1790 replies to this topic

[hayhu1m]

Hi guys i have just tested this with a couple of data east tables leathal weapon 3 and star wars and the dmd goes blank after a while can anyone else confirm reverting back to older vpm sorts it out

[Sixtoed]

Can confirm, blanks out in star wars after a while, also seems to reset the death star multiball as soon as the trench opens?

[Thalamus]

Only tried LW3 of the suggested and only one game. Didn't notice anything. Did you guys play many games before you saw issues ?

[Sixtoed]

I played one game of star wars (1.07) with the new beta and the normal score display vanished when I got tri-ball ready, and never came back. All the normal trigger sequences still work, but the fallback scoreboard didn't.

 

*edit* Just checked and lethal weapon 3 does the same

Edited by Sixtoed, 30 March 2020 - 09:27 PM.

[Thalamus]

I had no issues with the trigger. Actually played all the scenes and started over without issues. I'll try Star Wars soon.

[Thalamus]

I don't know. I might have a special setup compared to you. What I do know is that I need a firmware update to my pindmd3 to get the display to play at full speed - most noticeable on Fish Tales and Attach from Mars. I played both the mentioned tables with Freezy and colored dmd. I did not have any issues like you. I don't know if the video shooting sequences is perfectly timed or not since I don't have a real game to compare it to. But, it felt ok to me. Ref the firmware issue : https://github.com/f...mment-605646553

 

I tried SW without color, and it felt just the same to me. I'm of course running the Carny build in that I have color. A bit inclined to jump back to official in that I need to take away freezy and replace it with pindmd.dll.

[DJRobX]

There's definitely weird timing issues with Star Wars.  I had to add delays on triggers in the SW table to get the sequences to do what they're supposed to do.   Seems like the game logic is running too slowly (in the old "working" build).  It might be that the CPU needs to be sped up but maybe an IRQ interval slowed? 

Edited by DJRobX, 30 March 2020 - 10:34 PM.

[Sixtoed]

Yeah, there's definitely weird timing issues on Batman Forever too, it seems to spend a lot of time thinking about what it's going to do after a trigger is hit, the most obvious probably being the vertical "eject" wire run, and I'm pretty sure it's missing certain animations as the previous one is sometimes still playing when the new one is meant to be triggered. It's definitely not the same when directly compared to a real machine.

I've done a little digging today into the hardware, and the CPU boards are essentially identical, they do have different sound boards and different display boards though;

This is when things were changed in mainline MAME;

• 0.193: Replaced M6809 CPU2 with MC6809E (2MHz).
• 0.177: Started syncing with latest PINMAME. Will add new games directly in the wn [Osso].
• 0.150: Added video\decodmd3.c/h. Added Type 3B CPU board to machine\decopincpu.c (Data East Pinball CPU boards), fixed silly switch-related bug. Various DE_3B cleanups, hooked up CPU board device. Added DMD (Deco Type 3) to DE_3B. Corrected DECODMD3 lower dot intensities. Stop detest from bombing out with an invalid bank [Barry Rodewald]. Replaced M6800 CPU1 with M6808 (4MHz) CPU1. Added M68000 (12MHz) CPU4. Added DMD display (128x32 or 192x64, 60 Hz). Added Samples sound (bumper, chime1, chime2, chime3, coinin, hole and knocker.wav). Added Player 1 with 3x coin slots, tilt and 'Language' configuration and ramoption.
• 0.143u6: Added M6809 (1789790 Hz) CPU2, TMS32015 (24 MHz) sound, BSMT2000 (24 MHz) sound and 2x Speaker.
• 0.141u1: Added de_3b.c driver.

Mainline MAME runs the CPU's at the same speed for both the earlier games and the later games, so like you say, it's got to be something else, if you run stwr_106 in MAME it's 2 & 4MHz and it runs at exactly the same speed as pinmame at 1 & 2 MHz

 

Hope this digging helps...

Edited by Sixtoed, 30 March 2020 - 11:14 PM.

[toxie]

Great! VPM now on MacOS, too!?

macOS should've worked already, i think, this is specifically about iOS.

[toxie]

Hope this digging helps...

Indeed, always good to have multiple people looking at such difficult things with lots of HW variations involved.

 

The current theories are the following: The 6808 can be selected to run at 4 or 8 MHz clock rate, but this seems to be via a HW jumper, so unlikely to come into play. Also my tests on having it run at 8 MHz cuts off some animations in Baywatch for example. The other theory is that there are PALs that change the FIRQ rate dynamically, so that's why the measured rate is good for the standard case, but might be too slow for some scenarios like the video mode. Unfortunately these PALs are not emulated/read-out AFAIK.

[Sixtoed]

So at the moment in the VPinMAME33 beta I have from the 29th, the main control CPU M6808 is @ 1MHz and it needs bumping to 4MHz (not 8MHz), this is the case for all of the games on the hardware no matter what revision, everything else can stay the same (6809's @ 2MHz, 68K's @ 12, BMST's @ 24).

Like DJRobX says, Star Wars has problems in lagging switches, so there's probably still something wrong with the base hardware somewhere to do with this, but it's been fixed in normal MAME so there must be a solution, I can't read/understand the code though...

The easiest quick and dirty temporary hack would be to split the large DMD games out into a seperate 3b driver and have them running at the correct speed?

There must be a reason why MAME has them split, but I'm not a coder, just a researcher

Edited by Sixtoed, 31 March 2020 - 10:38 AM.

[toxie]

The 6808 is okay like is, as it divides the clock by 4, so the 1MHz is the same as MAMEs 4MHz (i was confused myself by this until yesterday ).

And how do you know that the switches in MAME actually work better? Is there some way to trigger/test them there during actual gameplay?

And the 192x64 already uses a separate driver (somehow), and bumping up the speed just leads to animations being cut off too early, so this is not enough.  :/

[Sixtoed]

OK, that's fair enough, I didn't know that about the clock divider thing.

 

I didn't mean the switches work better in mainline MAME, it was more an observation that you said Star Wars was running too fast with the higher clock speed and running through the settings and attract in MAME seemed to work at the "correct" speed with what I thought was a higher cpu clock speed (i.e. Star Wars didn't seem to be running too fast through attract, Batman Forever seemed to run at the right speed through attract), and then DJRobX said that the triggers were slow to respond in Star Wars and that's also the case in Batman Forever.

 

At this point I'm just guessing wildly, do the 6809's do the same with the clock speed divider? as both mame and pinmame have them as 2Mhz, which is odd if one's doing it but not any of the others? (or maybe not, again, not a coder!) Star Wars uses a 6809 for the display driver where Batman Forever uses a 68k, are these supposed to be clocked the same in MAME and PinMAME?

 

I'm just throwing out questions and ideas, if you want me to stop that's fine

[toxie]

There's definitely weird timing issues with Star Wars.  I had to add delays on triggers in the SW table to get the sequences to do what they're supposed to do.   Seems like the game logic is running too slowly (in the old "working" build).  It might be that the CPU needs to be sped up but maybe an IRQ interval slowed? 

Maybe you could quickly try the following patch in s11.c? (have to go back to work :/)

 

static WRITE_HANDLER(pia4b_w)
{
  if (locals.deGame) {
    int x;
    for (x=0; x<8; x++)
      if (data & (1u << x))
        break;
    locals.swCol = data & (1u << x);
  }
  else
    locals.swCol = data;
}

 

[toxie]

Here a quick test build:

 

Improve IRQ timings of Sys 11, DE/Sega/Sterns
Change switch writing of DE/Sega/Sterns

 

Please test again, especially all the previously mentioned machines (Baywatch, Batman Forever, Star Wars DE, Lethal Weapon 3), but please also some System 11 games (e.g. High Speed, Whirlwind, Black Knight 2000, Earthshaker, Diner, Radical!, .. ..).

 

And let me know if the switch changes did anything at all to improve the switch lag situation.

 

EDIT: removed file nowadays

Edited by toxie, 04 April 2020 - 08:27 AM.

[Sixtoed]

Batman Forever - Much better all round, seems to be pretty much spot on now, I ran it side by side with a few real machines on youtube and the animation speeds and everything else seem to match exactly now.

 

Jurassic Park - The T-Rex is totally broken, it's obvious on start-up, it should go down, left, right, bite a few times, down, left right, bite and start. It now goes down, then sometimes just starts, and if you get a ball in it is sometimes drops it, sometimes eats it then  immediately goes down again and bugs out.

*putting this here just in case - if table developers have put workarounds in place for switch / trigger lag then this could be an additional cause of this issue?*

 

Star Wars, Lethal Weapon 3: "Score" display goes blank, get the display board version appear randomly in game, but all "trigger" animations work perfectly, switches seem to work much better, death star door doesn't seem to close randomly anymore. It lost the ball totally once in star wars and I had to reset but then next boot up it was totally fine for 2 entire playthroughs.

 

High Speed, Black Knight 2000, F-14 - Seems to works fine, got to multiballs, Whirlwind is probably fine, I just suck horribly at it and can never get the ball past the spinning disc to the lock.

 

I'll add more as I test more and edit them in here, but I've got to get back to work!

Edited by Sixtoed, 01 April 2020 - 10:29 AM.

[hayhu1m]

Can confirm Batman Forever so much better.

 

 

Star wars, lethal weapon Score display still goes blank.

 

Just tested Jurassic Park T-rex throws a switch failure

Edited by hayhu1m, 01 April 2020 - 10:25 AM.

[hayhu1m]

Have tested all stern tables, williams tables, bally tables all seem fine looks like its just data east

 

Also Gottlieb/Premier tables seem ok too

Edited by hayhu1m, 01 April 2020 - 12:02 PM.

[Thalamus]

Can confirm issues with T-Rex on JP.

Not able on my desktop machine running Win7 64bit and native pinmame any of the issues reported for LW3 or SW.

 

LW3

https://www.dropbox....w3_208.avi?dl=0

 

SW

https://www.dropbox....wr_104.avi?dl=0

 

- Looking at this afterwards - at 2:20, is there a reset of the dmd of sorts ?

Edited by Thalamus, 01 April 2020 - 12:55 PM.

[hayhu1m]

https://youtu.be/bCC9Q8300go link to lethal weapon dmd