There were three official firmware releases by Sequential. r1.0 and r1.2 seem to be the most common, with users selecting one or the other based on if they care about the resetting LFOs or not. Click on the image(s) to the right to get the MS/LS *.bin files for a particular version.

While I haven't come across any evidence to support or deny it, there is a possibility that r1.3 was being developed by Sequential Circuits and never finished. Quoted here from Retrosynth Prophet VS Info:

If there did happen to be an official r1.3 it would be great to track down and work out the kinks.

I'll update this as more are discovered.

• r1.0:
  • In MIDI Mode 3, split programs with different channel numbers incorrectly steal voices across the split point.
    -- [Fixed in firmware ≥ 1.1]
• r1.0, r1.1:
  • Single waves cannot be stored correctly to a cartridge.
    -- [Fixed in firmware ≥ 1.2]
• r1.0, r1.1, r1.2:
  • Not technically a bug. Byte order of NRPN MSB and LSB are swapped from modern standards. This makes MIDI controllers like the BeatStep Pro, BCR2000, etc, not function correctly.
    -- [Fixed in modified firmware ≥ 1.4]
  • Loop count of "Get Basic Waveform (Enter + "Edit Waveform")" off by 1. Modwheel destination is ignored.
    -- [Fixed in modified firmware ≥ 1.4]
  • Loop count writing custom characters to the LCD (for the Voice Panning screen) too long by 24! Character table is 40 bytes long but loop iterates 64 times, writing values from a section of code describing Mod. Matrix pointers. Looks like it was just a hex vs dec typo (0x40 is 64).
    -- [Fixed in modified firmware ≥ 1.4]
  • NRPN control of ARP RATE nonfunctional. Code doesn't exist.
    -- [Fixed in modified firmware ≥ 1.45]
  • NRPN control of SPLIT POINT nonfunctional. Code doesn't exist.

More to come...

• Mid 2018:
  • Disassembled, compared, and analyzed code binaries for r1.0, r1.1, and r1.2 .
  • Created patched ROM file from r1.2 that disables LFO synchronization to first keypress.
  • Released OS v1.3 to users of VS Yahoo group, Muffwiggler (now Modwiggler), and Gearslutz (now Gearspace).
• Early 2019:
  • Created an m68k assembly file of r1.2 in Easy68k that compiles identically to the factory ROM. No more hex patching!
  • LFO key reset is selectable by changing the 8th patch name digit.
  • MIDI NRPN MSB/LSB order is swapped to current standard. Previously, many MIDI controllers would not work correctly with the VS.
  • Discovered several factory NRPN control bugs (present in all original SCI versions). Input control of SPLIT POINT and ARP RATE do not work, as the code does not exist... the routine locations just return (m68k "rts").
  • Released OS v1.4.
• Late 2019:
  • Discovered that some of my early byte patching caused Wave #126 (Silent) to, well, not be silent. Fixed the bug.
  • Found numerous word-sized branch instructions that could be changed to byte-sized.
  • Found several chunks of dead, unreachable, or internal debugging code.
  • Added a new page in the LFO Select menu to choose key sync for LFO 1, 1&2, or none. This option is currently global.
    
  • Added a new page in the ARP Option menu to choose key sync. This option is global.
  • Fixed NRPN control of ARP Rate bug.
  • Added new [Enter + 5] function: Load default global parameters.
  • Reincorporated Output test in diagnostics.
  • Release OS v1.45.
• 2020-ish/2021-ish:
  • Began building a complete memory map in earnest.
• Currently (late 2022):
  • Finally started some source control of the assembly code on GitHub.
  • Fixed selectable LFO reset bug in unison mode. It's possible that LFO 1 might always reset regardless of menu setting.
  • Improved diagnostic menu selection & switch debouncing. Dirty switch contacts or OLED users might experience the menu going nuts.
  • Added "EXIT" option in diagnostics. No need to power-cycle to return to normal operation.
  • Current code has ~1kb of free space, including additions and full diagnostics. Lots of wiggle room!
  • v1.5 in progress:
    • Testing:
      • Digital filter range tuning in diagnostics. Looking into scale as well.
      • Velocity to envelope attack rates. Will make new patches incompatible with older versions.
      • Per-patch LFO sync option. Will make new patches incompatible with older versions.
      • Modwheel to filter modulation. Will make new patches incompatible with older versions.
    • Wishlist:
      • "Freerunning" joystick option.
      • Full envelope circle mode from 0>3>0, instead of the jump using 0->3 or the backtracking using 0<>3.
      • Internal NV Ram expansion with selectable banks. The display would then look like this for Int., Cart., and Link:
        
        

• -- (r1.3) Initial Release, 2018.
  - Patched version of SCI r1.2. Has all the same functionality without the LFO reset.
  **WARNING** -- THIS VERSION WAS NEVER SOLD, AND THE ONES AVAILABLE ONLINE SEEM TO HAVE BEEN RIPPED FROM A POSSIBLE IN-PROGRESS WORKING VERSION FROM THE VS YAHOO GROUPS. DO NOT BUY A "SHALEBRIDGE v1.3" FROM ANYWHERE!!



Tools

Here are links to some of the tools that were invaluable in working on this project:

• IDA Pro (Interactive Disassembler) - This was my starting point for disassembling the binaries, until...
• Ghidra - When this was released, it totally blew IDA out of the water for me. It's now my main archiver for documenting the original OSes.
• Virtual VS - A VS emulator that uses the original OS binary images. This came in very handy testing binary builds without burning new ROMs or swapping EEPROMS. It also allows me to work on code and test when I'm away from a hardware VS. It isn't 100% 68k emulation accurate though, so some things that worked fine emulated caused weird problems in hardware. Since the author never responded to my inquiry for the source code, I used part of the Virtual VS core...
• Castaway 68k/Atari Simulator - The simulator used in Virtual VS. Used sparingly to quickly dump register values in sections of code.
• Easy68k Editor/Assembler/Simulator - I used the main program to adapt the disassembled code, and to do all current development. I also use EasyBin to output the split assembled program into an MS/LS pair. Working on automating some steps using the source code on GitHub and building a full cycle-accurate VS simulator with register monitoring.
• Eprom Emulator - Before designing my own split-EPROM emulator, I used two of these piggybacked, with modified firmware to automatically upload the MS/LS files to the correct device. I'll probably upload my design to the site some time in the future.