TIA differences between models

[JetSetIlly]

Has anyone ever compiled a comprehensive list of TIA differences between the different 2600 models?

 

I've been thinking about the work @alex_79 did on finding the issues in the 36 character demos and was wondering if we have similar comprehensive results of other variations, or maybe even just a list.

 

 

[Thomas Jentzsch]

 Some good info can be found here. As you can see, the number of TIA variations is quite large.

 

We tried to cover some more common glitches in Stella (Options/Developer/TIA). But it turns out, that modern mods also affect the timings (see James' console). So it will be very tough to cover each and everything.

 

 

[JetSetIlly]

29 minutes ago, Thomas Jentzsch said:

 Some good info can be found here. As you can see, the number of TIA variations is quite large.

 

We tried to cover some more common glitches in Stella (Options/Developer/TIA). But it turns out, that modern mods also affect the timings (see James' console). So it will be very tough to cover each and everything.

 

 

Interesting. Thanks.

[JetSetIlly]

3 hours ago, Thomas Jentzsch said:

 Some good info can be found here. As you can see, the number of TIA variations is quite large.

 

We tried to cover some more common glitches in Stella (Options/Developer/TIA). But it turns out, that modern mods also affect the timings (see James' console). So it will be very tough to cover each and everything.

 

Question: What is "James' console"?

[alex_79]

Just to complicate things a bit, a few years ago when testing the early bus stuffing demos, I tried swapping the TIAs between my 6-switch and vader consoles but I still got the positioning issue on the 6-switch and not on the vader. So the TIA variation is not the (only) variable.

Someday I should try installing one of the later TIA from my Jr or 7800 units into an older console. I don't know if anyone tried this before, as the later models have the IC which are soldered to the board, so removing them is not as easy as with older ones.

 

 

 

[alex_79]

58 minutes ago, JetSetIlly said:

 

Question: What is "James' console"?

James is @ZeroPage Homebrew, and the RGB mod installed on the 6-switch consoles used for playing the games during the show on Twitch seems to introduce some glitches:

 

[+SpiceWare]

@alex_79 beat me to it!
 

There's discussion throughout that topic about the issue.  Bottom of reply 34 has photos of James' 2600 using RF, such as this one:

 

 

Bottom of reply 56 has photos using the RGB mod, such as this one:

 

 

 

With the RGB mod in place the color bleeds over for the playfield pixels above the A in IMAGIC - click photos for full size if its hard to see the difference.

 

[JetSetIlly]

10 minutes ago, SpiceWare said:

 

With the RGB mod in place the color bleeds over for the playfield pixels above the A in IMAGIC - click photos for full size if its hard to see the difference.

 

 

Interesting. I've been working through the ataricompendium website that @Thomas Jentzsch linked. I've managed to replicate both conditions (playfield color in Quickstep, and playfield bits in Pesco) by moving the code that responds to playfield register changes to after the point where the TV signal is generated. So in effect, from the point of view of the TV, the playfield is updated a cycle late.

 

This makes sense because as @alex_79 says in reply 37 of the Light Sixer Repair thread, the RGB mod will add latency to the register write to the TIA. Causing any TIA on the edge of tolerance to respond as we can see in Quickstep and Pesco.

 

I've also been playing with the "2 or 3" sprite problem and have added a rudimentary emulation of the increasing operating temperature. Are there any ROMs that are affected by this phenomenon besides the one in the biglist archives?

 

https://www.biglist.com/lists/stella/archives/199901/msg00089.html

[JetSetIlly]

The "2 or 3" issue as described on http://www.ataricompendium.com/faq/vcs_tia/vcs_tia.html

 

 

The machines were that effect can be seen, do those machine also have (occasional) visible issues in Barnstormer? Does anyone have a machine of that type on which they can test?

 

If my solution to the 2/3 issue is correct then I believe we will occasionally see such issues (example image below). But if we don't see those issues in Barnstormer then that means my reasoning as to the cause of the issue is incorrect.

 

 

on edit: To clarify, the bug will only be seen during play and would be temperature related.

 

2nd edit: I can narrow down the conditions under which the bug manifests but it would be interesting to know if other ROMs (like Barnstormer) are affected.

Edited February 11, 2021 by JetSetIlly

[JetSetIlly]

Trying out the new demo of Ladybug Arcade. When running on Stella with TIA developer option "Inverted HMOVE clock phase for Missiles", the demo ROM reacts like in the attached video.

 

Is this what happens when the demo is run on a hardware 2600 with "inverted HMOVE clock phase"?

[DirtyHairy]

I don't have an affected 2600 to test, but that video looks weird. If this option makes a difference, then that means (barring a bug in Stella) that the ROM triggers HMOVE during the visible part of the scanline (either by strobing the register or by putting the chip into "starfield mode"). As a result, the sprite will receive less clocks, causing a shift. I fail to see how that can cause the effects seen on the video --- I'll have to investigate, maybe this is a bug.

27 minutes ago, JetSetIlly said:

Trying out the new demo of Ladybug Arcade. When running on Stella with TIA developer option "Inverted HMOVE clock phase for Missiles", the demo ROM reacts like in the attached video.

 

Is this what happens when the demo is run on a hardware 2600 with "inverted HMOVE clock phase"?

 

Edited February 21, 2021 by DirtyHairy

[JetSetIlly]

Just now, DirtyHairy said:

I don't have an affected 2600 to test, but that video looks weird. If this option makes a difference, then that means (barring a bug in Stella) that the ROM triggers HMOVE during the visible part of the scanline (either by strobing the register or by putting the chip into "starfield mode"). As a result, the sprite will receive less clocks, causing a shift. I fail to see how that can cause the effects seen on the video --- I'll have to investigate, maybe this is a bug.

 

It does look weird. Especially as there is no missile activity on the title screen.

[JetSetIlly]

2 hours ago, DirtyHairy said:

I don't have an affected 2600 to test, but that video looks weird. If this option makes a difference, then that means (barring a bug in Stella) that the ROM triggers HMOVE during the visible part of the scanline (either by strobing the register or by putting the chip into "starfield mode"). As a result, the sprite will receive less clocks, causing a shift. I fail to see how that can cause the effects seen on the video --- I'll have to investigate, maybe this is a bug.

 

 

This is definitely a bug. Even, if you set the Developer Settings but with the "Standard" chip type, the effect shown in the video is still visible. This is Stella 6.5.

 

 

[Thomas Jentzsch]

33 minutes ago, JetSetIlly said:

This is definitely a bug.

Nope.  The problem only appears when you enable "Drive unused TIA pins randomly on a read/peek". 

 

Looks like a programming error:

; Bank 6, f2eb:  
  ldy CXM1P ; should be ldy #1
  sty CTRLPF

 

Edited February 21, 2021 by Thomas Jentzsch

[JetSetIlly]

20 minutes ago, Thomas Jentzsch said:

Nope.  The problem only appears when you enable "Drive unused TIA pins randomly on a read/peek". 

 

Looks like a programming error:

; Bank 6, f2eb:  
  ldy CXM1P ; should be ldy #1
  sty CTRLPF

 

Heh. You're right. The same thing happens in Gopher2600 (when random pins is on)

 

I was confused by the Stella UI. I thought selecting Developer Settings in the TIA Tab would turn on developer settings only for the TIA but it turns them on everywhere.

Edited February 21, 2021 by JetSetIlly

[Thomas Jentzsch]

 

4 minutes ago, JetSetIlly said:

I was confused by the Stella UI. I thought selecting Developer Settings in the TIA Tab would turn on developer settings only for the TIA but it turns them on everywhere.

I agree, the UI is not optimal there. But the Debugger tab has no separate player/developer settings, so I had to repeat the radio buttons on each of the other tabs.