05276: witch, pbchmp95: Graphics corruption

ID	Category [?]	Severity [?]	Reproducibility	Date Submitted	Last Update
05276	Graphics	Minor	Always	Sep 2, 2013, 14:25	Mar 1, 2014, 13:21

Tester	D Go Go Fan	View Status	Public	Platform	MAME (Official Binary)
Assigned To	Osso	Resolution	Fixed	OS	Windows Vista/7 (32-bit)
Status [?]	Resolved			Driver	excellent/witch.cpp
Version	0.148	Fixed in Version	0.153	Build	Normal
		Fixed in Git Commit		Github Pull Request #

Summary	05276: witch, pbchmp95: Graphics corruption
Description	Layers of graphics missing from right place and put in wrong place. Does not happen on Sega version, and am 99% sure did not happen on "Pinball Champ 95" when Sega version was parent. Think it could be a bad dump of graphics rom.
Steps To Reproduce
Additional Information
Github Commit

Flags
Regression Version	0.147u3
Affected Sets / Systems	witch, pbchmp95

Attached Files	0000.png (11,900 bytes) Sep 2, 2013, 14:25 Uploaded by D Go Go Fan
Attached Files	witch.png (28,577 bytes) Sep 5, 2013, 16:46 Uploaded by M.A.S.H.

No.09773 Haze Senior Tester Sep 5, 2013, 13:25 edited on: Sep 5, 2013, 13:25	yeah, I thought this when the new set was added, but I assume the fresh roms were dumped from the new board, and they matched the set with the corrupt graphics? or at least I hope whoever dumped it did actually bother to check that.
No.09774 M.A.S.H. Senior Tester Sep 5, 2013, 16:46	The gfx corruption is a Z80 CPU1 speed problem. In MAME 0.147u3 Brian Troha did a minor Witch update based on eBay auction picture. He changed the Z80 CPU1/2 clock speeds to 3MHz (see snapshot), and clone Pinball Champ '95 has now this gfx problem, but not the old parent version Witch (Sega License). If you changed the Z80 CPU1 back to 8MHz or to 4MHz the gfx was correct in all version (witch and clones witchs and pbchmp95).
No.09775 Mamesick Senior Tester Sep 5, 2013, 19:20	I would say XTAL_12MHz / 3 then, which sounds more reasonable.... though it should be verified........
No.10144 Osso Moderator Jan 6, 2014, 18:44	Guesswork for guesswork, I applied the change proposed by M.A.S.H. I'll contact f205v and see if he still has the PCB and can check.
No.10334 M.A.S.H. Senior Tester Feb 27, 2014, 11:22	Reopen the bug report, because Roberto Fresca reverted Osso's fix in r28056: Witch: Changed both CPU clocks to 3 MHz, Both YM2203 clocks to 3 MHz, and confirmed the ES8712 clock (coming from pin 14 of M5205) in 8 MHz. All measured and traced from the PCB. Implemented all clocks through #define. [Roberto Fresca, ANY]
No.10335 Osso Moderator Feb 27, 2014, 11:37	Now that the clock has been measured, it's clear that the guessed 4MHz was wrong so there must be another problem that the "overclock" masked.