Retro-bit SEGA USB and Bluetooth Genesis, Saturn and Dreamcast accessories revealed!

Discussion in 'Sega Dreamcast Development and Research' started by Anthony817, Jan 10, 2018.

  1. TriMesh

    TriMesh Site Supporter 2013-2017

    Joined:
    Jul 3, 2008
    Messages:
    2,188
    Likes Received:
    663
    Yeah - but it looks exactly like the SLS USB Saturn pad I have, even down to the mismatched wire color. In fact ,the only thing that (to me) doesn't look original Sega is that pad that's been painted blue - which looks like crap.

    Sure, I would love it if they were actually making new controllers that were of equivalent quality to the originals - but the stuff I've seen so far from RetroBit have been anything but that.
     
  2. Plavix

    Plavix Rising Member

    Joined:
    May 17, 2012
    Messages:
    74
    Likes Received:
    9
    Whats going on with the wireless Dreamcast Controllers? Been waiting forever
     
  3. Anthony817

    Anthony817 Resolute Member

    Joined:
    May 12, 2010
    Messages:
    947
    Likes Received:
    399
    Well, the Maple bus is notoriously high bandwidth so they have to get it right, and they also won't be releasing the Saturn and Mega Drive wireless pads until Summer. So I reckon they aren't even close to finalizing the DC design yet, and they have to figure out how to transfer game saves through the air without getting corrupted.
     
  4. Plavix

    Plavix Rising Member

    Joined:
    May 17, 2012
    Messages:
    74
    Likes Received:
    9

    That's a bummer:(

    I hope the wait isn't too long?
     
  5. megavolt85

    megavolt85 Robust Member

    Joined:
    Jan 8, 2015
    Messages:
    272
    Likes Received:
    378
    2 Mbps = 256 kilobytes per second
    for current technology is low speed device
     
    LuizNai and americandad like this.
  6. Anthony817

    Anthony817 Resolute Member

    Joined:
    May 12, 2010
    Messages:
    947
    Likes Received:
    399
    Yeah not fast by today's standards, but for the time it was pretty decent. I still think that they will eventually come out with a wireless controller, but they have to get it right. And I think saving games wirelessly is going to be at least a stumbling block for them.
     
  7. Greg2600

    Greg2600 Resolute Member

    Joined:
    Jun 23, 2010
    Messages:
    913
    Likes Received:
    88
    My question is whether the DC controllers will correct the infamously bad analog control on racing games like Daytona USA or Midnight Street?
     
  8. Xerxes3rd

    Xerxes3rd Rapidly Rising Member

    Joined:
    Sep 1, 2016
    Messages:
    94
    Likes Received:
    33
    Not super fast, but when the host (Dreamcast) sends a Maple request to the peripheral, the peripheral must begin its response within 1ms. I realize that in “hardware time” 1ms is a long time, but when you add RF and other protocol layers on top of it, you could have issues.

    This is likely why the original DreamConn had “emulated VMUs” inside the dongle- the dongle itself was emulating a full peripheral with attached sub-peripherals, then would proxy the button inputs, jump pack commands, and VMU screen buffer to/from the wireless controller.
     
    Anthony817 likes this.
  9. TriMesh

    TriMesh Site Supporter 2013-2017

    Joined:
    Jul 3, 2008
    Messages:
    2,188
    Likes Received:
    663
    That's all pretty tractable though - another thing is that if the timing of the poll requests from the host is predicable then the adapter can just work out when it needs to request the data over the radio link to have it available in time for the next host poll (and if the host polling isn't predictable, then latency is not likely to be much of an issue anyway...)
     
  10. Xerxes3rd

    Xerxes3rd Rapidly Rising Member

    Joined:
    Sep 1, 2016
    Messages:
    94
    Likes Received:
    33
    It’s predictible when the host polls for button/stick inputs (60Hz), but everything else (saving/loading data, VMU screen writes, rumble, microphone) isn’t. If you take longer than 1ms to start your reply, the host resets the bus.
     
  11. TriMesh

    TriMesh Site Supporter 2013-2017

    Joined:
    Jul 3, 2008
    Messages:
    2,188
    Likes Received:
    663
    Yeah, I'm thinking more of how to minimize overall latency after you have implemented a local server system like that, since the quick and dirty solution of just running a polling cycle behind is going to be pretty unacceptable for some games. Rumble isn't too much of a problem - you just need to generate the acknowledge locally and then send the request to the controller. Stuff like LCD screen updates can be done on a best effort basis since it just needs to be "fast enough".

    It would be interesting to see if you could handle the Maple protocol entirely in software given a fast MCU like a Cortex-M4, although the easier approach would probably to use something like a CPLD to decode the wire protocol and pack it up into at least bytes.
     
  12. Anthony817

    Anthony817 Resolute Member

    Joined:
    May 12, 2010
    Messages:
    947
    Likes Received:
    399
    Honestly, I wonder if they are just going to do similar to the DreamCon anyways? Built in screen and virtual VMU pages? It would solve a lot of the issues and might make the price a little bit more but a small monochromatic screen and some cheap flash memory on board shouldn't drive up the cost too much.
     
  13. Xerxes3rd

    Xerxes3rd Rapidly Rising Member

    Joined:
    Sep 1, 2016
    Messages:
    94
    Likes Received:
    33
    It’d be pretty hard to keep up if you bit-bang it. I wrote some software using a PSoC5 that decoded Maple packets, and the resources of the CPLD-ish part of the PSoC filled up quickly. It worked well, even though it didn’t have DMA so the CPU was reading byte-by-byte.
     
    TerdFerguson likes this.
  14. TriMesh

    TriMesh Site Supporter 2013-2017

    Joined:
    Jul 3, 2008
    Messages:
    2,188
    Likes Received:
    663
    Yeah, I was thinking in terms of doing something really dirty like using two cross-connected SPI interfaces running in slave mode with the Maple bus signals going into SCK on one interface and MISO on the other. You would wind up with the data being spread across the two receive buffers, but it should be possible to sort them out in software.
     
    TerdFerguson likes this.
  15. Xerxes3rd

    Xerxes3rd Rapidly Rising Member

    Joined:
    Sep 1, 2016
    Messages:
    94
    Likes Received:
    33
    I wrote the Verilog to match how Sega described it in the Maple patent, which is similar- they feed the lines into two shift registers in the same fashion you’re describing. A lot of CPLD space ended up getting taken up by the code to perform start & end pattern detection (although I’m a novice at Verliog, so it could’ve been my fault).

    My only concern with using SPI peripherals is that there’s always an odd number of bytes, since the final byte is a checksum, and I’m not sure if the SPI peripheral would tolerate receiving only a nibble.
     
    TerdFerguson likes this.
  16. megavolt85

    megavolt85 Robust Member

    Joined:
    Jan 8, 2015
    Messages:
    272
    Likes Received:
    378
    what model of clpd did you use?
     
  17. Xerxes3rd

    Xerxes3rd Rapidly Rising Member

    Joined:
    Sep 1, 2016
    Messages:
    94
    Likes Received:
    33
    Sorry, I said CPLD, but meant the “UDB section” of the PSoC5, which is basically a very tiny CPLD.
     
    TerdFerguson likes this.

Share This Page