What turned up when we took it apart
A synthesizer inside the Z80
The cartridge has no sound chip: it has an 8-bit DAC at 0x4000, and the program computes the note sample by sample (0x5767): oscillator, two-stage resonant filter, amplifier, three ADSR envelopes and an LFO.
Measured in openMSX: the rewritten engine gives the sixteen tones byte for byte.
The pitch is a loop of cycles
PLAY sends the sample to the DAC byte by byte and waits between them: every key has one of thirteen routines and a delay D (0x5DA9), and routine r costs 59 + r + 14 × D cycles. One key at a time.
Measured: 254.37 cycles on average at A3, 254 and 255.
The sample lives in the computer's RAM
The cartridge has no RAM: 0x4105 puts the computer's RAM at 0x8000-0xBFFF to synthesize and play, and 0x410D brings the cartridge back to draw.
Measured: A8 = 0xF4 during SYN.
The noise comes from the BIOS
Wave 16 and the LFO's sample and hold read the BIOS ROM from 0x0000 (0x5AD4): tone 14 sounds different on every computer.
From the code.
The DAC starts without a thump
The first interrupt routine (0x4118) only raises the DAC from 0 to 0x80, one step at a time, 129 times, and then hands over the hook.
From the code.
TONE, LAYOUT and SAMP
Three tape files: the sixteen tones with their sixteen waves (4,608 bytes), the note of every key (176) and the sample with its loop page (up to 16 KB).
From the code, and the tape that came with the cartridge confirms it: seven SAMPs.
The second page of the menu
SHIFT changes the menu: copy a tone with its wave, list the sixteen, print the screen and LAYOUT, where every key gets its note.
From the code; these screens are not built.
What nobody calls, and the mark
Ten pieces of code nobody calls, among them a graphics loader driven by a table that does not exist (0x49CD) and three useless ld hl in the envelopes; RC741 86 5 31 at 0xBF62, and Konami's hidden mark at the end: シンセサイザー, RC-741 (Manuel Pazos's format).
