Best of Analog
Best of Digital
We are an instrument designer based in Palo Alto, California.
We are bringing the best of analog and digital together.
We are about to let you break new sound barriers.
Start with a 16×16 32-bit audio interface.
Add a powerful FPGA-powered synthesizer.
Wrap into a 16-channel control surface.
Sample analog audio and control voltage signals.
Patch and process them in the digital domain.
Reach unprecedented levels of polyphony.
Customize any of its printed circuit boards.
Develop user interfaces as web applications.
Open design hardware, open source software.
ISHIZENO i8 is currently under development. We are now in the process of gathering requirements and validating some of the core technologies that we are planning to integrate into the system, like the Zynq UltraScale+ MPSoC, or the XMOS XEF200 multicore microcontroller. If we do not hit any major roadblock, we hope to complete a first prototype in 2016, and a small production run the year after.
ISHIZENO i8 is designed by Ismael Ghalimi, who is neither a musician, nor an electrical engineer. Instead, he applies his expertise with software development and his unadulterated perspective on audio synthesis and sound processing to create something new and radical.
16×16 audio interface with 32-bit sampling at 768 kHz.
Built-in digital mixer and audiophile headphones amplifier.
32 × 32-bit audio sampling at 768 kHz.
32 × 16-bit CV sampling at 200 kSPS.
16 analog CV inputs and 16 analog CV outputs.
22 CPU cores, 3,528 FPGA DSP slices.
Use any smartphone, tablet, or PC as user interface.
Connect to the system through Ethernet, USB, or WiFi.
9 motorized faders with dual 64-LED RGB bargraphs.
Shuttle wheel, 24 RGB illuminated knobs, 78 switches.
324 illuminated keys sensitive to pressure and velocity.
8 controls and 16 drum pads. Playing and sequencing nirvana.
The ISHIZENO i8 is built in a modular fashion with customizable printed circuit boards: backplane, audio ADC, audio DAC, CV interface, channel controllers, headphones amplifier, and power supply. Above all, its design is optimized to reduce latency and jitter, while offering unprecedented levels of signal processing power and extensibility.
Each one of the 8 stereo inputs and 8 stereo outputs is driven by its own audio converter, and a dedicated set of FPGA I/Os. The AK5397EQ is used for analog-to-digital conversion (32-bit, 768 kHz, +127 dB SNR), while a discrete R-2R sign magnitude DAM is used for digital-to-analog conversion with 32-bit resolution and 3.072 MHz oversampling.
Every audio converter is driven by dedicated sets of ultra-low noise linear power regulators, with 8 regulators used for each ADC converter, and 4 regulators for each DAM modulator.
The patching and mixing of audio and CV signals is handled in the digital domain through the FPGA. Every audio input and output and every CV input and output can be driven by dedicated sets of DSP slices, while remaining DSP slices can be used for the synthesis of digital polyphonic voices.
All synthesis functions, such as high-frequency oscillators, low-frequency oscillators, delays, filters, reverbs, and mixers are implemented in VHDL, with 32-bit encoding and 768 kHz sampling.
The faceplate includes 9 motorized faders (100 mm) with dual 64-LED RGB bargraphs (one per mono channel), 24 RGB illuminated knobs with SPST switching, one 55 mm shuttle wheel, and 78 illuminated switches.
Mixing is done in the digital domain for the 8 stereophonic audio outputs, and a virtually unlimited number of digital audio channels produced by the 3,528 DSP slices provided by the FPGA, and the 22 CPU cores offered by the Zynq device and XMOS microcontroller.
The ISHIZENO i8 can synthesize sounds from scratch, or sample up to 8 stereo audio channels. For this purpose, it embeds a studio-grade audio interface made of 8 independent analog-to-digital converters. Each converter is mounted on a 102 × 42 mm board, making the design fully upgradeable, and allowing different converters to be used in parallel for different channels.
By default, each audio ADC module is built around the AK5397EQ analog-to-digital converter, which supports 32-bit sampling at 768 kHz, with an ultra-low signal-to-noise ratio of +127 dB. Two balanced audio inputs are fed to the converter through input buffers and four LME49990 ultra-low distortion, ultra-low noise operational amplifiers.
In order to preserve the purity and richness of the original signal, each module includes eight power regulators: one LM78L15 for +15 V, one LM79L15 for -15 V, and six ultra-low noise ADM7150 linear regulators (left & right 5 V analog, left & right 5 V bias, left & right 3.3 V digital). Furthermore, all resistors are aerospace-grade components with an ultra-low tolerance of 0.01%.
The printed circuit board (PCB) used for each module is made of two layers only, with all analog and digital signals being routed on the top plane. The bottom plane is used for power and grounding, with fully-separated analog and digital sections connected directly under the converter. Traces are never thinner than 16 mils, while most analog input and power traces are 32 mils, ensuring very low impedance. Additionally, no via is ever used by more than one trace, and all traces are carefully rounded and follow a 45 degrees routing pattern.
Each module is connected to the backplane through a 62-position card edge connector with a low impedance pitch of 2.54 mm and a 500 V rating. When using the AK5397EQ analog-to-digital converter, all 19 digital I/Os made available by the converter are routed to the backplane's FPGA through the edge connector, while leaving 6 top-plane positions available for future upgrades, and using all 31 bottom-plane positions for grounding.
This design enables the FPGA to control all parameters of the converter, in a fully programmable fashion, including 12 types of audio data interface formats, sampling speed, digital high pass filter, overflow detection, mono mode, power down, and reset. Most importantly, the 3,528 DSP slices of the FPGA fabric can be used to implement highly sophisticated filters in the digital domain, with ultra-low latency, 32-bit precision, and 3.072 MHz oversampling or more.
In order to reduce jitter from the femto clock to the converter, the master clock signal is not routed from the backplane to the audio ADC modules through their card edge connectors. Instead, 8 coaxial cables of equal length are connected from the backplane's clock distribution section to the 8 modules, using U.FL ultra small surface mount coaxial connectors.
1 × tablet holding groove
1 × 55 mm Avago shuttle wheel
9 × 100 mm Penny+Giles motorized faders
9 × 100 mm dual 64-LED RGB bargraphs
24 × Bourns rotary encoders with 32 PPR and SPST switching
24 × knob illumination rings with 32 RGB LEDs
78 × illuminated switches
Xilinx Zynq ZU15EG MPSoC
Quad-core ARM Cortex-A53 MPCore up to 1.3GHz
Dual-core ARM Cortex-R5 MPCore up to 600MHz
Mali-400MP GPU up to 466MHz
Xilinx Virtex 16 nm FPGA fabric
597K logic cells
3,528 DSP slices
XMOS XEF200 multicore microcontroller
16 × real-time processing cores
Pulsar femto clock
8 × AK5397EQ analog-to-digital converters on individual boards
8 stereo channels
768 kHz sampling rate
+127 dB SNR
–108 dB THD+N
Buffers for line inputs
U.FL coaxial connector for master clock
16 × discrete R-2R sign magnitude DAM on individual boards
8 stereo channels
3.072 MHz oversampling rate
16-channel 16-bit analog-to-digital CV levels converters
16-channel 16-bit digital-to-analog CV levels converters
16 × 3.5 mm TRS analog CV inputs
16 × 3.5 mm TRS analog CV outputs
32 × RGB LEDs
Automatic CV output calibration
Converter temperature control
2 µm Ag gold-plated headphones sockets
Kensington security slot
100-240v IEC power connector
Device USB 3.0 connector
Host USB 2.0 connector
Device USB-C connector
Host USB-C connector
AVB Gigabit Ethernet RJ45 connector
Gigabit Ethernet RJ45 connector
Coaxial S/PDIF input connector
Coaxial S/PDIF output connector
Optical ADAT + S/PDIF input connector
Optical ADAT + S/PDIF output connector
2 × DB25 8-channel balanced inputs connectors
2 × DB25 8-channel balanced ouputs connectors
2 × XLR male audio outputs connectors for studio monitors Δ
2 × XLR male audio outputs connectors for studio monitors Σ
2 × BNC connectors for world clock input
2 × BNC connectors for world clock output
16 × SMA connectors for 16.3Gb/s GTH transceivers
6.35 mm TRS stereo headphones jack
SD Card port
Length: 382 mm (15.04")
Width: 282 mm (11.10")
Height: 82 mm (3.23")
The g3 is an optional companion to the i8.
324 keys and 16 pads sensitive to pressure and velocity.
8 customizable controls for mode selection and track transport.
While the module can be used with any USB controller, it is being developed jointly with a grid controller, the ISHIZENO g3. This grid is made of 324 keys, 16 drum pads, and 8 controls. Keys, pads, and controls are illuminated by RGB leds and are sensitive to pressure and velocity.
The controller is made of a black walnut frame and is controlled by an embedded Raspberry Pi 2. It also provides a longitudinal groove capable of holding up to 8 tablets in a semi-vertical position, and can be extended with left and right docks for hosting one or two tablets horizontally.
8 rows of 36 keys for channel level setting, 1 row for master output level.
8 rows of 36 keys for playing across 3 octaves, 1 row for sequencing.
8 rows of 36 keys for multi-channel sequencing, 1 row for note selection.
Anticipated retail price: $1,995
Want to learn more? Contact us at info at ishizeno dot com or follow our blog