Shop OBEX P1 Docs P2 Docs Learn Events
Board settings for Pnut.exe vs FastGUI? [solved --> Terminal window was not closed] — Parallax Forums

Board settings for Pnut.exe vs FastGUI? [solved --> Terminal window was not closed]

potatoheadpotatohead Posts: 10,261
edited 2019-11-24 00:33 in Propeller 2
I've been running examples and coding today. (finally got more than a quick look!)

FastGUI ran with the Rev B eval board out of the box. (nice work Eric) Someone here mentioned flipping a switch, or doing something for Pnut.exe to work. Currently, it does not find the board.

What do I need to do here? Or, which doc do I need to re-read?

Either is fine :D

Edit: I need to power off, not just reset to load a new program via FastGUI.

BTW: The current HDMI code in the spiral thread works just fine on my Acer monitor I'm using right now. I've not gone through and tried variations on the driver Roger did just yet.

Comments

  • Hmm shouldn't need to do anything.

    Perhaps flexgui is keeping the terminal window open, thus preventing pnut finding the board? You can remove the -t -k options from the loader if need be (that doesn't keep the terminal window open)

  • evanhevanh Posts: 15,916
    edited 2019-11-24 00:01
    Pnut should just work too. Chip recently updated it to v33L, I've forgotten why. Lots of goodies at https://propeller.parallax.com/

  • TubularTubular Posts: 4,702
    edited 2019-11-24 00:32
    33L supports com ports up to com99. Previous versions supported com1 through com9

    What OS are you using @potatohead
  • potatoheadpotatohead Posts: 10,261
    edited 2019-11-24 00:36
    Right now, I'm on Win10. I've got some fairly old P2 code and assets on a Win XP laptop I need to copy over. May move over to Linux here soon. I've got an old ThinkPad free, and it's running MINT.

    But, for right now, quickie P2 time, it's a P50s on Win10, recent build.


    It's gotta be the terminal window. I should have known. That worked.

    What I was doing was running a program, then looking at various documents, browser windows. That terminal window was near or at bottom of the window manager stack!

    DOH!

    Thanks!

    Pnut, FastGUI working.

    evanh: Got the goodies, and the current v33L file. It's been a lot of fun today. Honestly, that Goertzel demo makes me want to fabricate a slightly bigger one. Too fun.

    Haven't gotten to the scope mode code yet.

    Next task is to get an SD card formatted and loaded with Peter's latest.
  • Seems like you're really having some proper fun now @potatohead ! Glad you are now taking the risk with connecting up the HDMI. IMO it's probably a fairly lowish risk to be able to damage the monitor giving the P2 output voltages involved, more of an issue if there is something bad in your monitor, that could fry directly attached P2 pins. Again hopefully rather unlikely to happen.
  • potatoheadpotatohead Posts: 10,261
    edited 2019-11-24 01:48
    It is a lower risk, but the screen was not mine. Had a chat and nobody was going to cry if it lost it's HDMI in. So... GAME ON!

    But you know how it is. Gotta be respectful with other people's gear. I ditched the switcher idea too. That's in use, and I just don't want to futz with it right now.

    I really like how lean a basic HDMI bitmap can be. Awesome sauce. I'll bet nobody else out there has that. And if one of us gets crazy, driving a handful of displays is a no brainer, save for how to manage RAM. Just like P1, this is going to be a lot of fun.

    I gotta run your driver. Haven't yet, but I'm all setup. Old school mono NTSC Amber, PVM, VGA, HDMI. I don't happen to have a component display handy. Thought the PVM had one, but it's RGB. We need to test that anyway.

    Also, who knew? That PVM has TTL, old school. CGA basically.

    I spent a likely embarrassing amount of time playing with the Goertzel today. Too damn much fun.



  • Lol CGA. Yeah I think even the 4 bit TTL will be quite doable at some point. I want to revisit it someday. Tubular mentioned he has some old school monitors I could test it on. Plus LCD CLK/DE output signals should ideally be part of too it with 24 bit RGB parallel output pins, or maybe an 8/16 pin subset for simple LUT mode translation in pin constrained setups.

    Glad you have a good working environment now. If/when you do get the chance to play with NTSC and my driver, take a look at its output to see if there is something obvious I should be doing to improve the CVBS signal with better XFRQ/CFRQ/timing values or more/different xzero's etc. I didn't dig into CVBS in detail as I wanted to get this beta out for people to use and provide feedback of any problems etc. We can add a CVBS/S-video fix to the driver if required - I have 5 spare longs now. Plus I still need to add in the correct PAL colour flip values as well.
  • I plan to. My fear is we are seeing a phase error between the color system and the pixel clock due to the frequencies not being even multiples.

    I am not sure we have a way to reset the color like we do the streamer with an xzero.

    One answer might be to move the shimmer. If the color burst was done with pixels, that and the pixel clock would be stable. Colored pixels may still present artifacts, but monochrome things would not. Text, etc may be improved.

    Or, we use some common frequency.

    Or? Who knows. Some exploration is needed.

    Re TTL: One thing I have wondered for a while is just how fast their response is. With high speed pixel clocks, "sub pixels" could potentially mix to get more color. Save that for a fun day.

  • roglohrogloh Posts: 5,787
    edited 2019-11-24 05:20
    I am not sure we have a way to reset the color like we do the streamer with an xzero.
    Yeah, I was sort of wondering if the SETCFRQ might possibly do a phase reset when issued perhaps?
  • roglohrogloh Posts: 5,787
    edited 2019-11-24 05:28
    The thing is that in Chip's demos, which use 400 pixels per line, the colour looks really solid. My version with 858 pixels per line doesn't. I'm not sure if that is somehow related or I'm just doing something wrong in the NTSC sync timing somewhere that is throwing it off, or have bad XFRQ/CFRQ values, quantization errors etc.
  • rogloh wrote: »
    The thing is that in Chip's demos, which use 400 pixels per line, the colour looks really solid. My version with 858 pixels per line doesn't.

    You're talking about ntsc, right? I wouldn't expect even 320 pixels wide to look good. Is there a reason for such a high expectation I'm missing? :smile:
  • potatoheadpotatohead Posts: 10,261
    edited 2019-11-24 06:17
    Well, that number is the total pixels per line. Active will be lower. Probably 640. It is the active area that counts.

    In general, non phase change color has a baseline of 160 pixels in the active area. Those will match up with the period of the color burst.

    That can be pushed to 256 with only minor league artifacts. The better the signal, the more it can be pushed to a point.

    Phase change, alternating on frames is good for 320 pixels, and can be pushed to 480, again with minor league artifacts. This does depend on the quality of the comb filter. It also depends on signal quality. On P2 we have better DACS.

    Above that can work, but color management should be done. Think TV weather maps. Those are often done higher, but color changes are rarely at that resolution. Luma changes tend to work.

    Current P2 drivers are doing phase change, alternating frame.

    One exception to all this is text. The greater the color contrast, the fewer characters are resolvable. Better signals is a super big deal here.

    On P1, I tried phase change, but static. No alternating frames. This works very well for text. It is a hack, just like fixed phase, 240P was on home computers back in the day.

    On very good displays, like a PVM, the circuits are really good. A 640x480 display tends to be fairly good. I have one. It is crazy good. Consumer sets had much coarser CRT shadow masks and generally less performant circuits. I will have to get a photo or two.

    Pro grade video is generally 720x480, full interlace, phase change, alternating phase color.

    An 858 scan line is not unreasonable. Btw, the HOT chip, with its PLLs would render that perfectly. A good display would show 80 column color text in most all but the most aggressive color combinations. And it did that on composite on better consumer grade sets.

    This one can too, but we are going to need to get both the color frequency domain and base pixel clocks in consistent phase, IMHO.

Sign In or Register to comment.