After much pondering, I believe that our first model should be packaged as a standalone device, instead of a Euroack module. There are many reasons for it, some marketing, and some technical.
First, the device that we are considering now shares very little with the Eurorack spirit, beside its modularity. It is much more digital than analog, it might use hi-fi components for audio conversion, and it would be priced twice as much as the most expensive Eurorack module currently available on the market. Clearly, we have some impedance mismatch there.
Second, what we are planning is not just a synthesizer anymore. As our design evolved, it morphed into some kind of multi-channel audio interface tightly integrated with powerful digital signal processing engines. In other words, audio synthesis is just one of the many applications for something like the ISHIZENO i8. But the main application might be found in sound production.
Therefore, what we are building could be seen as a portable crossover between an Apogee Symphony I/O 16×16 ($3,995) and a UAD-2 Satellite Thunderbolt ($5,999), with CV inputs and outputs for direct connectivity to Eurorack modules. If something like that could retail for a third of an Apogee+UAD combo, there might be a reasonable market for it, which is an attractive prospect.
Of course, I’m not going to pretend that we could build such a device, not to mention bringing it to market. But if I had my way, this is the kind of device that I would like to use. Something portable that I can put in my backpack, with excellent sound quality, and plenty of processing power. And most importantly, something that can be hacked with software, because it is built around a set of tightly packaged FPGAs.
From a mechanical standpoint, the device would be made of a metal frame, a faceplate mounted on a front-side horizontal hinge, a backplate attached to the frame, and two wooden side panels carved in black oak. This design would provide easy access to the submodules, while allowing us to reuse the Mars PM3 base board for most of the connectors that would be exposed on the backplate:
- 12V Power
- USB 2.0 host interface
- USB 3.0 device interface
- Gigabit Ethernet RJ45
- Mini HDMI (for PCIe or LVDS)
To these, we would simply add our MD68 connector for the 16 audio inputs and 16 audio outputs, and a 1⁄4 inch connector for headphones. The device would measure 314 × 128 × 50 mm. Therefore, it would have the same length as a MacBook Pro 13″. Nevertheless, it would be as wide as a Eurorack module is tall, and with sideboards removed, it would be 58-HP wide. Therefore, repackaging our device into a Eurorack module should not be too difficult, should we decide that there is a market for this form factor.
Why do we keep this constraint to remain compatible with the Eurorack format? To make sure that we have constraints. Otherwise, we will end up with one of these “battleship” synthesizers that look really impressive in pictures, but end up being impractical. Instead, by forcing ourselves to be no wider than a Eurorack module and no longer than a MacBook Pro, we keep our project under control.
Within these specs, we can fit 2 horizontal rows of 4 SO-DIMM modules, leaving quite a bit of room for other components on the backplane. With such a design, we would have three main boards within the enclosure: the faceplate backplane on which 8 SO-DIMM submodules would be mounted, the base board on which the core fabric would be installed, and the audio board with the audio converters, the MD68 connector, and the headphones connector. The audio board itself might host several breakout boards for the actual converters if we manage to outsource this piece of our design to companies like Ackolabs and/or Twisted Pear Audio.
From a connectivity standpoint, the backplane would be linked to the base board through an FMC connector, which is a standard format for FPGA mezzanine cards (hence the acronym). As a result, the pinouts for both our backplane interconnect and our submodule interconnect would be already defined, according to the Enclustra Mars specifications. The only thing that we would have to create from scratch would be the interconnect for analog modules, and the interconnect between the backplane and the audio board, but these should be relatively simple.
Conclusion: this latest iteration is much more ambitious than anything we’ve gone through so far, because it relies on FPGA components for both the fabric and the digital submodules, and because it plans on using Hi-Fi components for audio conversion (both ADC and DAC). By the same token, it is much simpler than anything we’ve considered up until now, because most of the hardware complexity is encapsulated into off-the-shelf components that can be sourced at a reasonable cost and are designed with an Open Source approach. Therefore, I think we have a winner here…
I like it. I really like it…