WillFA11 - Lethal Weapon 3, round 4 Results of v1.28 and test build v1.29 ===================================================================== Thanks a lot for the clips, the gameplay video and the detailed notes. 1. WHAT YOUR v1.28 CLIPS SHOW The speed in attract mode is now exactly where we predicted it: Setting time per entry predicted ------- -------------- --------- B 2.23 s 2.22 s C 2.39 s 2.39 s D 0.87 s 0.85 s original 2.33 s (your original MPU clip) So the clock fix worked and setting A (2.30 s) matches your original MPU in attract mode. The gameplay video shows the problem you describe, and it is bigger than it looks at first: during normal play the DMD stays dark again and again for 8 to 16 seconds (the score display is missing, the animations still play). Twice the DMD shows its start-up screen "LETHAL WEAPON R2.05" - at 43 s and at 133 s. Both times this happens 8.5 s after the DMD went dark. That is a watchdog in the game software: when the DMD reports "not ready" for 8.5 s while the game has messages waiting, the game resets the DMD. The weird high scores (4,<02,500,000 and 4,:03,000,000) are the same in all your clips, so they are stored in the NVRAM and not a display problem. 2. WHY WE CHANGE THE APPROACH Until now we slowed the CPU down by a fixed amount. That can only match one situation (the attract mode we measured). In a real game the interrupt routine has more work to do, and the rest of the game then gets much less time than on the original, so the game runs slower than the original during play. v1.29 does it properly: every single CPU instruction now takes exactly as many clock cycles as on the original 6808 CPU. There is no speed setting any more - it should simply behave like the original in every situation. v1.29 also fixes an error in the sound output that could send a wrong sound command now and then (a candidate for the "wieeuw"), and the sound strobe line no longer sits in the "active" state while WillFA11 boots. The DMD lines are kept quiet during boot now, which may also help the PIN2DMD. 3. YOUR SETTING D GAME Thanks for the quick check. It tells us a lot: - the "wieeuw" comes in setting D as well, so it is not caused by the speed setting - it fits the sound output error fixed in v1.29 - the DMD also goes blank in setting D, only shorter (3-5 s instead of 8-16 s). The blank phase ends after a certain number of game loops, and in setting D the game loops run 2.5x faster. - no resets in setting D: the game only resets the DMD after 8.5 s of "not ready", and the blank phases there are shorter than that. So the blank DMD has its own cause, which we are now looking for. v1.29 will most likely not cure it completely, but should bring the timing back to the original. 4. TESTING v1.29 - file: WillFA11_v129.jic - option DIP 4: OFF (ON would switch the new timing off) - option DIP 5: ON (DMD link monitor on the WillFA11 digit, see 5.) - option DIP 6: OFF - the boot message shows version 1.29 a) High score list, exactly like before: power cycle, let it go to attract mode, film the DMD until the list has been shown at least once (#1 to #6). We expect about 2.33 s per entry. b) Play one or two games and film them like your gameplay video, with the DMD in the picture. - Does the DMD stay on during play (score display)? - Any DMD reset (start-up screen)? - Any "wieeuw", and when? - Does the game feel like your original MPU? c) NVRAM: the stored high scores are broken. Please reset the NVRAM once: option DIP 1 ON, power up, leave it in attract mode for about 30 seconds, power off, DIP 1 OFF again. (This clears all settings and audits of this game on WillFA11.) Do the weird digits come back later? d) PIN2DMD: does it show a picture now? e) Did you ever use other game numbers than Lethal Weapon 3 on this WillFA11 board? 5. THE DMD LINK MONITOR (DIP 5 ON) We decoded the complete data exchange between the CPU board and the DMD controller from both ROMs. If a single byte gets lost on its way to the DMD, the DMD waits for it and times out after 3.3 seconds - exactly the "blank for 3-5 seconds" you saw in setting D. A byte that arrives twice gives garbage on the display. What we do not know yet is WHERE the byte gets lost. v1.29 watches the connection and counts the problems. The result is shown on the single 7-segment digit on the WillFA11 board (the one that shows the boot phase at power-up). Once the game runs, it shows this sequence again and again: 2 s dark, then five numbers, 1 s each, with a short dark gap between: 1st number byte sent while the DMD was still busy (byte lost) 2nd number byte sent, but the DMD did not react 3rd number the DMD reacted without a byte being sent (extra byte) 4th number the DMD waited more than 1 s for the rest of a message 5th number DMD resets done by the game (1 after power-up can be normal) Each number counts up to 9 and stays there. They start at 0 at every power-up. Please: - after power-up and attract mode, note (or film) the five numbers - play a game; whenever the DMD goes blank, note the numbers again (or keep the digit in the picture of your gameplay video - even a second phone next to the digit is fine) - after the game, note the five numbers once more WHAT TO SEND BACK clip / check result ------------------------------ -------------------------------- v1.29 high score list clip name v1.29 gameplay clip name, DMD ok? reset? wieeuw? link monitor after attract _ _ _ _ _ link monitor after the game _ _ _ _ _ NVRAM reset done, digits back? .............. PIN2DMD .............. Thanks again - your clips made it possible to find all of this.