User talk:eighty5cacao/misc/Unverified dot clock rates

From Pin Eight
Jump to: navigation, search

11.289 guess[edit]

Is the "11.289" in Exidy a 44.1k*256 crystal commonly used with CD audio in the same way that the 6.144 is used with DAT? And is "4.914" in Mr. Do's Castle really a 4.9152 MHz (128*38400 baud) crystal like that used in GSM? --Tepples (talk) 21:52, 13 February 2014 (UTC)

Respectively:
  1. Nothing I've seen contradicts this, but all xtal.h says is that the same crystal is used somewhere in SNK's 6502-based games like Vanguard (that driver makes no mention of raw video timings though).
  2. No, XTAL_9_828MHz means exactly what it says according to xtal.h. --Eighty5cacao (talk) 23:49, 13 February 2014 (UTC)

System M vs. System MK[edit]

Aren't the Y Unit and T Unit using a vertical scan frequency substantially slower than the 60 Hz of system M? --Tepples (talk) 03:34, 27 August 2014 (UTC)

Yeah, sorry I forgot. I knew I had looked at those drivers many times in the past but not added them here for some good reason. IIRC (I haven't run the numbers), the refresh rate is somewhere around 53–54 Hz (not the roughly-half-NTSC that some MAME versions/websites[?] claim it as, though the software may only be able to draw on every other frame). That would be enough to cause issues with most superguns though. --Eighty5cacao (talk) 06:30, 27 August 2014 (UTC)
Raiden II has a field rate slower than NTSC but faster than PAL. So does a notoriously violent arcade fighting game with a character named Raiden. Coincidence? I hope there isn't an arcade game based on characters in Metal Gear Solid not named Snake. --Tepples (talk) 04:23, 23 November 2014 (UTC)
As I understand it: I still need to verify the guesses I brain-dumped above. In general, (pixel clock)/(HTOTAL*VTOTAL) should equal the refresh rate, unless there's something I'm missing. I assume you posted that mainly for the humor value rather than to request any other action from me. Yes, the Mortal Kombat series was my original reason for reorganizing the page. --Eighty5cacao (talk) 06:47, 23 November 2014 (UTC)
But back on the topic of Midway T-Unit: The correct refresh rate based on the MCFG_SCREEN_RAW_PARAMS macro call is 8000000/(506*289) = 54.71 Hz — not quite as far from NTSC as I originally thought. I'm not sure whether any of the games reconfigure the TMS34010 registers at runtime though. --Eighty5cacao (talk) 06:52, 23 November 2014 (UTC)

Rainbow Islands[edit]

Regarding the guess of 340.5 px/line in Rainbow Islands: "not to mention that the pixel clock of 5.34 MHz and corresponding hblank time are implausibly low." For a 256px wide display, it's not that much lower than 5.37 MHz of TMS9918 and its descendants. Most TMS9918 descendants have a 342 pixel line; NES and Super NES have 341. The 340.5-pixel line is also plausible if the pixel clock is generated by dividing a master clock. On the NTSC Super NES, two lines in every other vblank do have 340.5 dots, as they're each missing two master clocks. (This produces less of a bend artifact near the top than the NES PPU, which deletes the first dot of every other field.) If MAME's lack of fractional lengths is the problem, then perhaps treating it as 512px with every pixel sent twice might be doable. Super NES is again precedent, as each pixel is divided into two half-pixels, with an optional "pseudo-hires" mode showing each pixel's unblended subscreen color in the right half of the previous pixel. How does MAME handle Nintendo Super System and Super NES? --Tepples (talk) 23:52, 20 December 2015 (UTC)

Apparently you failed to understand that the clean aperture is 320px wide; that's what makes the first guess implausible. The screenshots on TCRF are sufficient to demonstrate this; perhaps you were thinking of the NES or FM Towns ports or the title screen of the Mega Drive port. But to answer your question about SNES, there are some explicit * 2 factors in snes.cpp that result in always rendering at 512px.
My example invocations of the MCFG_SCREEN_RAW_PARAMS macro did have 320 in them, but perhaps I need to explain what the parameters mean. The parameters are defined (or at least conventionally labeled) as (PIXEL_CLOCK, HTOTAL, HBEND, HBSTART, VTOTAL, VBEND, VBSTART) where the full frame is HTOTAL by VTOTAL and the clean aperture is (HBSTART-HBEND) by (VBSTART-VBEND). (The names denote end of vertical blank, etc., and they have nothing to do with the word "bend.") I'll eventually copy this explanation onto the content page.
But in light of SNK (ED: and/or Alpha Denshi) keeping the same pixel clock from their previous hardware when designing Neo Geo, and NMK possibly doing the same for their 256→384px transition, there may not be any real problem here other than the lack of verification of the crystal source for the Rainbow Islands pixel clock. --Eighty5cacao (talk) 01:32, 21 December 2015 (UTC) to 02:47, 21 December 2015 (UTC) (+ 18:39, 13 December 2016 (UTC))
As for the issues with the visible area that I've been mentioning: TCRF appears to require registration to edit any namespace, including talk namespaces. --Eighty5cacao (talk) 19:32, 2 March 2016 (UTC)