Pass
FETs are often said to combine the sonic characteristics of tubes with the reliability of solid-state. In the M1, FETs from International Rectifier intended for linear amplifier applications have been selected. The attention to detail continues with our commitment to using the best passive components—optimized for their specific application in the electronic design—from the likes of AVX, Bourns, Vishay etc.
To ensure isolation from all interference sources, the amplifier is encased in a completely shielded, machined aluminum enclosure (see below).
The amplifier provides the following inputs:
1. USB 2.0 (digital)
2. coaxial S/PDIF (digital)
3. ‘high-level’ RCA (x2) jacks (analog)
4. ‘low-level’ 3.5 mm stereo jack (analog)
The USB 2.0 input accepts all Hi-Res Audio formats up to 32-bit/384 kHz PCM and DSD via DoP (DSD64/DSD128), while the S/PDIF input accepts all PCM formats up to 24-bit/192 kHz. The high-level RCA inputs operate with a maximum input signal of 2.1 V (rms), while the low-level 3.5 mm jack accepts a maximum signal of 850 mV (rms).
ESS SABRE Reference DAC
The Sonoma M1 system was developed to deliver true high-resolution performance. For the critical digital-to-analog conversion stage, we turned to an established leader in the field. ESS Technology is universally recognized as the world’s premier DAC chip manufacturer, and we have opted for their 32-bit Reference DAC. Two stereo DAC chips are used in a special mono mode to deliver a measured 129 dB signal-to-noise ratio (SNR).
Custom 64-bit Fixed-Point Digital Signal Processing
Anyone who is familiar with loudspeaker or amplifier measurements will undoubtedly have seen plots of flat frequency response from a few tens of Hz all the way up to 20+ kHz. A flat frequency response is the ultimate goal for those components, and is quite easily achieved in an amplifier, but more difficult to achieve in a loudspeaker, especially in a real-world listening room (as opposed to an anechoic chamber).
Unfortunately, the situation with headphones is a lot more complicated. The way in which a sound field interacts with our ears gives rise to a non-flat frequency response known as a Head-Related Transfer Function (HRTF). Worse, the HRTF changes depending on the direction from which the sound arrives.
Further complicating matters, there is no universal agreement on what a headphone target frequency response should be. Historically, there were two options: free field or diffuse field. Free field is akin to listening to a pair of loudspeakers in an anechoic chamber, something very few listeners do. In contrast, diffuse field is akin to listening to a pair of loudspeakers in a highly reflective (reverberant) room. While this may be a bit closer to most listening conditions, it is still less than ideal.
“Frequency response is the single most important aspect of the performance of any audio device. If it is wrong, nothing else matters.”
— Floyd Toole
Studies have found that listeners actually prefer something different, and more like listening to a pair of ‘flat’ loudspeakers in a good listening room. This is the approach we have taken with the M1, and the target frequency response is termed a ‘modified, pseudo-diffuse field response’.
Helping us to achieve the desired response at the output of the headphone, we digitally process all signals using custom 64‑bit double-precision fixed-point arithmetic, running within a high performance, multi-core XMOS processor. It is well‑known in the field of audio processing that fixed-point arithmetic is best, and the 64‑bit arithmetic used in the Sonoma M1 exceeds the performance of most professional audio workstations.
All the filter responses within the DSP are minimum-phase, slow roll-off and are optimized for excellent time-domain response.
Taking further advantage of the incredible accuracy of the 64-bit DSP, we were able to implement within the amplifier a fully-digital interpolated volume control which dramatically outperforms all the pure-analog, stepped attenuators we evaluated. The benefits include:
1. no loss in fidelity and no loss in dynamic range
2. perfect left/right channel matching
3. no potentiometer/attenuator non-linearities at the end of range
4. no noise issues like ‘zipper’ noise, clicks/pops, etc.
In short: it sounded better!
AKM 32‑bit/384 kHz Premium ADC
Due to the need for DSP to achieve the target frequency response, all incoming analog signals must first be converted to digital. This is undertaken in a multi-channel 32-bit/384 kHz AKM Premium ADC chip. As the amplifier has both low-level (3.5 mm) and high-level (RCA) inputs, separate ADC channels are used depending on the input selected. That is, there are two independent, fully-optimized signal paths, one each for the low- and high-level inputs. In this way, the full dynamic range capability of the ADC is achieved regardless of the input selected. The measured SNR of the ADC stages exceeds 120 dB.
Crystek Ultra-Low Phase-Noise Oscillator
The ultimate performance of any digital system is determined by the quality of its master-clock. To guarantee the highest performance regardless of the input signal format, the Sonoma M1 is built around an ultra-low phase-noise oscillator from Crystek. Operating at 100 MHz, it features extremely low close-in phase-noise (<90 dBc/Hz) and an industry-leading rms jitter level, at 100 MHz, of 82 femtoseconds (82 x 10-15 s). All clocks used throughout the M1 are derived from this exceptionally accurate master oscillator via a sophisticated clock distribution scheme using Texas Instruments frequency dividers.
Optimum PCB Layout
The use of premium parts alone does not guarantee high quality performance. In the design of the M1 system, particular attention was paid to circuit board layout to ensure the lowest noise and distortion levels possible. Through careful component positioning, we were able to maintain an ultra-low noise-floor of -129 dB!.
CNC-Machined Aluminum Amplifier Enclosure
In contrast to the M1 headphones, where every effort was made to minimize weight, the M1 amplifier is a very substantial unit. Starting with a solid billet of high purity Aluminum 6063, the material is first extruded, and then CNC-machined to produce a shallow U-shape comprising the base and side walls of the amplifier. The metal walls which remain are 5/16th of an inch thick. A top-plate of the same thickness is similarly produced, along with ½ inch thick end-plates to form the complete enclosure. A special 3D-wave pattern is then cut into the base and top-plate by CNC-machine to facilitate heat dissipation.
To obtain the desired finish in the metal, the machined parts are then blasted under high pressure with fine glass beads before being clear anodized. Finally, all logos and labels are laser-etched into the metal, so will not fade or rub off!
A beautiful case would be of little use if it did not help to improve system performance. The entire M1 chassis is electrically conductive, and acts as an ideal EMI/RFI shield. Care was taken to ensure a low impedance ground path between every mechanical component of the chassis and the power ground plane, which extends to the Custom Universal Power Supply Unit (see below) and earth ground. The signal ground is kept isolated from this protective shield, and all these features together contribute to the ultra-low noise floor and freedom from external interference.
Superior USB Data Cable
As USB is the only connection which accepts all the high-resolution digital audio formats which the Sonoma M1 supports (PCM up to 32‑bit/384 kHz and DSD64/DSD128) it was vital to ensure that there was no compromise in performance when using this connection. Consequently, in collaboration with Straight Wire Inc., we developed the supplied USB cable. Featuring gold-plated connectors and a silver-plated data path, it is the ideal connection between your digital music source and the Sonoma M1 system.
Custom Universal Power Supply Unit
It should, perhaps, be obvious that stable, clean power is essential to the operation of any audio circuit. For the M1 we have opted for a two-part power solution in order to keep as much noise as possible out of the sensitive audio path.
The first-stage involves a custom-designed, universal voltage, outboard switch-mode power supply unit. While it may look similar to the units provided with laptop computers, etc., internally it is very different. To avoid any headroom issues, our device is capable of delivering about 3.5 times the maximum power the amplifier is designed to draw under steady-state conditions. In addition, the unit uses a fixed frequency-switcher (operating at over 85 kHz) to avoid the possibility of the switching-frequency dropping into the audio band as the power draw changes (i.e., the switching-frequency remains above 85 kHz under all operating conditions!). This unit is designed with improved internal filtering to yield extremely low noise and ripple (<50 mV peak-to-peak). It is connected to the amplifier via a custom, high-quality, shielded cable (with braided cover to match the headphone cable), and, to ensure an ideal connection, it is fitted with a high-performance, Switchcraft locking DC power connector. A massive ferrite core filter has been added to the DC cable to reduce EMI.
Inside the amplifier, all audio circuitry is supplied by multiple stages of ultra low-noise, high-current linear regulators from Analog Devices. Isolated power regulation stages are used for both analog and digital sections as well as high- and low-level circuit stages.
This two-part power solution is expensive, but, we feel, necessary to meet the design goal of delivering audio performance without compromise.