Changeset c523a67 in buchla-emu


Ignore:
Timestamp:
12/31/2017 08:37:44 PM (7 years ago)
Author:
Thomas Lopatic <thomas@…>
Branches:
master
Children:
49efb91
Parents:
f0af168
Message:

Updated FPU info.

File:
1 edited

Legend:

Unmodified
Added
Removed
  • readme.txt

    rf0af168 rc523a67  
    276276-----------------
    277277
    278 Here's what we emulate:
     278Here's what we currently emulate:
    279279
    280280  * Motorola 68000 CPU. This is actually the Musashi CPU emulator by
     
    283283      https://github.com/kstenerud/Musashi
    284284
     285  * Intel 82716: Video chip.
     286
     287  * Epson SED1335: LCD controller.
     288
     289  * Western Digital WD1772: Floppy disk controller.
     290
     291  * Rockwell R65C52: Serial console and MIDI ports.
     292
    285293  * Motorola MC6840: Timers.
    286294
    287   * Rockwell R65C52: Serial console and MIDI ports.
    288 
    289   * Epson SED1335: LCD controller.
    290 
    291   * Intel 82716: Video chip.
     295  * Unknown item #1: A program running on a microcontroller. It
     296    converts the analog signals from the Buchla's controller pads to
     297    digital values.
     298
     299    Neither the program, nor the microcontroller are known, but the
     300    protocol (known from the firmware source code) is pretty simple
     301    and self-explanatory.
     302
     303The next development milestone will hopefully emulate the following
     304additional components:
    292305
    293306  * National Semiconductor LMC835: Equalizer.
     
    297310    to the above equalizer chip.
    298311
    299   * Western Digital WD1772: Floppy disk controller.
    300 
    301   * A few LEDs.
    302 
    303   * Item X: A program running on a microcontroller. It converts the
    304     analog signals from the Buchla's controller pads to digital
    305     values.
    306 
    307     Neither the program, nor the microcontroller are known, but the
    308     protocol (known from the firmware source code) is pretty simple
    309     and self-explanatory.
    310 
    311   * Item Y: The actual sound generator, referred to by the firmware
    312     source code as "the FPU." This could actually be two chips:
    313 
    314       1. One chip, maybe a DSP, for generating the 15 different
     312  * A few indicator LEDs.
     313
     314  * Unknown item #2: The actual sound generator, referred to by the
     315    firmware source code as "the FPU." This is the biggest unknown so
     316    far. Judging from the firmware source code it consist of two
     317    parts:
     318
     319      1. The function generator that generates the 15 different
    315320         parameter envelopes for each of the 12 voices:
    316321
     
    327332           - 1x Output signal stereo location.
    328333
    329            - 1x "Dynamics." (TBD - currently not emulated.)
     334           - 1x "Dynamics" - whatever that is.
    330335
    331336         Over time, the chip interpolates between the points of the
    332337         envelopes drawn in the MIDAS VII instrument editor.
    333338
    334       2. A second chip for the actual sound generation. This is likely
    335          a DSP.
     339      2. The digital oscillator.
    336340
    337341         XXX - Details to be filled in.
     
    343347         Obviously, the emulator does everything digitally.
    344348
    345     This "two chip" hypothesis would be in line with the "four
     349    This "two part" hypothesis would be in line with the "four
    346350    computers" marketing claim from the Buchla 700 marketing copy. The
    347351    four "computers" would be the Motorola 68000, the microcontroller
    348     that does the A/D conversion of the pad inputs, plus the two CPUs
     352    that does the A/D conversion of the pad inputs, plus the two parts
    349353    that constitute "the FPU."
     354
     355    The firmware source code archive indicates that the FPU is based
     356    on micro-programmable hardware. We recently ran this by Lynx, the
     357    developer of the firmware, who generously agreed to meet up with
     358    us in Oakland, CA. While he never worked on the FPU and thus was
     359    not familiar with its implementation details, he was able to
     360    confirm that the FPU was based on AMD's Am2900 family.
    350361
    351362If you have access to an actual Buchla 700, please do contact us. It
     
    353364
    354365If your Buchla is non-functional, this is also fine. We might be able
    355 to gain some insights from reading out the FPU microcode PROMs.
     366to gain some insights from reading out the FPU microcode PROMs or from
     367figuring out how the FPU chips are wired together.
Note: See TracChangeset for help on using the changeset viewer.