PAL 7800, bug with all multi screen lists

[+Andrew Davie]

Actually... the above is rubbish.  How about this...

 

timestart = now
while (now-timestart < MINTIME) {}
addr = ADDR_IN
                               

 

[+Andrew Davie]

Placing this here from the thread in 7800 touching on this subject...

batari wrote...

 

Quote

As for the issue, one difference with the PAL 7800 is how it detects a 2600 cart. It doesn't run the signature check but instead reads a range of addresses from the cart to check whether a 7800 cart or a 2600 cart is installed, and whether to run the built-in game Asteroids or not. It also runs through a few addresses that we typically use as hotspots. I am wondering if the cart is responding to these hotspots? If so then a possible fix is to ignore hotspots until we are sure the 2600 code is actually running.

I think we should look at this closely.

[+Al_Nafuur]

 

26 minutes ago, Andrew Davie said:

Placing this here from the thread in 7800 touching on this subject...

batari wrote...

 

I think we should look at this closely.

The UnoCart and the PlusCart ignore these initial 7800 reads.
Additionally the PlusCart menu has much less hotspots than the UnoCart menu..

[+Al_Nafuur]

4 hours ago, Andrew Davie said:

Actually... the above is rubbish.  How about this...

 

timestart = now
while (now-timestart < MINTIME) {}
addr = ADDR_IN
                               

 

I just tried the "more robust" address test (that is used in several bankswitching of the UnoCart and PlusCart) for the firmware ROM too. But then the PlusCart doesn't boot anymore (tested with PAL 2600Jr)

?

 

[juansolo]

On 10/5/2020 at 5:11 PM, marauder666 said:

I've had a chance to test with a NTSC 7800 I'm modding for someone. and it works as expected.

Seems just a PAL issue, also tested with a friends PAL 7800 with the same problem.

I am said friend

 

Though I'm now on an older firmware than Cleggy we still are both experiencing the same issues on our PAL 7800s.

[+Andrew Davie]

Just now, juansolo said:

I am said friend

 

Though I'm now on an older firmware than Cleggy we still are both experiencing the same issues on our PAL 7800s.

The issue is not forgotten.

Apparently the PAL and NTSC '7800 machines start up differently.

I am trying to find a '7800 so that I can do some testing/fixes.

 

[juansolo]

Nice one. Well you have another willing PAL guineapig should you require one.

[+Al_Nafuur]

I just bought this one:

https://www.ebay.de/itm/Atari-7800-Konsole-Spielekonsole-PAL-ohne-Zubehor/174329415163

 

But it is missing the powersupply ?

 

I'll look at my electronic scrap box or maybe I'll also buy this:

https://www.ebay.de/itm/Atari-7800-Netzteil-Adapter/254347183672

 

 

 

[marauder666]

Easy enough to wire up a 2.1mm DC socket onto the board in the power input area, and use a sensible 9V PSU with barrel jack.

[+Andrew Davie]

21 minutes ago, marauder666 said:

Easy enough to wire up a 2.1mm DC socket onto the board in the power input area, and use a sensible 9V PSU with barrel jack.

It is, but as we saw in the other thread where somebody did this on my behalf, it's also easy enough to blow up your 7800.

Me, I'd buy that converter linked in the post above - it looks neat, and isn't too expensive.

 

[marauder666]

I'm not sure how he got it wrong way round, unless the PSU was centre negative.

[+Andrew Davie]

1 minute ago, marauder666 said:

I'm not sure how he got it wrong way round, unless the PSU was centre negative.

I believe it was!

[+Al_Nafuur]

@Andrew Davie

I was able to track done the issue to the point where the VCS writes to the PlusCart command hotspot:

https://gitlab.com/firmaplus/atari-2600-pluscart/-/blob/master/source/STM32firmware/PlusCart/Src/cartridge_firmware.c#L342

 

here the wrong value is read. It should be 32 ($20) for page up, but it is 47 ($2F) or in my case 37 ($25). Maybe it is the hotspot address ($1FE6 ) or maybe it is because the write is done from code in the VCS RAM?

I was able to setup a workaround in main.c main function:

  while (1){
    int ret = emulate_firmware_cartridge();
    enum e_status_message menu_status = none, main_status = none;
    if (ret == CART_CMD_ROOT_DIR){
	  system_secondary_init();

	  d->type = Root_Menu;
      d->filesize = 0;

   	  input_field[0] = 0;
      curPath[0] = 0;
      menu_status = buildMenuFromPath( d );
    } else if((ret & CART_CMD_PAGE_DOWN)== CART_CMD_PAGE_DOWN ) {
    	act_page--;
    } else if((ret & CART_CMD_PAGE_UP) == CART_CMD_PAGE_UP ) {
    	act_page++;
    } else {
  .
  .
  .
  

But what scares me a bit is, that the code for reading the VCS write was the same than in every bankswitching with writes (e.g. SC RAM). So they may fail too and the workaround only works in this special situation.

We better find out why the wrong value is transmitted..

 

 

Edited October 20, 2020 by Al_Nafuur

[+Andrew Davie]

39 minutes ago, Al_Nafuur said:

@Andrew Davie

I was able to track done the issue to the point where the VCS writes to the PlusCart command hotspot:

 

The DEV version has some changes to the timing of emulation address changing.

Did you try release or dev in your testing?  If not DEV, I would like to know if it behaves the same.

 

[+Al_Nafuur]

2 minutes ago, Andrew Davie said:

The DEV version has some changes to the timing of emulation address changing.

Did you try release or dev in your testing?  If not DEV, I would like to know if it behaves the same.

 

tried Dev and Release, same issue

Edited October 20, 2020 by Al_Nafuur

[+Al_Nafuur]

Does the NTSC 7800 also has a inbuild Asteroids ROM?

can someone check whats at address $1FE6 in the inbuild PAL Asteroids ROM, or where I can find the binary?

[marauder666]

On 10/20/2020 at 12:59 PM, Al_Nafuur said:

Does the NTSC 7800 also has a inbuild Asteroids ROM?

can someone check whats at address $1FE6 in the inbuild PAL Asteroids ROM, or where I can find the binary?

NTSC 7800 does not have the Asteroids inbuilt as standard, but theres a later eprom, which does, but was never officialy released.  

 

The NTSC 7800 I video modded a few weeks ago, owned by mimo, had this eprom, and it worked with the pluscart

[marauder666]

Pluscart works on the PAL 7800 I've been sent to video mod, it's got an EPROM and not a mask ROM.  Later OS revision maybe, its a 1993 machine.

Need to also check some of Juansolo's homebrew 2600 carts which aren't detected on both mine and his PAL 7800s.

[marauder666]

Same result as my 7800 with the 2600 homebrews.

Wish I had the ability to dump this bios eprom.

[marauder666]

Menus now working on both 7800s with latest firmware.

Now I can select other titles, i’ve discovered some other issues, posted these in another thread.

 

 

[+Andrew Davie]

1 minute ago, marauder666 said:

Menus now working on both 7800s with latest firmware.

Now I can select other titles, i’ve discovered some other issues, posted these in another thread.

 

 

 

I believe you have a version which is a workaround for the issue.

Since that release, the *actual* bug has been discovered and fixed.  That will be in the next release, hopefully soon.

The issue was a "lda VALUE" instead of "lda #VALUE" in the menu/page switching code.

 

 

 

[marauder666]

I do have the workaround version.

 

Great news on finding the actual bug,  any reason it worked on ntsc, and the eprom pal i tried it on?  Or just a fluke?

 

 

[+Andrew Davie]

Just now, marauder666 said:

I do have the workaround version.

 

Great news on finding the actual bug,  any reason it worked on ntsc, and the eprom pal i tried it on?  Or just a fluke?

 

 

IMHO a complete fluke that it ever worked on anything ever. (!)

 

[+Al_Nafuur]

1 hour ago, Andrew Davie said:

IMHO a complete fluke that it ever worked on anything ever. (!)

 

"lda VALUE" instead of "lda #VALUE" is indeed a bug. But the nasty thing about that bug is, that it mostly works as expect (for zeropage write addresses).

 

The value the 6507 reads on the data bus is the fading shadow of the zeropage address that has been set by the "Cart" because of the 6507 last address bus request.

 

What happen on the data bus in that situation is totally unreliable, and may also depend on the used cartridge, hardware or peripherals:

 

CPU Cycle	1	2	3
ROM data	LDA	$10	-
Data Bus	A5	10
Data Bus set by	Cart	Cart	(no one)
Address Bus (6507 PC)	$1800	$1801	$10

 

@Mr SQL has made a test ROM that steps through all zeropage values. As far as I remember it might be used for paddle detection.

Edited November 1, 2020 by Al_Nafuur

[Mr SQL]

5 hours ago, Al_Nafuur said:

@Mr SQL has made a test ROM that steps through all zeropage values. As far as I remember it might be used for paddle detection.

 

I think it would work -there is one low zeropage value that changes immediately when the paddle is plugged in and the value would stick on the PlusCart, paddle cap stayed charged I think. I haven't checked if this is resolved in the latest build for it.  

 

EDIT: Meant to say the value would stick on the UnoCart.

 

I noticed I also saw this same issue Andrew observed on the other thread with the SuperCharger (bidirectional loading is actually working as long as the button press is light, it seems to be only the button latching and one other issue I can't figure out where KC proto should just load KC II after all lives are exhausted but it does not).

 

I think the latching is definitely an issue too though:

 

On 10/10/2020 at 10:59 AM, Andrew Davie said:

"12.2 Latched Input Ports (INPT4, INPT5) These two ports have latches that are both enabled by writing a "1" or disabled by writing a "0" to D6 of VBLANK. When disabled the microprocessor reads the logic level of the port directly. When enabled, the latch is set for logic one and remains that way until its' port goes LO. When the port goes LO the latch goes LO and remains that way regardless of what the port does. The trigger buttons of the joystick controllers connect to these ports."

 

 

Edited November 1, 2020 by Mr SQL
typo