A project dedicated to create an FPGA implementation of the Philips CD-i to be usable for the MiSTer FPGA project. As every Philips CD-i player has a different hardware, this project focuses on reverse engineering the "Mono I" PCB. This mainboard is used in models like the 210/00, 210/05 or 220/20.
This repository is very experimental! Use at your own risk!
The first games are booting. Expect a certain amount of bugs!
Place cdi200.rom as boot0.rom in /media/fat/games/CD-i.
Place zx405042p__cdi_slave_2.0__b43t__zzmk9213.mc68hc705c8a_withtestrom.7206 as boot1.rom next to it.
The save files containing the NvRAM is compatible with the CD-i emulation of MAME.
Digital gamepads, Analog gamepads and mice are supported for use with this core. To play a title, load a CD and press on the play button at the start screen. CD images can be stored as CHD or CUE/BIN format.
Core Utilization:
Logic utilization (in ALMs) 13,520 / 41,910 ( 32 % )
Total registers 15626
Total block memory bits 630,471 / 5,662,720 ( 11 % )
Total DSP Blocks 66 / 112 ( 59 % )
- Find a better solution for reducing CPU speed
- Fix regression: Audio hiccups during Philips Logo in Burn:Cycle
- A workaround is CPU overclocking
- Investigate mysterious non loading behavior
- Investigate input responsiveness (skipped events?)
- Investigate screeching sound effect in the menu of "Golf Tips"
- Fix hang on audio track stop or change in media player
- Investigate "Gray border glitch" at the top of "Myst" gameplay (seems to be only one plane)
- Fix reset behaviour (Core is sometimes hanging after reset)
- Add auto start of titles using front panel "Play" button
- Find a way of detecting Audio CDs
- Investigate desaturated colors / low contrast in "Photo CD Sample Disc"
- Probably fixable with 16-235 to 0-255 scaling
- More investigation needed
- Find a solution for the video mode reset during system resets
- The ST flag is the issue here, causing a video mode change
- Add SNAC support (IR remote + wired controller)
- Refurbish I2C for the front display and show the content as picture in picture during changes?
- It might not even be required at all.
This core is tested with these ROMs:
2969341396aa61e0143dc2351aaa6ef6 cdi200.rom
3d20cf7550f1b723158b42a1fd5bac62 zx405042p__cdi_slave_2.0__b43t__zzmk9213.mc68hc705c8a_withtestrom.7206
Due to legal reasons, these files must be sourced separately.
This MiSTer core would've probably never been possible without the reverse engineering efforts of certain people. Thanks to CD-i Fan for the insights into his closed source CD-i Emulator. Also Thanks to MooglyGuy, which took on the task of implementing a CD-i emulator into MAME, which I used to analyse the program flow of the CD-i boot process.
- https://github.com/TobiFlex/TG68K.C
- https://opencores.org/projects/68hc05
- https://github.com/cdifan/cdichips
- http://www.icdia.co.uk/microware/index.html
- https://github.com/Stovent/CeDImu/blob/master/src/CDI/OS9/SystemCalls.hpp
The production quality of the CD-i hardware and the software running on it is sometimes questionable. For this reason, I've created a small list of some known quirks, someone might suspect of being caused by emulation errors but are also present on the real machine.
- Is the "Digital Video Cartridge" supported?
- No! Please stop asking!
- Potential development on the DVC might only start after everything works without bugs on the base machine.
- The map of "Zelda - Wand of Gamelon" has micro jitter during scrolling
- This also happens on real 210/05 hardware
- "Hotel Mario" seems to have the first samples of every ingame song repeated
- You have good ears as it is barely noticeable. This also happens on real hardware.
- Some earlier CD-i titles have both stereo channels swapped
- Yes, according to an internal memo from Philips there were manufacturing issues and some early players have the left and right channel swapped. This might explain discrepancies.
- One known quirk is inverted stereo on the "Philips Logo Jingle" of "Zelda - Wand of Gamelon"
- During the rotating transition in "Myst" there are glitched lines at the bottom
- This also happens to some extent on a real 210/05 hardware and is caused by a misplaced Video IRQ The video data is changed while it is displayed.
- When I load a save game in "Burn:Cycle" it plays a short and unclean loop of noise until I do something
- I thought that was a bug in the core at first too. However, it is actually like this on a real CD-i too.
- The music of Burn:Cycle sometimes has "pops" and "clicks"
- This game is special. It doesn't play music from CD like most games for this system would do.
- Small loops of sampled music are loaded from CD, stored in memory and randomly concatenated together to create the background music. These samples are sometimes not very "loopable" creating a pop at looping points.
- This issue is reproducible on a real 210/05 as well
- The music during the Philips Logo animation of Burn:Cycle has broken audio
- This issue is reproducible on a real 210/05 as well
- For some reason, it seems to be absent on other models with different hardware, like the 450/00
- The problem can be fixed by overclocking the CPU
- Burn:Cycle - Random flickery animated text in front of the "Psychic Roulette" credit card terminal
- This actually happens on the real machine. I also thought this might be a CPU speed issue, considering that the flickering disappears if the CPU is slightly overclocked.
- Flashback: The audio and video during the intro are asynchronous
- This curiously happens on the real machine as well and doesn't even depend on 50 or 60 Hz
- It seems to be an oversight by the developers when the game was ported to CD-i
- When dying in "Zelda's Adventure" the accompanying sound effect doesn't match the audio data on CD
- Good catch! This is a programming error which can be reproduced on a real CD-i 210/05 as well, which causes the audio playback to start one sector too late.
- The same phenomenon exists in the "Help cutscene" of "Zelda - Wand of Gamelon"
- It is not audible in that game, because of silence in the beginning
- The intro music of Zenith is played only on the left audio channel
- Yes, this happens on a real 210/05 too. To be sure, I've tested it as well on a 450/00. It seems to be an oversight by the developers.
