Any interest in a DIY MSX hardware group

Página 3/7
1 | 2 | | 4 | 5 | 6 | 7

Por Grauw

Ascended (10306)

Imagen del Grauw

25-07-2015, 11:22

I guess PS/2 in stead of USB because the protocol is much simpler to decode? Finding nice keyboards and mice that still support PS/2 can be a bit of a challenge nowadays though, if you don’t want to go for the first el cheapo thing you find. I ended up getting an expensive USB to PS/2 converter for my Zemmix Neo.

Por anonymous

incognito ergo sum (116)

Imagen del anonymous

25-07-2015, 20:59

A question guys, from your point of view what would be a feasible configuration for a hypothetical new MSX being 100% compatible? IMHO to guarantee the compatibility is a must if not all of this has no sense at all.

There are many opinions but I'm quite purist about MSX so for me a new design should be as close as possible to the MSX standards, I mean including a real Z80 CPU and other MSX IC's if this were possible and logical. I know that maybe the FPGA usage would be mandatory to replace some parts, the MSX IC stock is not eternal...

Por RetroTechie

Paragon (1563)

Imagen del RetroTechie

26-07-2015, 01:27

PAC wrote:

Many times we have discussed about a new possible MSX computer and blah, blah, blah. Why not do we use this thread to discuss about a credible and feasible proposals once and for all? What people want? Don't get me wrong, I'm not criticising the existing ones like 1ChipMSX (or clon), Orbit, MSX Arm or M5X, etc but we know all that this kind of products are not normally so popular among us and we should ask ourselves why. Distrust? Price? Compatibility? Requeriments? Specs? Necessity?

A combination of end-user price, required amount of diy, features, and availability. If it runs most MSX software, cheap enough, ready to go out of the box, and available for sale, people will buy. And keep doing so.

1chipMSX'es are only expensive because no-one builds them anymore. Similarly there were a few runs of machines based on 1chipMSX hardware, they're expensive because very limited # built, relatively high price for a small batch, and not available after a short while. General purpose FPGA boards like Altera DE1 are relatively affordable, but require diy to add things like joystick ports, cartridge slot(s) etc. Gr8bit - very much diy, too big, too expensive. Orbit: still in design phase? Or have they moved onto testing on 'final' hardware? Again: not available, likely too expensive because everything & the kitchen sink is included.

Like PAC: not criticizing those builders, just some of the reasons people aren't buying. If...

megatron wrote:

The aim is to get people into designing Z80/MSX hardware and learning about how to build a microcomputer system on the way.

Then I think a reasonable setup would be:
Discrete Z80 (CMOS version, pick speed according to system size & what you're aiming for)
One or two 512KB SRAMs, or a pair of 4M*4 bit DRAMs
A single biiig Flash ROM (AM29F040 or similar)
Discrete PSG + YM2413 (and perhaps other soundchips)
V9958
A large CPLD or small FPGA to glue everything together
IC housings (DIP / PLCC / Flat Pack) etc chosen depending on preferences & availability.
Analog RGB output, cartridge slot(s), PS/2 keyboard
Primarily aimed at MSX users, 100% software compatible
Doable by 1 person if enough time is invested

If the aim is to put an affordable, brand new MSX into as many people's hands as possible, then:

A big FPGA (in case of Altera: 15k+ LE's) that includes Z80, glue logic, soundchips, and (highly preferable) the VDP as well.
Analog RGB output and one of VGA or HDMI (preferably the latter)
Optional: composite video output
Cartridge slot(s) optional (to reduce cost / size / power / IO voltage issues)
USB keyboard (and if necessary, a small microcontroller to handle the USB interfacing), or even a custom made case that includes keyboard
Optional: a Raspberry Pi style parallel port for hooking up diy projects, sensors, motors, LEDs, LCD displays, Arduino shields etc.
If possible, aimed at a larger group of people than just MSX users. 100% software compatibility not required as long as most software works fine.
End user price: ~$100, $150 maximum
This would be a crowd-funded, multi-person team, >1 year effort I think.

Por hit9918

Prophet (2911)

Imagen del hit9918

26-07-2015, 01:44

@PAC,
compatibility is about the bios used and the slot layout.
troublemakers are: subROM in slot 0 or RAM in slot 0

Subslot-RAM:
it is best when the subslot is already toggeled to RAM when clueless MSX1 code toggles only the primary slot.
The TurboR has RAM in subslot 0. Good with secondary register powering up with 0, so there is RAM.
But I wonder whether the bios ROM search in pages 0/1 leaves the register in subslot 3?
Would be best if it would leave the search with register back to subslot 0.

About FPGA
whether it is real machine cannot be answered formaly, it is a question of practice.
The z80 is known in detail to hordes of people, the 9958 is not.
PSG PPI and secondary slot registers are simple logic.

Por hit9918

Prophet (2911)

Imagen del hit9918

26-07-2015, 01:49

@RetroTechie, the PSG is primitive and known in detail, it is purely digital till the DAC.

Por RetroTechie

Paragon (1563)

Imagen del RetroTechie

26-07-2015, 04:26

Choose a suitable slot layout + what ROMs to use, and most software will run. Even on 1chipMSX, for example the infamous POKE -1,xx is used, and few MSX users would complain about that. So this is a non-issue.

When I say "compatible", I mean "does software have the exact same effect as on the original/real hardware?". For the Z80, several FPGA implementations exist that are damn near perfect from a software point of view. If not perfect, and/or cycle-accurate as well. So no reason you can't include an FPGA-implemented Z80 and still have very true-to-the-real-thing behavior.

For simpler/older soundchips like PSG / SCC / YM2413 it's a similar story. What exists is more than good enough, and imperfections are not the kind that software trips over. Merely sound quality or accuracy issues - similar to analog problems like mixing volumes, sound clipping, etc. Again: no reason to stick to discrete chips if it can be avoided.

FPGA version of the MSX1 VDP is similarly accurate as Z80 / PSG etc. But MSX2/2+ VDP's are another beast. Both V9938 and V9958 implementations still have some kinks. Most MSX2/2+ software will run fine, but not always. Which could be a reason to include a V9958 in a new MSX design. Personally I think existing FPGA implementation V9938 is very good, V9958 implementation is good enough, and both are still being improved by people like KdL.

So for nostalgic reasons, or (perhaps) to provide a nice "learning system", including these discrete IC's is okay (as long as working specimens can be found!). But to obtain a low-cost, modern 'MSX for the masses' ? No way, go the all-included FPGA route.

Por hit9918

Prophet (2911)

Imagen del hit9918

26-07-2015, 04:48

Quote:

Cartridge slot(s) optional (to reduce cost / size / power / IO voltage issues)

I suggest a rear slot where an expander can be plugged without ruining the look and feel.
A cost-effective all-FPGA machine still should have the ability to plug things.
What if the user has 9990 + moonsound and can't plug it -> that is not cost effective.

A brainstorming for a project with discrete chips. openmsx machine file below.
9958
z80
PPI
AY
RTC
Are all chips needed for this machine available? RTC?
It has 64k RAM to avoid mapper and 1 external slot to avoid subslot gear for subROM, just to brainstorm a minimum MSX2 board.

Well and the title says "MSX 3" Smile A board with fast z80 is the main deal, all the great 9990 and moonsound can be plugged. (I couldn't model a fast z80 in the emu file)

Hit quote to get at the ascii file without errors.

<?xml version="1.0" ?>



  
    Realist
    <code>MSX3 Mini PPI</code>
    2015
    
    MSX2
  

  

  

    
      true
      
        16000
      
      
      jp_ansi
      true
      true
      <code locks>false</code>
      false
    

    
      AY8910 
      
        21000
      
      
      
    

    
      hb-f5.cmos
      
      
    

    
      V9958
      128
      
      
    


    
        
          
            hb-f5_basic-bios2.rom
            0081ea0d25bc5cd8d70b60ad8cfdc7307812c0fd
          
          
        
    
    
        
        
          
            hb-f5_msx2sub.rom
            b8e30d604d319d511cbfbc61e5d8c38fbb9c5a33
          
          
        
    
    
            
          
        
    
  


Por Grauw

Ascended (10306)

Imagen del Grauw

26-07-2015, 15:17

An FPGA-based MSX with discrete V9958 and YM2413. The FPGA will fulfill the role of the MSX-ENGINE including Z80, PSG, etc., while having the VDP and FM chip discrete for perfect, accurate and reliable video and sound. Then, add an R800 to the FPGA. So that there’s finally a turboR-based MSX clone available.

Include a casing that’s super pretty and compact (like Zemmix Neo Wink). Include external USB keyboard and mouse support and RGB output, ideally also HDMI (but may be difficult with discrete V9958). Cost should be less than €300.

Oh yes, and make sure the audio signal path is high quality and not noisy like the OCM.

Por Fabf

Champion (264)

Imagen del Fabf

26-07-2015, 20:36

RetroTechie wrote:

Then I think a reasonable setup would be:
Discrete Z80 (CMOS version, pick speed according to system size & what you're aiming for)
One or two 512KB SRAMs, or a pair of 4M*4 bit DRAMs
A single biiig Flash ROM (AM29F040 or similar)
Discrete PSG + YM2413 (and perhaps other soundchips)
V9958
A large CPLD or small FPGA to glue everything together
IC housings (DIP / PLCC / Flat Pack) etc chosen depending on preferences & availability.
Analog RGB output, cartridge slot(s), PS/2 keyboard
Primarily aimed at MSX users, 100% software compatible
Doable by 1 person if enough time is invested

Something like that ? LOL!

Por mars2000you

Enlighted (6097)

Imagen del mars2000you

26-07-2015, 21:27

Fabf, you are the best ! Smile

Página 3/7
1 | 2 | | 4 | 5 | 6 | 7