PlusCart(+) + SD

[JetSetIlly]

52 minutes ago, Thomas Jentzsch said:

@Kupakai That's very interesting. Can you please test the attached ROM too?

 

@Kupakai Can you also test Halo (if you can find the binary). Move the player around a bit and see what happens. Halo uses early HMOVE too but in a different way.

[Thomas Jentzsch]

2 hours ago, JetSetIlly said:

@Kupakai Can you also test Halo (if you can find the binary). Move the player around a bit and see what happens. Halo uses early HMOVE too but in a different way.

And Cosmic Ark would be nice too.

 

@JetSetIlly Is that broken with your modification too?

[JetSetIlly]

Just now, Thomas Jentzsch said:

And Cosmic Ark would be nice too.

 

@JetSetIlly Is that broken with your modification too?

Yes.

[RFV22]

9 hours ago, Kupakai said:

I don't think the development board you linked to is the right one. Do a search for "STM32F407VGT6 discovery". The correct one has double rows of pins on 3 sides. See the post at the top of this page for a picture of what it looks 

That's it, I got the link wrong, I'll agree there. 
https://a.aliexpress.com/_mNTk9YC

Thank you very much for letting me know. 

But are certain components

[DirtyHairy]

@JetSetIlly I **think** I have implemented the same effect in Stella by gobbling a hmove pulse (MOTCK) when it happens on the first cycle of a line. Without knowing your TIA implementation I suspect this is what moving MOTCLK between the phases does for you: the separation between MOTCLK and the last regular clock supplied to the sprite counter gets lost, combining them into a single clock which effectively gobbles the MOTCK.

 

Those are the results for the menu, @Thomas Jentzsch's test and the Cosmik Ark starfield:

 

Does this match what you are seeing? The starfield seems plausible to me: usually the sprites get 17 pixels of movement per line, leading to an irregular pattern, while gobbling the first clock reduces this to 16 pixels, which is one tenth of a line giving a regular pattern.

[JetSetIlly]

21 minutes ago, DirtyHairy said:

@JetSetIlly I **think** I have implemented the same effect in Stella by gobbling a hmove pulse (MOTCK) when it happens on the first cycle of a line. Without knowing your TIA implementation I suspect this is what moving MOTCLK between the phases does for you: the separation between MOTCLK and the last regular clock supplied to the sprite counter gets lost, combining them into a single clock which effectively gobbles the MOTCK.

My Cosmic Ark result is different. I think we need to see the results from the hardware. There are several explanations I think.

 

The thing is, I've never seen the Cosmic Ark pattern that you show but then again, I've never seen the Cosmic Ark pattern that I show. But I suppose this is because the Janice chip is so rare.

 

With my method as described above, the change of HMM0 value happens at the wrong time relative to the motion clock. Meaning the HMOVE ripple counter catches the wrong value and ends the HMOVE almost immediately. This causes a single column of missile pixels.

 

 

Normal result

 

 

Known TIA Variation. The Stella GUI says this is caused by an "inverted HMOVE clock phase". I think of it as a "lost motion clock".

 

[Kupakai]

19 hours ago, Thomas Jentzsch said:

@Kupakai That's very interesting. Can you please test the attached ROM too?

 

It should look like this, but I suppose it doesn't:

Note: This program tests early HMOVEs for all 5 objects (@74 at the top half and @73 at the bottom half).

HMTest.bin 2 kB · 5 downloads

I tried the test rom in the Atari and it looks like this:

 

I also tried the hmenu.bin and it looks like this:

 

When I tried launching Halo and Cosmic Ark, both briefly shows a screen and then crashed back to Pluscart menu. Halo did load the start up screen and played the music, but when I start the game it would crash.

[Kupakai]

I don't know how helpful it is but here are the videos of the games crashing:

Cosmic Ark

 

Halo

 

[JetSetIlly]

2 hours ago, Kupakai said:

I don't know how helpful it is but here are the videos of the games crashing:

Thanks for that. I think there's more going on with this machine than just a funny HMOVE.

 

Can you launch any games from the Harmony menu?

 

 

@DirtyHairy The right hand side of the HMTest screen is interesting

 

 

This is mine:

   

Not quite, but close. I think the cause of this is a delay but not directly caused by a phase skew. I'll play around with it some more.

 

 

Edited August 14 by JetSetIlly

[Thomas Jentzsch]

55 minutes ago, JetSetIlly said:

Thanks for that. I think there's more going on with this machine than just a funny HMOVE.

Agreed. But I guess the crash problems are not TIA related. 

 

The briefly shown Cosmic Ark starfield looks different than everything we have seen so far.

Edited August 14 by Thomas Jentzsch

[JetSetIlly]

14 minutes ago, Thomas Jentzsch said:

Agreed. But I guess the crash problems are not TIA related. 

 

The briefly shown Cosmic Ark starfield looks different than everything we have seen so far.

It does but I'm wondering if whatever is causing the crash is also causing these artefacts in the TIA.

[Kupakai]

I tried Cosmic Ark and Halo on another Janice Atari and they both crashed the same way. Just in case it’s an issue with my Pluscart, I’ll try to get a real cartridge of Cosmic Ark or try to make an eprom cart. 
I don’t have an Harmony cart. I was just trying the rom image you posted.

Edited August 14 by Kupakai
Additional info

[Thomas Jentzsch]

Usually these games should not crash, even if the TIA is not compatible. There is something else wrong here.

[+Al_Nafuur]

18 hours ago, Kupakai said:

I don't know how helpful it is but here are the videos of the games crashing:

Cosmic Ark

 

Halo

 

You have to disable the Exit Function in Setup before you start the ROM

[Kupakai]

7 hours ago, Al_Nafuur said:

You have to disable the Exit Function in Setup before you start the ROM

Thanks! Once I turned it off, the games do run. Both machines now run Halo fine. However, the star field in Cosmic Ark and the right edge of the HMTest looks different in the two machines. 

First machine:

 

Second machine:

[JetSetIlly]

2 hours ago, Kupakai said:

Thanks! Once I turned it off, the games do run. Both machines now run Halo fine. However, the star field in Cosmic Ark and the right edge of the HMTest looks different in the two machines.

That's very interesting. I'm not sure what to make of the fact that two different consoles of the same type are producing different results.

 

I'm a little reluctant to spend much time on this until we understand more about the Janice chip. The fact that we have two different results from two different machines makes me think there's something about these chips that we don't know about it. For instance, do we even know if the 6507 part of the chip performs the same as the MOS 6507?

 

To be clear, I don't think the CPU is causing what we're seeing on the screen in the above examples. But establishing that the 6507 is functioning the same as in other consoles would be a good first step. I don't have anything specific for 6507 in a ROM format though - only code that's intended to be run in a 6507 test harness. Does anyone have a good test binary focused on the CPU?

 

With regards to the TIA output I'm fairly confident that with a bit of effort, we can find something explaining each example output and then predict how other binaries might react on the hardware. What puzzles/worries me is that this is two example ROMs that perform differently on two different machines. How many variations of Janice are there?

[Thomas Jentzsch]

Yes, very interesting. I wonder if the video changes when the machine warms up. Wouldn't be the first time.

 

The first Cosmic Ark screenshot looks familiar.

 

@Kupakai How many of these consoles do you have?

 

BTW: In which way could the 6507 be different?

[Thomas Jentzsch]

13 hours ago, Al_Nafuur said:

You have to disable the Exit Function in Setup before you start the ROM

Do you have a list of games affected? If not, maybe we should create one.

Edited August 15 by Thomas Jentzsch

[+Al_Nafuur]

1 hour ago, Thomas Jentzsch said:

Do you have a list of games affected? If not, maybe we should create one.

I don't have a list, but a search for ".noexit" in the PlusStore repository finds only three ROMs:

• Bad Apple Demo
• Indy 500
• Crazy Ballon

I just retested "Cosmic Ark" (Public ROMs/Classic Roms/NTSC/BY ALPHABET/A-G/Cosmic Ark.bin) on my system and it is not affected by the unintended exit. So maybe @JetSetIlly is right that there are more differences with the Janice chip. @Kupakai can you retest your ROMs with the PlusCart on a "normal" 2600?

 

btw: The discussion about the Janice chip is very interesting, but I think it should be in an own thread. Maybe one of the moderators can move the posts to a new thread?

 

[Thomas Jentzsch]

1 hour ago, Al_Nafuur said:

I don't have a list, but a search for ".noexit" in the PlusStore repository finds only three ROMs:

• Bad Apple Demo
• Indy 500
• Crazy Ballon

Please add VROOM!

1 hour ago, Al_Nafuur said:

btw: The discussion about the Janice chip is very interesting, but I think it should be in an own thread. Maybe one of the moderators can move the posts to a new thread?

100% agreed.

[+Al_Nafuur]

2 hours ago, Thomas Jentzsch said:

Please add VROOM!

Are you sure?

 

[JetSetIlly]

6 hours ago, Thomas Jentzsch said:

BTW: In which way could the 6507 be different?

Who knows? 🙂 I suppose it depends on who made the chip and how. I'm not ready to discount the possibility of bugs in the CPU core. This chip came out after the 65C02 was available so I'm even prepared to believe it's based on the 65C02 core.

 

I honestly don't know but the results we've seen so far are wild, it's causing me to wonder about the entire chip.

 

As a simple test we could try a ROM which we know uses undocumented instructions. If the console crashes then that would be a good first clue that something is different.

 

 

[Thomas Jentzsch]

1 hour ago, Al_Nafuur said:

Are you sure?

It is a combination:

 

[Kupakai]

Sorry I haven't replied. Was a bit busy the last week.

 

On 8/15/2024 at 7:36 AM, Al_Nafuur said:

I don't have a list, but a search for ".noexit" in the PlusStore repository finds only three ROMs:

• Bad Apple Demo
• Indy 500
• Crazy Ballon

I just retested "Cosmic Ark" (Public ROMs/Classic Roms/NTSC/BY ALPHABET/A-G/Cosmic Ark.bin) on my system and it is not affected by the unintended exit. So maybe @JetSetIlly is right that there are more differences with the Janice chip. @Kupakai can you retest your ROMs with the PlusCart on a "normal" 2600?

 

The Cosmic Ark and Halo work fine without turning off noexit on a normal 2600. 

Vroom also seems to require noexit setting on the Janice 2600, but not on a normal 2600. Also Vroom's startup screen is messed up on the Janice 2600

 

On 8/15/2024 at 6:04 AM, Thomas Jentzsch said:

@Kupakai How many of these consoles do you have?

I have two Janice 2600.

 

Just as an interesting aside, I also tried the Pluscard SD on Sear Video Arcade II (Sears version of the 2800) and got an interesting result in the menu and Halo. The second video is when I ran Halo again and the video corruption was different from the first time. The second video showing the vertical line is odd because you can only shoot horizontally in Halo, and it sometimes glitches and sometimes does not.

 

Cosmic Ark looked like this:

 

 

On 8/15/2024 at 2:45 AM, JetSetIlly said:

That's very interesting. I'm not sure what to make of the fact that two different consoles of the same type are producing different results.

 

I'm a little reluctant to spend much time on this until we understand more about the Janice chip. The fact that we have two different results from two different machines makes me think there's something about these chips that we don't know about it. For instance, do we even know if the 6507 part of the chip performs the same as the MOS 6507?

I wasn't thinking anyone would spend time on the glitch, especially, Video Arcade II, since both are pretty uncommon variants. I just thought I'd post the issue just in case someone else ran in to it. However, if someone wants to look into this, I can do any testing they'd like.

[JetSetIlly]

19 minutes ago, Kupakai said:

Also Vroom's startup screen is messed up on the Janice 2600

Is this screen moving? Based on what we know so far, this is what I think it will look like. Is this close?

 

 

21 minutes ago, Kupakai said:

I wasn't thinking anyone would spend time on the glitch, especially, Video Arcade II, since both are pretty uncommon variants. I just thought I'd post the issue just in case someone else ran in to it. However, if someone wants to look into this, I can do any testing they'd like.

It's an interesting problem and I'd like to better understand what the causes are. Even if we don't find a solution, it's worth investigating I think 🙂 All information adds to our knowledge of the 2600.

 

Can you test the homebrew game "astar"? It's in the PlusROM Homebrew section.

"Public ROMs/Homebrew/NTSC/A-D/astar_ntsc.bin"

 

Start a game and see if it plays correctly.

Thanks