CPS2 - Important information to anyone using the hardware.

Razoola

Divine Hand of the UniBIOS,
Staff member
20 Year Member
Joined
Nov 12, 2002
Posts
4,662
The problem
A CPS2 system which is not supplied enough voltage can work but at a slightly reduced speed. This can affect suicide free boards and also boards that have darksofts multikit attached. It also affects normal non suicide CPS2 games. Results so far indicate the multikits are prone to this situation, this may even out however as more tests are received.

Symptoms
The result of under power is games not playing as designed on a fully operating CPS2 system. The symptoms will be different from game to game and some games may show no symptoms at all. QSound problems can also happen (pops and crackles) due to potential fluctuating timings.

How to check your CPS2 setup
I have updated the free suicide tester tool to check for this situation for anyone who wants or feels the need to test their CPS2 suicide free or multicart system. The expected result is always a number or either 0x1EAE 0x0x1EAF. Numbers below this show the CPS2 system is not running as it should for the frame tested. Instructions are in the readme on how to use.

http://www.www.cps2shock.com/download/suicidetesterv2.zip

http://www.www.cps2shock.com/download/suicidetesterv2multikit.zip

The Fix
If your test results show you are suffering this problem your power supply may be on its way out or set incorrectly. Upping the voltage if available may fix the problem. Devices attached to the CPS-2 'b' board may affect the voltage required, care should be taken regardless to unsure you do not damage your system when changing voltages.

More information is always better. if your able to do these tests, please post your results.

-------------------- original post below --------------------

Hi, I am trying to obtain some information on CPS2 hardware over all PCB revisions. If you own one and are able can you please run the test code and report back. I'm supplying it ready to be programmed onto a 03 EPROM for use on a phoenix board or as a file that can be injected into the SD card of a SSF2 set on the multi kit.

http://www.www.cps2shock.com/download/testcps2eprom.zip

http://www.www.cps2shock.com/download/testcps2multi.zip

To test simply power on the unit and you should see a number (may be hard to read depending on the GFX held on the system). Make a note of the first number. Pressing button one will make the number will update. Over the period of a minute keep pressing button 1 and take a note of the range the numbers fall into (max and min number). Finally also report the CPS2 'b' board revision tested.

So I need...

The first number displayed, The number hi/low range over a minute and the 'b' board revision number.

Thanks in advance, Raz
 
Last edited:

mainman

CPS2 Person.,
20 Year Member
Joined
Mar 26, 2001
Posts
3,716
Hi, I am trying to obtain some information on CPS2 hardware over all PCB revisions. If you own one and are able can you please run the test code and report back. I'm supplying it ready to be programmed onto a 03 EPROM for use on a phoenix board or as a file that can be injected into the SD card of a SSF2 set on the multi kit.

http://www.universebios.com/testcps2eprom.zip

http://www.universebios.com/testcps2multi.zip

To test simply power on the unit and you should see a number (may be hard to read depending on the GFX held on the system). Make a note of the first number. Pressing button one will make the number will update. Over the period of a minute keep pressing button 1 and take a note of the range the numbers fall into (max and min number). Finally also report the CPS2 'b' board revision tested.

So I need...

The first number displayed, The number hi/low range over a minute and the 'b' board revision number.

Thanks in advance, Raz

You wanted the software tested on multiple revision cps 2 boards. I could have done that but you didn't specify that what you wanted initially.
 

Razoola

Divine Hand of the UniBIOS,
Staff member
20 Year Member
Joined
Nov 12, 2002
Posts
4,662
Yes, I did not think I would need that info initially but now it looks like I will because there is a variance I want to try and track down.
 

kahel

Armored Scrum Object
Joined
Apr 4, 2014
Posts
251
Just Ran the test.

Revision 6 board (93646B-6)

Start: 00001E9A
Low (not including start): 00001EAB
High: 00001EAE
 

Razoola

Divine Hand of the UniBIOS,
Staff member
20 Year Member
Joined
Nov 12, 2002
Posts
4,662
Just Ran the test.

Revision 6 board (93646B-6)

Start: 00001E9A
Low (not including start): 00001EAB
High: 00001EAE

Many thanks, can I have a little more information from you.

Did the high and low numbers give random results in that range during the minute of did they slowly rise from 00001EAB to 00001EAE to fall from 00001EAE to 00001EAB? Also as the minute test went on did the numbers become stable?

Were the numbers easy to read, Are you 100% sure the number was 00001EAB and not 00001EAF?

Finally, If you were using a EPROM what speed was it rated at? Or were you using a multikit?
 

kahel

Armored Scrum Object
Joined
Apr 4, 2014
Posts
251
It went like this:

1E9A 1EAC 1EAB 1EAC 1EAD 1EAC 1EAD 1EAE 1EAD 1EAE and then it alternated only between AD and AE for the rest of the test.

The number were clear and i am pretty sure it was AB but i only saw it once at the start.

This was done with the multi-kit.
 

Razoola

Divine Hand of the UniBIOS,
Staff member
20 Year Member
Joined
Nov 12, 2002
Posts
4,662
Thanks, something is strange with your results, while you were testing was music playing? Also, if you power off and on after the test has run for a minute, is the first number 1E9A again or does it start with a higher number close to 0x1EAE?

Thanks for the time to do the tests btw.

Raz
 

kahel

Armored Scrum Object
Joined
Apr 4, 2014
Posts
251
Yep , a nice SF2 tune was playing .

I'll re-run the test tonight to see if i get the same result, make sure it was AB and then cycle power and redo it right after.

Glad to help.
 

kahel

Armored Scrum Object
Joined
Apr 4, 2014
Posts
251
I ran 5 more tests. The number were very clear so i am sure about all of them.

The first one gave really weird result.
It went like this: 98 A5 A6 A7 AB AC (then it varied between AB and AC)

Second test started at AB and varied between AB and AF

Third test started at AE and only varied between AE and AF.

Fourth started at AD and varied between AB and AE.

Last test also started at AD and varied between AB and AF.

Let me know if you want to do more test or do them differently.
 

Razoola

Divine Hand of the UniBIOS,
Staff member
20 Year Member
Joined
Nov 12, 2002
Posts
4,662
Thanks this is perfect information and it has left me in a bit of a situation so I better explain what I am testing as it may end up being in the interest of those people using multicarts.

Way back when the phoenix edition fixes were created there were a few people that claimed the phoenix edition fixes ran at a different speed than original encrypted games. I checked this at the time and confirmed that they did all work at 100% identical speed. I did this using the test program you are using compared against the same test program what was encrypted to run on a non suicide PCB. That was the end of the story.

Just over a month ago I got an email again claiming that non encrypted games were not running at the same speed as the originals.... Specifically SSF2T series. As I have free time I decided to look at the speed again, Did the tests again on the 2 'b' boards I have here and asked a couple of people to do the same test privately. One was on EPROM the other like yourself a multicart.

All the results on EPROM give the same values... Power on value of either 1EAE or 1EAF and all other values in the next minute the same 2 values. This matches the results I got over multiple 'b' boards way back in the mid 2000's. The multicart this was tested on was different.... That gave a power on value of 1E9F and then the values over a minute went from 1E9D to 1E98. Very strange as these values should always only show a variance of 1 maximum. What is more worring however is the test on this multikit was reporting that it was loosing about 35000 cpu cycles a second... That is enough to affect gameplay given the way the SSF2 series is programmed (SSF2, SSF2T and HSF2).

More testing was needed hence this thread and then your result using a multikit. Like the other multikit test yours shows an issue, unlike the other test however yours is not loosing cycles at all most of the time, but it can sometimes, nothing like the amount of the other test. I don't know what is happening and more testing is needed to say the cause (is it the 'b' board, the hardware on it or the power supply because of the extra power draw). I think more test results are needed but clearly there are CPS2 games out there not playing at the correct speed.
 
Last edited:

mainman

CPS2 Person.,
20 Year Member
Joined
Mar 26, 2001
Posts
3,716
I take it you no longer need me to run the program across different B boards.

So essentially only the multiboot is affected by this issue.
 
Last edited:

Razoola

Divine Hand of the UniBIOS,
Staff member
20 Year Member
Joined
Nov 12, 2002
Posts
4,662
I think it would still be good to test other 'b' boards with EPROMs just to be sure but at the moment it looks like the issue is with the multis only, but some worse than others. What would really nail it down though is if someone with the issue on a multi then takes the multi kit off and trys with a EPROM on the 'b' board to see if they get the same reading as the multi or the expected reading of always 1EAE or 1EAF.

It may be for example that issue only affects the multis with a set 'b' board revision or it may not be down to the multi directly but rather the fact that power supplies cannot power the multi and cps2 together... The fix being to supply power the multi pcb separately.

This is something really Mitsurugi-w and darksoft need to ask their users to check to nail down the cause and find a fix if required.
 

Lemony Vengeance

Mitt Romney's Hairdresser,
Joined
Jan 30, 2012
Posts
4,204
I think it would still be good to test other 'b' boards with EPROMs just to be sure but at the moment it looks like the issue is with the multis only, but some worse than others. What would really nail it down though is if someone with the issue on a multi then takes the multi kit off and trys with a EPROM on the 'b' board to see if they get the same reading as the multi or the expected reading of always 1EAE or 1EAF.

It may be for example that issue only affects the multis with a set 'b' board revision or it may not be down to the multi directly but rather the fact that power supplies cannot power the multi and cps2 together... The fix being to supply power the multi pcb separately.

This is something really Mitsurugi-w and darksoft need to ask their users to check to nail down the cause and find a fix if required.

I've spent considerable time with my multiboard and I haven't experienced any slowdown. Besides, these things shouldn't be running the PHX roms anyway. Have you tested lag on the multis with the avalaunch roms?
 

Pasky

Fug:DDDDD,
Joined
May 5, 2013
Posts
2,844
I think you're missing the point that Razoola is concerned with and that is the accuracy of the games being played back on the multi cart. It may not be noticeable to a casual player but perhaps to a high level player that can execute combos that require 1-frame links and can do them consistently. As an example, you could be losing 1 frame per 1000 in terms of speed (just an example) and you most likely wouldn't notice. 35,000 cycles being lost is not a minuscule amount, a single instruction on a 68000 could take 4-158 (MULU/S and DIVU/S reaching the higher cycles needed), that's almost 220 ASM instructions lost assuming the impossible situation that they are all DIVU's. I'd agree that this does need further testing to be sure.

For example, if you ran a legitimate SSF2T board side by side next to one on the Multi cart with the exact same romset and let them play for 24 hours, the multi cart board would slowly drift out of sync dude to the lost cycles.

I don't think he is saying it's unplayable but rather wants to ensure the integrity of the games are intact as I doubt the casual play would mind.
 
Last edited:

kahel

Armored Scrum Object
Joined
Apr 4, 2014
Posts
251
It voltage... It explains my first weirder results because my supergun was lower when i started it.

I set the power to 5.08V and ran 3 tests, now i always get between AE and AF.

Seem i need to run my cps2 board at 5.08 with the Multi cart.
 

Tyranix95

Chang's Grocer
Joined
Jun 30, 2010
Posts
4,593
... I did this using the test program you are using compared against the same test program what was encrypted to run on a non suicide PCB. That was the end of the story. ....

Post the link for the encrypted test program rom.
 

ForeverSublime

6400|!!|Kyo Clone
20 Year Member
Joined
Oct 23, 2001
Posts
6,416
Thanks for working on this fix, Razoola. I don't own one of these boards, but I love reading about this type of problem solving.
 

Lemony Vengeance

Mitt Romney's Hairdresser,
Joined
Jan 30, 2012
Posts
4,204
I think you're missing the point that Razoola is concerned with and that is the accuracy of the games being played back on the multi cart.

It's not his creation.
He objected to the compatibility of his work being played on it due to his dissatisfaction of it even existing.
Now he's saying that the PHX roms (which should be incompatible with this device) run slowly due to a defect on the device (a device that he himself doesn't like).

Sounds legit to me.

Would be nice to have the creator of the multiboard provide his own findings on it.... fat chance of him coming back here though.
 

Razoola

Divine Hand of the UniBIOS,
Staff member
20 Year Member
Joined
Nov 12, 2002
Posts
4,662
The code I created myself is testing the cycle speed, its neither a phoenix ROM or an avalaunch set, its a custom piece of code totally (counter calculation part below). Its only 0x460 bytes in size which is simply pasted over one of the files darksoft provided to be uploaded to the multi. This test intended to again check the speed of the phoenix sets (non encrypted versions of games) running on real hardware over a larger base so I included a multi version of the test as it is a way for people to test it who don't have EPROM programmers. Unfortunately it has highlighted a potential issue that so far is only affecting multicart users. More tests are needed to lock down the cause but there is an issue somewhere

This is all about non encrypted games playing at identical speed to the original. It has highlighted an area where they do not.

Code:
vbl	addq	#1,d7			vblank vector comes here
	rte	


again	move.w	#$2700,sr		disable interrupts
	moveq	#0,d0
	move.w	#$fffe,d7
	move.w	#$2000,sr		enable interrupts
stloop	move.w	d7,d2
	bne.s	stloop

loop	addq.l	#1,d0			add to counter
	cmp.w	#$0001,d7			
	bne.s	loop

	move.w	#$2700,sr		disable interrupts
	lea	$0092CCD8,A1
	bsr	pval			Print counter (D0)
	bsr	wbutton			Wait button press
	bsr	clear			Clear entire screen
	moveq	#127,D1	 
	bsr	wait			Small delay
	bra.s	again
 
Last edited:

Razoola

Divine Hand of the UniBIOS,
Staff member
20 Year Member
Joined
Nov 12, 2002
Posts
4,662
@Lemony Vengeance

Your setup may be playing at the correct speed, but it also may not. There is no way for you to know unless your a very experienced player with some of the CPS2 titles or you do the test for yourself on your setup.
 

Razoola

Divine Hand of the UniBIOS,
Staff member
20 Year Member
Joined
Nov 12, 2002
Posts
4,662
It voltage... It explains my first weirder results because my supergun was lower when i started it.

I set the power to 5.08V and ran 3 tests, now i always get between AE and AF.

Seem i need to run my cps2 board at 5.08 with the Multi cart.

This is good information, Its a case of if you would have been required to make that adjustment if the multi was not present on the CPS2 'b' board (probably you would not have due to the lower power draw). Everyone with a multi is going to have to check if they want to be sure games are running at an accurate speed.
 

Razoola

Divine Hand of the UniBIOS,
Staff member
20 Year Member
Joined
Nov 12, 2002
Posts
4,662
Post the link for the encrypted test program rom.

Its not that simple as a different encrypted build would be needed depending on the non suicide 'b' board you wanted to test on, I never kept the encrypted test bin used because of this. I have spent the morning trying to find the tool created that would re encrypt code but I can't find it, its going to be on a old lost HDD somewhere I guess.

Talking about this now I remember the encrypted test program had the added watchdog kicks in place (those needed changing dependant on game also), these were intentionally not in the decrypted build. They did not steel cycles from the timing calculation loop though so the results equal. Back then I was testing a system with a working protection/watchdog against a system with default protection and no triggered watchdog (as happens when a phoenix game is running).
 
Last edited:

Razoola

Divine Hand of the UniBIOS,
Staff member
20 Year Member
Joined
Nov 12, 2002
Posts
4,662
Original post has been updated to make this information easier to understand.
 

Dion

Known Scammer, NeoGeoFreak Co-Founder
Joined
Aug 24, 2000
Posts
2,070
Wow. You are really stretching with this one. Makes you look really bad actually. One test and its conclusive? Testing and adjusting voltage a bit is part of the hobby. If you think the multikit is drawing more power than a fully populated eprom daughterboard then you aren't as smart as you try to appear. This has been out for awhile and there are hundreds of these in the hands of gamers. Many of them "elite" fighting game players. NOT ONE PERSON has complained about any timing issues. So right now it just looks like another witch hunt by Razoola.

Grow up.
 

ForeverSublime

6400|!!|Kyo Clone
20 Year Member
Joined
Oct 23, 2001
Posts
6,416
Razoola, can you produce one person who has complained about a timing issue?
 
Top