Atari VCS: Interviews with Joseph Decuir and Ron Milner
These are two separate interviews conducted by Henner Thomsen of Stay Forever with Joseph Decuir and Ron Milner who both worked on the original Atari VCS. Stay Forever is Germany’s leading retro gaming podcast.
Interview with Joseph Decuir, who worked on the original Atari VCS, its graphics/sound chip and some of its games, before moving on to Atari’s 8-bit computers and the machine that ended up as the Commodore Amiga.
Stay Forever: Did you know you would be assigned to the VCS project when you applied for the job at Atari/Cyan in December 1975? Some sources claim the Stella project was top secret (which makes sense), so I guess they wouldn’t have mentioned it in the job description.
Joseph “Joe” Decuir: I had an idea, but I did not know.
I knew something of the history of the company. They had pioneered arcade video games. They had taken a game concept from the Magnavox Odyssey, and radically improved it to make Pong. They then designed more complex arcade games, like TANK (which I encountered that summer at Disneyland) and designed a consumer version of Pong – connected to home television sets. Based on the questions I answered, I correctly deduced that Atari would bring the complex video games home.
I chose Atari vs. another job offer, to learn large scale integrated circuit design from Jay Miner; I chose a mentor.
Stay Forever: What were the development goals defined by Atari for Cyan’s VCS project – a specific type of game that should be playable on the system (like TANK), a production cost limit? Or did you have free rein in your design process? Was Atari concerned of home consoles rivalling their coin-op games, possibly limiting the VCS’ power?
Joe: We were constrained.
First, Atari was determined that this machine display a decent TANK game. The initial hardware and software designs reflected that. My first job: debug their prototypes of that hardware and software. They also wanted it to play other complex games, and simpler games line Pong.
Second, they had to make an affordable system. Marketing people specifically looked around at other consumer entertainment products that cost around $200 retail. Candidate: audio cassette players.
They decided that they cost of manufacturing had to be 1/3 of that, or around $66 USD in 1977.
At that time in history, an inexpensive new car cost about $2,500, and a used house cost around $25,000. In fact, when it was introduced in 1977, an Apple II personal computer listed at $1,200.
We did not think the VCS would approach the arcade games for complexity. The only concern was that the consumers would spend their money on a VCS and cartridges vs. quarters in the arcade.
On the other hand, the Atari PCs (8-bit computers) were designed so that arcade game ports were easy.
In fact, the first animation hardware design that clearly overtook the arcade machines was the Amiga. Atari signed a deal to license the Amiga chipset for arcade games, but that was terminated when Commodore bought Amiga.
Stay Forever: Before you started working on the VCS, did you know about Fairchild’s Channel F project or the RCA Studio II – or when did you hear about them? Were you inspired by them (apart from the cartridge design which was made by the same person as the Channel F’s, if I’m not mistaken)?
Joe: The Channel F was announced at CES in June 1976. In fact, I was acquainted with [Channel F developer] Jerry Lawson from Homebrew Computer Club, but I did not know what he was working on.
In the architecture stages, we considered how to deliver programs. The main candidates:
- Enough RAM to load data from an audio cassette
- ROM
The latter solution was much cheaper. Channel F validated that decision.
IMO, the reason that the Channel F failed in competition with Atari VCS:
- Atari was an entertainment company that did electronics
- Fairchild was a semiconductor company looking for markets
As an entertainment company, Atari knew about people:
- They want decent resolution for their avatar, less resolution for the background images. The VCS had low resolution for the ‘playfield’ and higher resolution for the ‘players’ and ‘missiles’ [sprites], with hardware motion control.
- Fairchild used a simple bit-map (more cost) with the same resolution for avatars and the background. All moving objects were animated in software, not in hardware.
Result: Channel F games were ‘clunky’ to play. Atari’s were a lot faster and smoother to play.
I have a copy of the 1977 Consumer Reports Magazine that gave the VCS top rating.
I never heard of the RCA Studio II – or forgot it.
Stay Forever: Did you ever consider including a ROM chip for some basic included games, a character set or something else?
Joe: We barely made the budgeted cost of manufacturing. Since it was just a game player, we decided to bundle a game – no ROM included. The Atari version came with Combat; the Sears version came with Video Olympics.
The hardware did not support displaying characters at all. The character displays developed later, for chess and for the BASIC cartridge, which were interlaced and sometimes flickered.
Stay Forever: Looking at the machine itself, I find the decision to include difficulty switches particularly surprising. It’s a brilliant idea, but considering that you tried everything to minimize component count and cost, it’s an odd decision. Could this functionality not have been implemented via software, as an in-game switch?
Joe: Again, Atari was an entertainment company that did electronics. We were thinking about how to create fun experiences. We understood that there would be disparities in gaming ability, so we included the difficulty switches to help: e.g., father vs son, older brother vs younger brother, etc.
This was personally important several years later. In 1988, my oldest (6-year-old) son came home from school. He told me that he had learned to play handball. I pulled my VCS out of the garage, and set it up to play the handball games in Video Olympics. We started out with a handicap for me: my paddle was half the size of his, etc. Within a few games, we were both playing at the same level. He is still a committed gamer, who says ‘Dad, you were cool once’.
Stay Forever: Did you expect game developers to repurpose the Difficulty, Select and even Color/BW switches for in-game functions (as Activision’s Space Shuttle does, for example)?
Joe: Yes. Again, Atari was an entertainment company, thinking about how to make fun experiences.
Also, the Color/BW switch was used to make the colors distinguishable on a BW TV set.
Stay Forever: Did you expect the console to be flooded by third-party games?
Joe: No, that was a big surprise. The original design was so complex that we thought it might be difficult to reverse engineer. Later, in the court battle over the foundation of Activision, I testified that it could be reverse engineered, buy running a game with a commercial logic analyzer (e.g., HP 1611) attached.
We always expected the competition would be fierce, and it was. Nolan [Bushnell] insisted that we start working on the sequel immediately, which became the PCs. It was designed to make porting existing arcade games straightforward, and it met that requirement.
Stay Forever: When you switched from a console to a real computer with the Atari 400/800 in 1979, did you expect home computers to fully replace gaming consoles?
Joe: Jay [Miner] always said that a great home computer would play great games. We had a dilemma: are we designing a great home computer, or a great follow-on game machine, or both? We ended up designing two systems with a common new chipset. These became the 400 and 800.
Doug Neubauer, who designed the POKEY chip in that system, got bored after the chip worked the first time. He then wrote the Star Raiders video game cartridge for the PCs [400/800]. Atari marketing was so impressed that they changed the design of the 400; it came bundled with a membrane keyboard – so users could play Star Raiders.
Big problem for Atari: the new machine could not play VCS games; it was too big a departure. Meantime, the VCS games market turned out to be much larger than originally expected. Why: the game programmers were very clever, and found new ways to exploit the hardware. We (in hardware) were trying to be cheap. The result was that we put control of the vertical display in software.
We originally expected to sell the VCS for 3-4 years. It sold over 30 million units in 10 years.
Because we had made VCS games obsolete, we created market openings for other systems. Mattel’s Intellivision did well. Atari created the 5200, using the PC chipset, but it couldn’t play VCS or PC games. They upgraded the VCS hardware to the 7800, which could play VCS cartridges, but it was too late. In 1985, Nintendo stepped in to establish a new game console programming model.
Stay Forever: Did you – or anyone in Atari/Warner management – expect home computers and/or home consoles to replace coin-up arcade games in the eighties?
Joe: We were creating home gaming to supplement the arcade revenue.
Stay Forever: What was the biggest difficulty you faced when you developed the PAL version?
Joe: All the timing for the USA/NTSC systems were based on the NTSC color clock: 5×63/88 MHz = 3.579545 MHz. The color clock in the PAL systems is different. I don’t remember the derivation, but it is around 4.4 MHz. Our solution: create a 4/5 divider, so most of the logic runs around 3.5 MHz. Obviously, the systems drive a 50 Hz frame rate vs 60 Hz NTSC frame rate. This design sometimes needed software modifications, too. We designed a PAL version of the TIA chip for the VCS, and PAL versions of the ANTIC and GTIA chips [for the 400/800 computers]. Later, Jay & I designed PAL versions for the Amiga chips, too.
Stay Forever: Some sources claim that Atari envisioned selling about a dozen different games total for the VCS/2600 (more than 500 were actually released by all parties), is that true?
Joe: Yes. The marketing people expected that the design would last maybe 3-4 years in the market before the competition killed it, selling maybe 4-6 game cartridges per system. They did not understand that the hardware put control of the vertical in the hands of the game designers. (See [the book] ‘Racing the Beam’.)
Stay Forever: You did most of your VCS work at Atari HQ, not in the Cyan offices, correct? Considering Atari was well known for its lax work environment, with cannabis use and whirlpool meetings, how did you get any work done?
Joe: Focus. I spent 2 months at Cyan, and 16 more months working on VCS at HQ. Last task: creating QC test software for production. I then spent two more years working on PCs, again ending with QC ROMs. Exercised helped. I bicycled 7 miles each way to work and back. I would get good ideas on those rides.
Stay Forever: You once said that you regretted limiting the VCS’ address space to 4K and the ROM cartridge pin count. Is there anything else – in hindsight – that you should have changed about the VCS architecture?
Joe: I had an epiphany in Winter 1977: we should have spent an extra dollar on the product, with a 40-pin CPU (6502) instead of a 28-pin CPU (6507 – same chip, with less pins brought out) and a 30-pin game cartridge connector instead of a 24-pin cartridge. Why: allow the VCS to evolve to a computer and/or a better game player, rather than abandon the VCS game market, and make the customers buy new games for a new machine. Nolan and I still have this argument.
Otherwise, I would have left the original design intact. However, we should have also sketched out the design of our next machine such that a new machine could recognize and play an original VCS game OR a new game for a successor machine. We call that a roadmap; it could have led to the 7800, which plays the same VCS cartridges.
Stay Forever: What part or aspect of the VCS are you most proud of?
Joe: We put the design of the vertical in the hands of the game designers, who were a lot smarter than we (hardware designers) realized.
Stay Forever: Do you still own – or even use – a VCS/2600 console?
Joe: I have three original Atari VCS consoles of various vintages, starting with a ‘heavy-sixer’ and ending with an ‘Atari Jr’. All play game cartridges, but they generate RF output; you need an obsolete TV to play the games. I also own three Flashback 2.0 consoles, modified to play game cartridges. They have composite video outputs, so they can be played on some modern monitors. Why: I have several children, and they all want one. With luck they will play with my grandchildren. On the other hand, they don’t seem interested in PCs or Amiga machines.
Stay Forever: What do you think of the new ‘Atari VCS’ revival console?
Joe: I don’t think of it at all. I don’t own one.
Interview with Ron Milner, who worked on the original Atari VCS at Cyan Engineering, co-creating the very first prototype machine.
Stay Forever: Did anyone at Cyan have any experience with video games like SpaceWar!, which was played on quite a few universities back then? Bushnell claims to have played it in the sixties (it’s more likely he played it later though).
Ron: I played Snakey in the physics computer lab in 1966. They had an oscilloscope rigged up to an IBM 1620 to give a 32×32 array of dots Each player controlled which way their snake (a line of 16 dots) went trying to run through the side of their opponent.
No idea if others had played computer games before.
Stay Forever: I’ve read that Atari was Cyan’s first and only customer before the merger, is that true or did Cyan work for other customers as well?
Ron: Cyan started as Larry Emmons and Steve Mayer’s consulting company aiming to make a video loop recorder. After some successful consulting for their old friend from Ampex, Nolan Bushnell, Atari bought out the consulting company but kept the name.
Stay Forever: Atari back then was well known for its lax work environment, with some cannabis use and whirlpool meetings. Is that true or just a myth, and how was working at Cyan in comparison? I watched a TV documentary about Cyan from 1982 (here: www.youtube.com/watch?v=bfFGrQLuY8s) and you seemed pretty relaxed.
Ron: Cyan was quite relaxed but never the party and pot environment Atari was. We were all good friends and helped each other build our houses and other less exciting stuff.
Stay Forever: Cyan worked on non-gaming projects for Atari in the eighties, including a video phone. Did Cyan’s engineers start projects at will, or did you always follow Atari’s orders?
Ron: Most of our projects were our ideas and inventions at Cyan.
Stay Forever: Before you started working on the VCS idea, did you know about Fairchild’s Channel F project? Did you start in 1975 (probably even before Home Pong was released) or 1976, since several sources give different years?
Ron: We knew nothing of Channel F. Our work on the VCS (later referred to by the model number on the package, 2600) was only a very few months culminating with Joe Decuir taking the project to Sunnyvale Atari.
I think we got serious on the project after going to Wescon 1975 – I think October – and seeing Chuck Peddle’s [the MOS Technology engineer’s] 6502.
Stay Forever: The March 1983 issue of IEEE Spectrum said about the VCS’ development, “Atari and other video-game companies had been making microprocessor-based arcade games for some time before the VCS was developed.” The earliest microprocessor-based video game that I could find, though, is Midway’s Gun Fight from November 1975. Did you work on CPU-based games before or was the VCS the first one?
Ron: We made an 8-player tank game with a 6800 running it before VCS.
Also computerized a Bally “Delta Queen” pinball machine with an Intel 4004 before that.
Stay Forever: Who came up with the idea of a CPU-based home console, was this an Atari idea or an original Cyan invention?
Ron: Cyan – mostly Steve Mayer.
Stay Forever: The IEEE Spectrum article I mentioned before claims that developing a custom, non-CPU based game took you over a year and $100,000, would you say that’s accurate? How does this compare to developing a similar game for the VCS architecture?
Ron: A discrete logic game like my Qwak video shooter took 2-3 months for a working wirewrap prototype, then a lot of work making pc [printed circuit] boards, controls, and cabinets. I would believe the $100k.
Stay Forever: Various information about the system’s internal ROM is circulating. Just to be clear, the machine doesn’t have any ROM, it’s all inside the game cartridge, right?
Ron: Yes, only ROM in the cart.
Stay Forever: The VCS was the first home console with sprite graphics support (even if it wasn’t called that), how did you come up with this concept? Was it inspired by earlier arcade machines?
Ron: Our car racing game Gran Trak used “sprites” of a sort from a PROM, and Nolan’s original Computer Space game had the graphics for the ships in a diode matrix.
Stay Forever: You personally developed (or co-developed) games like Gran Trak 10 or Combat, if I’m not mistaken. Did you enjoy developing games or did you prefer designing circuits?
Ron: Both. I didn’t do the Combat game – just put the hardware together.
Stay Forever: What kind of machines were used for VCS game development, I would guess minicomputers like DEC’s? Did you use CAD for the hardware design (which was available at the time)?
Ron: We had a DEC PDP-8 that I think Steve used for assembling the first program, then we used a Jolt 6502 development board.
Stay Forever: Looking at the machine itself, I find the decision to include difficulty switches particularly surprising. It’s a brilliant idea because it allows players of different skill levels to compete, but considering that you tried everything to minimize component count and cost, it’s an odd decision. Could this functionality not have been implemented via software, as an in-game switch?
Ron: Nobody knew about soft switches then.
Stay Forever: The VCS was shipped with the first recognizable joystick in home gaming history. Who came up with that design? Instead of bundling the machine with two different controllers (joystick and paddles), couldn’t you have combined both into one device, as Fairchild did with the Channel F?
Ron: The joystick came straight from the Tank coin-up game. Too hard to mechanically integrate a pot.
Stay Forever: Did you ever consider releasing a keyboard and other peripherals to turn the VCS into a ‘real’ computer? Peripherals like that were released by third parties, but not by Atari.
Ron: No, our group went right on to the 400/800 home computer design.
Stay Forever: The VCS was projected to have a lifespan of 3 years (it lasted 15), at least that’s what I’ve read – true?
Ron: True.
Stay Forever: You once said that you regretted limiting the VCS’ address space to 4K. Is there anything else – in hindsight – that you should have changed about the VCS architecture?
Ron: It was all completely limited by how much we could put on a reasonable cost custom chip. Optimized to the nth degree.
Stay Forever: What part of the VCS are you most proud of?
Ron: I’m proud we accidentally came up with a platform that clever programmers figured out how to do much more than we intended it to do.
Stay Forever: Do you still own – or even use – a VCS or 2600 console?
Ron: I have my original production sample. I don’t play it!