Intellivision III and Intellivision IV

[ghsqb]

Most of us here are aware that in the past Mattel had begun work on an Intellivision III and Intellivision IV.

 

For those that weren't aware, of if you just want a refresher, you can find details here:

Intellivision 3: http://www.intellivisiongames.com/bluesky/hardware/intelli3_tech.html

Intellivision 4: http://www.intellivisiongames.com/bluesky/hardware/intelli4_tech.html

 

So to all the Intellivision fans here, in a world with no video game crash, what would the Intellivision IV have been?

Same features as the Intellivision III, but evolved, or completely reimagined?

 

Would features would you want?

[TrekMD]

Cool, thanks for posting those links!

[sqoon]

Yes, good reads. The more I read, it looks like Mattel was really an indecisive mess at that time. It really is amazing they ever got Intellivision off the ground, considering their overall wishy washy attitude toward gaming related projects.

[intvnut]

The Intellivision IV was a completely new system based around the MAGIC display processor and the 68000 CPU. Originally this was called the Intellivision II, but its name got bumped up a couple times.

 

The MAGIC display processor is described in detail in the spec available on the Dave "Papa Intellivision" Chandler memorial site.

 

MAGIC stands for Mattel Advanced Graphic Interface Circuit. It was to be the heart of the new system, alongside a Motorola 68000 processor. The system would be completely incompatible with all existing Mattel games, except maybe through emulation. You could make the argument that the MAGIC-based Intellivision IV was a revolutionary machine as opposed to evolutionary. Based on the specs, it would have been in some ways beyond the NES and almost to the Sega Genesis in capability.

 

The Intellivision III, which came along later, was to be an upgraded, compatible-to-Intellivision-I machine. It featured an upgraded CP-1610-2 CPU that ran much faster than the CP-1610 (by demultiplexing the bus and other updates), and an upgraded STIC that had more MOBs. It really was an evolution of the existing Intellivision line. You can read some about about the Intellivision III at Dave Chandler's site also.

 

Based on what I've heard from folks like Don Daglow (IIRC), the Intellivision IV was the earlier concept (as I said, starting off as the Intellivision II), and the other guys came in as pragmatic stops in-between. When you look at the Intellivision IV and the Keyboard, you can see that some such as Dave Chandler extremely forward-thinking. But, pragmatics kept drawing them back. I wouldn't call this wishy washy. This is very typical in the high-tech world. I should know, having worked at a high tech company the last 16 years... these forces are still at play everywhere, just some of the flameouts are more public than others.

 

EDIT / Addendum:

 

And one add'l aside: I buy my microprocessors for Intellivision boards from a company that inherited all the electronics from General Instrument (the company that made most of the chips in the original Intellivision). That company is Microchip, based in... Chandler, AZ. And my boards themselves are manufactured by a board shop also based in Chandler, AZ.

 

I *DID NOT* plan this. I just realized it now.

Edited November 3, 2012 by intvnut

[ghsqb]

So if we accept improved sound & graphics, integrated Intellivoice & ECS, and wireless hand controllers what other features would Intellivision fans want?

 

Would you want the hand controllers to have a joystick or the disc?

What about other controllers like a driving controller?

 

Any other features?

[atari5200dude82]

That's very cool. I wish they would released it back in the day. Is it still possible for someone to still make this kind of console. One can dream.

[+Gemintronic]

It seems like the INTVs biggest asset was the BIOs calls. I bet they could have gone without a single hardware change except a bigger ROM chip. Packed it with the routines programmers craved and painted the case a different color.

[ghsqb]

It seems like the INTVs biggest asset was the BIOs calls. I bet they could have gone without a single hardware change except a bigger ROM chip. Packed it with the routines programmers craved and painted the case a different color.

 

Hmmm thats an interesting idea.

 

With the Colecovision out, and the Atari 5200 & 7800, and even the NES on the horizon, you don't feel that they would have had to improve graphics or add any flashier features?

[ghsqb]

That's very cool. I wish they would released it back in the day. Is it still possible for someone to still make this kind of console. One can dream.

 

What features would you have wanted to see?

[+DZ-Jay]

What features would you have wanted to see?

 

Higher resolution, bitmapped graphics, and multi-coloured sprites. Also, more than 8 moving objects at a time.

 

-dZ.

[intvnut]

That's very cool. I wish they would released it back in the day. Is it still possible for someone to still make this kind of console. One can dream.

 

You can do great things with emulation... The system isn't the tough part. Getting a catalog of game on it is the real trick. If I gave you an Intellivision III today, could you pony up developers to write games for it? It's hard enough to find folks willing to spend time on systems that already exist. :-)

 

And, if you're building a "new, old system" in software like that, why not just code the games directly on the new hardware? Kinda like these retro-looking games that are actually flash-based games or similar.

 

The only reason I could see to build one of these machines would be to prove that it could be built, much like the folks trying to build the Difference Engine.

 

It seems like the INTVs biggest asset was the BIOs calls. I bet they could have gone without a single hardware change except a bigger ROM chip. Packed it with the routines programmers craved and painted the case a different color.

 

Hmm... I've got all the ROM in the world available to me now, but I still feel a little cramped in the 160x96, 16-color, 8 single-color MOB display. I got pretty good at the on-screen shell-and-pea game writing Space Patrol to make you think it was doing more, but I'd really rather have more actual MOBs, more color on the MOBs, more pixels, etc.

[atari5200dude82]

You can do great things with emulation... The system isn't the tough part. Getting a catalog of game on it is the real trick. If I gave you an Intellivision III today, could you pony up developers to write games for it? It's hard enough to find folks willing to spend time on systems that already exist. :-)

 

And, if you're building a "new, old system" in software like that, why not just code the games directly on the new hardware? Kinda like these retro-looking games that are actually flash-based games or similar.

 

The only reason I could see to build one of these machines would be to prove that it could be built, much like the folks trying to build the Difference Engine.

 

 

 

Hmm... I've got all the ROM in the world available to me now, but I still feel a little cramped in the 160x96, 16-color, 8 single-color MOB display. I got pretty good at the on-screen shell-and-pea game writing Space Patrol to make you think it was doing more, but I'd really rather have more actual MOBs, more color on the MOBs, more pixels, etc.

That would be cool. Have more pixels and enhanched graphics . Maybe built in Intellivoice. I love the original Intellivision. But i would like to see a upgrade on it. Like either a super pro expansion , that plays new games . Or a new console with new and superior graphics,and sound , better than the Colecovision. Edited November 4, 2012 by atari5200dude82

[Walter Ives]

On 11/2/2012 at 9:17 PM, intvnut said:

It featured an upgraded CP-1610-2 CPU that ran much faster than the CP-1610 (by demultiplexing the bus and other updates)

Nope. It used the same bus-multiplexed CP1610A Valeski used on the last INTV IIIs. The original CP1600 and the CP1610 devices used exactly the same chip, the only difference was that former was packaged in ceramic and the latter in plastic. That part was designed to run at 5 MHz and could in fact do so until it got too hot. It ran just fine at double Intellivision speed (3.58 MHz) as long as you arranged to dissipate the generated heat. The Intellivision III emulators used at the 1982 Winter CES used ceramic packaged CP1600s, as the CP1610A was still in layout. The CP1610A was actually created for use in the cost-reduced Intellivision II and would have been swapped in as a running change once the existing inventory of CP1610s were used up. It was implemented in a newer 5-volt process that didn't need VDD or VBB but was otherwise functionally identical to the original. It could run at double speed without a heat sink.

[Walter Ives]

On 12/3/2022 at 5:06 AM, Walter Ives said:

The Intellivision III emulators used at the 1982 Winter CES used ceramic packaged CP1600s, as the CP1610A was still in layout.

Should read "1983 Winter CES," which took place in January 1983. Sorry.

 

WJI

[Walter Ives]

On 11/2/2012 at 6:56 PM, ghsqb said:

Most of us here are aware that in the past Mattel had begun work on an Intellivision III and Intellivision IV. https:

 

//www.reddit.com/r/intellivision/comments/l4ghde/comment/gkpjozw/ DerpDerp3001 (January 24, 2021): This is what the Intellivision III (the unreleased system, not the INTV System III) sprite would have looked like according to hardware specifications.

    Well, that was the obvious next step, eh, and STIC 1b could certainly do that. But the design team was in fact a bit more clever.

    Moving object capability was constrained by the number of GRAM fetches one could make during horizontal retrace and the width of GRAM. The original STIC could do eight 8-bit fetches per horizontal retrace—one per moving object. The other constraint was the number of shift registers one could fit on a chip: parallel load bi-directional shift register cells are not exactly tiny. Eight 8-pixel-wide objects was right balance for that generation of technology. The TMS9918 used in Colecovision also did eight pixel fetches per horizontal scan line but it needed two per sprite because its sprites were sixteen bits wide.

    When time came to design Coffee/STIC 1b (the graphics chip used in Aphix/Intellivision III) semiconductor technology had advanced to using 2 micron lithography HMOS so it was practical to increase the size of GRAM, put more bits of shift register on chip and run everything at twice the speed. The big constraint was time to market: the project was given the go-ahead in the summer of 1982 because it was becoming clear to Mattel's management that MAGIC/DECADE could not be ready before 1985 and the company was going to have nothing to sell for all of 1984.

    Given that the design and internal timing of the cost-reduced STIC 1a had all been worked out and the chip was in layout, it was proposed that its design could be expanded relatively quickly and with minimal risk into what came to be called STIC 1b, with that part following six months behind STIC 1a. More-or-less working STIC 1a parts were ready by Spring 1983. The plan was to have enough STIC 1b parts for a shakedown product test distribution by December 1983 with volume shipments to follow at the start of the new year.

    MAGIC (the graphics chip used in DECADE/Intellivision IV) was a much more ambitious design that was intended to last for a decade. It was intended more for home computers than for dedicated game consoles. As to objects, the design called for up to sixty-four 16-bit fetches during horizontal retrace to fetch pixel data for sixteen different 16x16 three-color objects with sub-pixel edge-resolution ("super resolution"). The large number of fetches was made possible by using page-mode addressing of dynamic RAMs. As to super resolution: on the image shown above, the position of the edges of, say, the blue pixels could be adjusted to produce a smoother outline of the dress. This unusual encoding scheme was chosen because NTSC encoding allows the luminance to change more quickly than the chrominance. The resulting chip was not only uncomfortably large, but the complexity of its design rivaled if not exceeded that of contemporary microprocessors. The MAGIC/DECADE design team also embraced the new paradigm of going straight to silicon without first building an emulator. As a result Mattel had difficulty finding a semiconductor vendor willing to undertake the project.

    The STIC 1b design did not come near to pushing the limits of the then current technology. This meant that even with STIC 1b, Mattel would still vulnerable and it was likely Aphix/Intellivision III would face stiff competition in 1985. In the course of its design the design team recognized that it would be practical to compatibly extend the STIC architecture for several more generations. Aphix II, the next generation, added among other things an additional fetch per object to enable it to produce sixteen-bit-wide 7-color objects. The STIC 1b design team moved on to that project at the end of January, 1983, immediately after delivering the STIC 1b design documents and emulator to Toshiba for layout.

 

    WJI