Besides the Supers I remember both Summit and Kyra using FPGA. Reading the specs it seems Summit has a high samplerate while Kyra got a very high voice count. Is this a tradeoff one must make with this processor architecture?
Now there is another FPGA-based synth that looks interesting which is a prototype by Paula Maddox (formerly Modal) called Delta. Its currently 8 voices and has an analog signal path after the oscillators. It also supports wavetables which sounds really nice. There’s a video going through what’s currently in prototype on https://youtu.be/jpkybukyXss?si=9mPUO83KB1PcYXi0.
Not really. I think the main reason Peak / Summit is limited to 8 / 16 voices comes down to the analog signal path. The VCFs / filter pre and post distortion stages / mixer / VCAs are all analog.
Kyra is all digital and emulates the filters and everything on the chip, quite impressive for 128 voices with 8 parts. No idea how the FPGA is configured to allow for all if this, but many individual components (even filters) can be done with very little logic elements if there’s need.
The two main benefits of FPGAs (in my view):
Reconfigurable architecture. Meaning that the processor logic elements as well as memory are reconfigurable. This allows for a lot of flexibility when it comes to I/O (such as separate DACs for all voices) and hardware design (peripherals such as external memory).
Concurrency in signal processing. Since the chip is basically a collection of gates and flops, there is potential for parallelism in signal processing (multiple channels operating simultaneously in high clock speeds - this is more of a challenge with traditional CPU or DSP chips.
FPGAs are not a magic bullet and they are computationally inefficient compared with modern microcontrollers, but they lend to polyphonic synthesizers well.
Don’t forget Roland JD-XA, ?discontinued?. FPGA was a selling point to myself. It’s a bit polite, compared to my Super 6, but I won’t be selling the 'XA. I wish I knew more about electronics and processors, but saw an article comparing CPU, FPGA and DSP processors. Freds Lab Manatee uses the latter, not sure if there’s other DSP synths out there. From a photo (not opening it up), Super 6 has X6 voice cards, so they must be stereo, dual filter. It’s not clear if there’s a single, X6 or X12 DACs? Most likely X12 DACs, as the filter’s analogue. What other functions these cards perform is not clear. So the big question, is Moog Muse a bigger sound than Super 6? It’s a hard act to follow.
