A few questions

[applekevin]

First...Are they're any issues with the FB2 mod? Can I fix them? What games don't work?

Is it possible to make a 2600 cart without ruining another cart?

Can I play the ROM from my flashback 2 on Stella (the emulator)?

 

Second...Is there anyone archiving the ROMs for all these 2600 games? I know that most people don't care about pirates but I'd still like to play them.

 

Thanks!

[autumn_drag0n]

You might want to post your questions to the Dedicated Systems/Atari Flashback Consoles forum.

 

As far as archiving ROMs, do you mean putting them all into one .zip or .rar file? If so, yes I am sure with enough searching you can find people have done this (check filesharing networks: P2P/Gnutella, or BitTorrent)

 

Or where you can download ROMs? If so, yes you can download almost every Atari 2600 ROM ever made here on AtariAge.

 

Or maybe I just don't understand the questions

[BassGuitari]

Welcome to AtariAge!

 

As far as I know, there are no compatibilty issues with the FB2 cartridge modification. And I suppose it'd be possible to make a cartridge without recycling an old one, but it'd be difficult (how do you manufacture new cartridge shells?). PCBs are available though, so if the shell isn't a big deal you could just use those and go caseless.

 

Most of the ROMs from the FB2 are up here already - and have been for ages. However, some of the FB2 exclusives like Adventure 2, Yars' Return, and Return to Haunted House are not. But, I haven't looked in a while, so someone feel free to correct me.

 

 

 

Oh, FYI, you should post Flashback related stuff in the Dedicated Consoles forum. Might save you some grief in the future.

[applekevin]

Thanks for the tip!

I know that there is almost every ROM on AtariAge, but do they archive the pirates and what not?

 

Also where do you buy the circuitry to make carts?

 

P.S. Are the carts in the AA store recycled or made fresh

 

P.P.S. I could make the cases

 

P.P.P.S Are they're any games that require or greatly benefit from controllers other than the joystick?

 

THANKS!!

[supercat]

As far as I know, there are no compatibilty issues with the FB2 cartridge modification.

 

There are compatibility issues, at least with rev A and B and I believe C units. I think Curt talked about a new rev that would fix some of the known problems, but I don't know if it was ever built.

 

Issues:

 

-1- Some games use undocumented opcodes on the 6502. From what I understand, some of these opcodes--but not all--are implemented on the FB2. Games which use such opcodes will behave oddly if not outright crash on the FB2.

 

-2- Some games use multiple RESPx operations on a scan line to display extra sprites. When multiple sprite copies are enabled, a real 2600 will not display the first one following RESPx but will display subsequent ones. The FB2 would display them all, causing extraneous objects to appear in some games.

 

-3- Some games use tricky write sequences to jam one or more of the horizontal pulser circuits. This approach is used for things like the starfield in Cosmic Arc. The FB2 does not support this; games which rely on this may show a solid vertical line instead of a "starfield".

 

-4- Some of the memory timing is different from a normal 2600. I haven't looked at the actual waveforms, but carts which act upon address changes (rather than just address patterns) often won't work. These include the Supercharger and, I think, Robot Tank.

[applekevin]

Wait...What do you mean by rev A B and C? Are you talking about the FB2? Also are there anyway to fix these issues?

[atwwong]

Wait...What do you mean by rev A B and C? Are you talking about the FB2? Also are there anyway to fix these issues?

There were three different "runs" (actually, there were at least 4 as Rev.A went through at least one revision) of the FB2. Each run has a different version of the hardware. If you look through the posts on FB2 mods in this forum, you can see the different look of the hardware circuitry and boards. Rev.B correct some of the Atari TIA chip inaccuracies on the Rev.A; Rev.C has some cosmetic changes to make adding a cart slot to the board easier, but AFAIK the hardware works the same as a REV.B. To fix the TIA inaccuracies, a Rev.D would have to be produced. Maybe the FB2 Portable will have an even more accurate TIA chipset "emulation".

[applekevin]

Will this Rev D ever be released?

[atwwong]

Will this Rev D ever be released?

Keep your fingers crossed for the FB2 Portable.

[applekevin]

So I'm assuming this FB2 portable will be 100% compatible

[supercat]

So I'm assuming this FB2 portable will be 100% compatible

 

It will likely fix known incompatibilities. That doesn't mean it will be 100% compatible, however.

 

Both the 6507 and TIA are built using NMOS technology with dynamic logic. Today, logic is all done with CMOS technology, and constructing NMOS chips is no longer practical. Consequently, it's not possible to simply copy the 6507 and TIA designs; they must be re-engineered. This in turn requires an accurate assessment of exactly how the chips work, and that's not as simple as it sounds.

 

For example, the TIA has a number of race conditions within it, where a stimulus will trigger two chains of events, the outcome of which will depend upon which one completes first. Some of the TIA's race conditions are unpredictable and can vary with temperature, the phase of the moon, etc. Others, however, are absolutely consistent. Someone studying schematics of the part won't be able to tell which way race conditions will go. Only experimentation with actual parts will show how real systems handle race conditions.

 

The number of potential race conditions in something like the TIA is huge. Most of them, however, don't really matter. Unfortunately, there's not really a good way for automated tools to examine a circuit like the TIA and determine which race conditions are important, nor to test all the potential race conditions. Consequently, it's necessary for someone familiar with the device to select for examination the race conditions that matter; if the person misses one whose outcome is important to a particular game, and if the designer of the new chip didn't happen to guess properly, incompatibilities may arise.

[ijor]

Both the 6507 and TIA are built using NMOS technology with dynamic logic. Today, logic is all done with CMOS technology, and constructing NMOS chips is no longer practical. Consequently, it's not possible to simply copy the 6507 and TIA designs;

 

Any internals of the original NMOS 650X are publicly available at all (internal schematics, net-list, or whatever) ???

[jbanes]

Any internals of the original NMOS 650X are publicly available at all (internal schematics, net-list, or whatever) ???

How's This?

 

Edit: I suppose that's not what you're looking for. 6502.org has a lot of the types of documentation you're talking about. But the problem isn't in understanding the differences, it's in compensating for them. CMOS technology is simply more efficient than NMOS technology. While that's usually a good thing, it fails in the case of the 2600.

 

You see, the 2600 is one big hack to reduce gate-counts. By using some pretty wonky logic combined with hand-tweaked timing delays (almost analog in nature), Jay Miner was able to produce an exceptional graphics chip using very, very, very little silicon. The problem comes in when trying to reproduce those timings. Since CMOS is faster and more efficient than NMOS, gate changes propogate much too fast to reproduce the original timings. As a result, some of the more advanced features of the TIA are hard to reproduce using modern manufacturing methods. You'd almost need to build a *new* schematic that intentionally implements all the tricks that the original 2600 programmers used. Which is a rather daunting task when you think about it.

Edited December 28, 2006 by jbanes

[ijor]

Any internals of the original NMOS 650X are publicly available at all (internal schematics, net-list, or whatever) ???

How's This?

 

Thanks for the link, but that's just a VHDL binary code compatible core. No relation with the original internal design.

Edited December 28, 2006 by ijor

[jbanes]

Thanks for the link, but that's just a VHDL binary code compatible core. No relation with the original internal design.

Look at my post again.

[ijor]

Hi jbanes,

 

Yeah, sorry, I replied too fast

 

Edit: I suppose that's not what you're looking for. 6502.org has a lot of the types of documentation you're talking about.

 

Well, I was going to reply that I am a regular visitor to 6502.org. And that there are no internals whatsoever there. Only datasheets, manuals, etc, that they were available all the time. No internal or other undisclosed info. But ...

 

...since you posted the reference, I took the link, checked the "news", and found this:

 

Dave Dixon sent in a link to Reverse Engineering the 6502. These Hungarian-only pages include high-resolution scans of the actual 6502 silicon and an impressive attempt to make a schematic from them

 

Difficult to follow because it is not in english. But seems there are impressive internal schematics done by physically reverse engineering the silicon!

 

Not the same as official internal schematics (again, are there available at all?), but I had no idea somebody did that.

[jbanes]

FWIW, the 6507 isn't really the problem. The 6502 design has been done to death, and is pretty well understood. (Though I imagine that the interleaving of the clock drives chip copiers up a wall.) The only issue I've heard of are some issues with not supporting unsupported opcodes used by several 2600 games. As you're probably aware, those opcodes were accidental decodings rather than official instructions. Support could be added if one was so inclined, however.

 

The TIA is far more problematic, despite the fact that detailed schematics exist. I seem to remember that the RIOT wasn't exact either, but I don't remember why. Something about some of the lesser used controller pins not being implemented for output?

[ijor]

FWIW, the 6507 isn't really the problem. The 6502 design has been done to death

 

It might be done to death, but last time I checked (a few months ago) I couldn't find any free IP core compatible with the original NMOS part. They were all compatible with the newer CMOS ones.

 

Anyway, I wasn't asking specifically for the purposes of this thread. It was just out of curiosity and general interest to know if the original 650X schematics are available or not.

[supercat]

It might be done to death, but last time I checked (a few months ago) I couldn't find any free IP core compatible with the original NMOS part. They were all compatible with the newer CMOS ones.

 

Out of curiosity, which quirks of the 6507/TIA are relied upon by one or more back-in-the-day or homebrew titles?

 

Some TIA quirks that are certainly used are cycle 73/74 HMOVE (homebrews only I think), multi-RESP (used in Galaxian and some other games BITD), and confusion mode (used only with Missile 0 BITD, but equally applicable to all objects). As for the 6507, I'm aware of LAX, SAX, DCP, SBX, and ASR. Probably INS and maybe some of the other RMW+ALU ops. A test version of the Toyshop Trouble kernel used LAS (anyone else ever use that?) What other things have people used?

 

Do any games (old or new) rely on the exact flag operations of ADC/SBC in decimal mode? Do any rely upon the bizarre behaviors of some other undocumented opcodes in decimal mode? I would guess decimal mode would probably be one of the hardest things to get "100% right" in silicon, but I would also guess it would be the least likely to be exploited beyond the obvious features (e.g. the behavior of BCD math when one or more input digit is >=$A).

[jbanes]

I would guess decimal mode would probably be one of the hardest things to get "100% right" in silicon, but I would also guess it would be the least likely to be exploited beyond the obvious features (e.g. the behavior of BCD math when one or more input digit is >=$A).

Wouldn't decimal mode be a non-issue though? The 6507 didn't support it, so no one relied on it there. And I presume that the 6502C (Sally) *did* support BCD on the 7800, so we know that no one was hacking any of the dummy BCD opcodes/flags for their own nefarious purposes.

 

BTW, I bumped into this thread:

 

http://www.atariage.com/forums/index.php?showtopic=23176

 

It might answer some of the questions you posed.