In 1983 the first laser disc-based coin-op game, Cinematronics' Dragon's Lair, came out and was a big hit, so Atari had to have one, too.

        For several years, Owen Rubin had agitated to be allowed to do a laser disc game and had always been told, "no."

        You might think he would be the one to do Atari's laser-disc game, but he wasn't.

        I believe it was because of the overly aggressive schedule. It had been decided that Atari's laser disc game would be shown at the 1983 AMOA Show to be held in October in New Orleans. AMOA stands for "Amusement and Music Operators Association" and is the main trade association for coin-operated amusement machines like video games, pinball machines, mechanical games like coin pushers and cranes, and jukeboxes.

        The schedule allowed only a few months to design the game and Owen was an experienced game designer who would have known better than to commit to such a ridiculous schedule. Somehow, Atari persuaded Mike Hally to be the Project Leader. He persuaded Greg Rivera to be the Head Programmer and Norm Avellar to be the Associate Programmer. The project was done under the aegis of Roy Machamer's research department. It was the department's first project.
 

        Then there was the hardware.

        The hardware would be a completely new design that required decoding the composite video from the laser disc player, combining it with digital graphics, and controlling the player.

        None of the experienced hardware engineers thought they could meet the schedule and declined the assignment. (We were all working on other projects anyway.)

        The job was accepted by a new hire who had successively claimed to be: a  Programmer, a Hardware Engineer, and then a Systems Designer, although he had never designed a coin-op game before.

        He became the Project Engineer which meant he was responsible for the hardware, which included getting it into production.

        The hardware that resulted was a three-board stack:

1. a Main Board with the processor, memory, sound system, and laser disc control interface;

2. the Graphics Board that produced the CGI;

3. the Video Board that demodulated the composite video from the laser disc player, recovered the video sync signals, and produced the game and bit clocks synchronized to the laser disc video.

        Each board was designed by a different engineer. The Project Engineer designed the Main Board and was responsible for coordinating the design of the other two boards so everything would work together. Tom Patten designed the Video Board and Ross Cox (I think) designed the Graphics Board. The project also had two technicians: Minh Nguyen and Dave Wiebenson.

        For some reason, the game was also selected to use the new Switching Power Supply being developed, as well as an in-house raster monitor under development.

        The game was to be based on the then-recent movie Firefox starring Clint Eastwood.

        Even though it has been over seventeen years since the Firefox game was released, it remains a highly emotional subject to those of us who worked on it. The original Project Engineer was not able to get the hardware to work, leading to a very embarrassing incident at the AMOA trade show in the Fall of 1983. (This was before I was conned into becoming Project Engineer)

        The game was scheduled to make its debut at AMOA that year, but they couldn't get the hardware to work as a standalone system. The game was ready to show but it only worked on the Development System. The Project Engineer (not me) said he only needed a few more hours on it, so they arranged with Nolan Bushnell to have his private jet fueled and ready to take off at San Jose Airport. They had already sent the cabinet (a specially designed sitdown with a Ricaro seat) to the Show so all they needed was a working boardset.

        A few hours went by and the hardware wasn't ready. By then it was approaching 10pm, at which time San Jose Airport would be closed to jet traffic out of noise considerations. The Project Engineer said he needed only a few more hours, so they had Nolan Bushnell's jet fly to San Francisco Airport which is open to jet traffic 24 hours.

        The entire night went by and still no hardware. The first day of AMOA opened without Firefox.

        The Project Engineer promised it would be ready by that afternoon, so they had Nolan Bushnell's jet fly back to San Jose Airport.

        Guess what?

        The afternoon and evening came and went and still no hardware.

        But since the Project Engineer absolutely promised it would be ready, they had Nolan Bushnell's jet fly back to San Francisco Airport.

        Again, the hardware was a no-go. The second day of AMOA opened without Firefox.

        At this point things become a little hazy.

        Since Mike, Greg, and Norm were my teammates on Star Wars, I stayed with them during the long nights in case there was something I could do to help.

        After a few days and nights I decided there was nothing more I could do to help so I went home.

        When I came in later that day, I was told that in order to make the last day of the Show they had loaded the hardware and the Development System into Nolan Bushnell's jet and taken off because the Project Engineer promised he could get the game running on the Development System. They got to the Show, set up the Development System, and even that didn't work.

        There was no Firefox at AMOA other than a really spiffy, but dead, cabinet.

        The Firefox Project Engineer couldn't take the pressure, which was considerable, and walked away from the project. When the company asked me to take over, I politely declined since it had been only a few months since Star Wars had gone into production. The company became desperate and finally resorted to outright lies in order to get me to do it. Later, when I found out, I was furious and I never forgave them or trusted them again, even though I worked there for another nine years.

        It was during the Firefox fiasco that Don Osborne, the VP of Marketing, died of a heart attack. He was way too young to die. We put a tribute to him in the game.

        I have been careful not to mention the name of the original Firefox engineer and his name does not appear in the game. Every trace of his very existence has been expunged. We were really pissed at him.  

        When we did Star Wars, the team members' names appear on the Death Star on every other wave. The dots making up the letters are not connected, so they are difficult to read unless you put the game in freeze frame mode and turn up the monitor brightness. Nonetheless, we took quite a bit of heat over it because Atari's policy was that the names of the team members were supposed to be a total secret lest we be hired away by our competitors. Of course, everyone in the coin-op industry knew everyone anyway. Officially, Lyle Rains (VP of Engineering) designed every game.

        Once we did it, the other teams all wanted to do it, too, and the policy was eventually changed. Firefox came out before the policy was officially changed.

        In Firefox, the names appear in the High Score Table; you read them vertically. The way you get the High Score Table to appear is, while in Attract Mode, turn the control yoke sharply clockwise, then counter-clockwise, and hold it in that position. Moving the control handles forward and backward causes the High Score Table to scroll up and down.

        After I was conned into becoming Project Engineer I did a quick review of the hardware system and determined:

1. Some of the problems were due to poor coordination of the design of the three boards. In other words, the signals produced by Board A weren't exactly the signals that Board B was expecting.

2. The Main Board was so poorly designed that it was beyond hope.

        There wasn't time to do a wirewrap prototype. There wasn't time to get a prototype PC Board. The PCB Department didn't even have time to draft the schematics. They worked directly from my hand-drawn inputs to produce Rev A (Production) film. Then Atari ordered the production boards.

        I will assume this reflected their confidence in my abilities as a hardware engineer. (The alternatives are either that they were desperate, stupid, or looking for a Fall Guy.) Imagine going from schematic to production without having the opportunity to test the board.

        This was also hard on the PCB Department. They only had one shot at laying out a production board from scratch. I believe that by then the PCB Department was under the direction of Art Jackson and the board was done by Leon Fritts.

        We managed to get the PC Board vendor to produce a few boards before starting the production run. I received a board and had to give an answer by the next morning. There wasn't time to stuff and test it, so I was limited to a visual inspection. I discovered that several pins of an interface connection had been inadvertently grounded in the layout process and was able to have it corrected before they started the production run. All in all, Leon did a great job.

        The first board available for testing was a production board. Amazingly, my Main Board worked. Not only that, it worked with the Video and Graphics boards. There were a few things that weren't exactly right. We ended up with about 6 cuts and 6 jumpers, which is fewer than most production boards which have had the opportunity of having had several prototypes.

I never want to go through that again.

        The final board-stack hardware totally worked. What sank the game was the new Switching Power Supply, the new Raster Monitor, and the Laser Disc Player.

1. The Switcher

    I was around during its development.

    A typical switcher test consisted of:
            1. Unnamed Engineer: "Everybody ready?"
            2. Sound of bench power switch being turned on.
            3. "BANG"
            4. Unnamed Engineer: "o SHIT!"

        Eventually the Switcher project was moved to another lab where it was mostly by itself (and we could close the door).

        Afterwards, a Switcher test was evidenced just by:
            1. "BANG"
            2. Unnamed Engineer: "o SHIT!"

        After I got sucked into Firefox I changed the power distribution so it could use either the Switcher or the Linear Supply with two ARIIs.

        All of the prototypes were built with Switchers except the one I had them build with the Linear Supply. That's the one I took home.

        We might have built some production games with Switchers; I think that most were built with Linear Supplies.

        A year or so later, Tom Patten did a post mortem on the switcher and found at least one fatal design error.

        The standard method for designing a switcher to work at either 120VAC or 220VAC is to have the rectifier circuit operate as a voltage doubler at 120VAC input, or straight half-wave at 220VAC. This way the DC input to the switching circuit is the same, about 340VDC. The selection is made with a single jumper.

        It was necessary to have it handle 220VAC-240VAC to sell games internationally. (In the Linear Supply the transformer has two primary windings with multiple taps.)

        In order to reduce costs even more, the Unnamed Engineer worked out a method of providing 120VAC for the game fan and the fluorescent light even with a 220VAC input. The idea was to save the costs of having to stock 220VAC fans and fluorescent lights.

        What Tom discovered was that the fan circuit had a tendency to make the switcher operate as a doubler even when configured for 220VAC.

        The Switcher also had audio amps. When it was first tested in a Firefox they motorboated. The problem turned out to be because the Unnamed Engineer filtered the voltages used to power the audio amps the same way as he filtered the +5V supply. (The advantage of a Switching Power supply is that by using a high switching frequency, you can get away with a much lower value filter capacitor.)

        However, if you are going to power an audio amplifier, The Supply Must Be Filtered To Be Effective At AUDIO Frequencies. In this case it meant replacing some 47 uF capacitors with 470 uF capacitors.

        When I discovered the reason for the problem I had been working 12 hours a day for three straight weeks.

        The Unnamed Engineer, who had not worked 12 hours a days for three straight weeks, railed bitterly at having to stay late this one (and only) time because he was missing a sporting event on television. (I think it was a football game.)

        And that is why he has earned the title "Unnamed Engineer".

        The worst punishment is to be forgotten.

        To put the icing on the cake, when the Unnamed Engineer was hired, he was hired-in at a salary that was higher than I was making at the time.
 

2. The Raster Monitor.

        The new in-house designed Raster Monitor had an annoying tendency to blow up when it did not receive sync from the game.
 

3. The Laser Disc Player

        Although we used Philips' Industrial Strength players, it was not up to Arcade use. After a few months the platter motor (a brush-type motor) would give up. Tom Patten (who designed the Video Board) did the post mortem on the players, discovered the problem, and came up with a brushless dc motor for a replacement but it was too late. The operators, who paid almost $5K for the game at a time when most other games cost under $3K, refused to have anything more to do with LD games. The next time they were willing to buy a LD game was when Mad Dog McCree came out, years later.

        That is why the LD version of Road Runner was switched to straight video. The LD version would do instant replays when you screwed up, using actual cartoon footage. It was really fun. The hardware was by Pat McCarthy; game design and software was by the legendary, now almost mythical, Ed (Asteroids, Centipede) Logg. Greg Rivera and Norm Avellar did the conversion to straight video, sans instant replay. They were really under the gun to get the game out, so one day, to relieve the stress a little, I waited until they were at lunch, removed the Road Runner Laser Disc from the game they were working from, and put in a disc containing pictures of naked ladies. When they came back and saw the 'improvement' they had a good laugh.

        By the way, the Philips player was fast enough to move up to several hundred tracks during VSYNC so that the actual footage has several threads interleaved. Playing the disc linearly may be confusing. (It uses the standard CAV format and can be played on a standard laser disc player.)

        About the laser disc control panel. We did not save money by having a player without its own control panel. It was done so that if someone broke into the game to steal the LD player they would get a useless player. Presumably, the word would be spread. In any event, LD player theft was not a problem.  

        When the game was initially released it was too hard. Marketing insisted on it in order to keep game times down. A few months later, in an attempt to "salvage" the game they released an update that made it easier to play.
 

        I met Clint Eastwood one time when he was there about something having to do with Firefox. This was before I joined the Firefox team and I was working on a project called "The Powered Moving Seat" which was a seat that could rotate around the pitch and roll axes, controlled by some healthy motors. I designed the motor controller, which controlled the two motors and had a serial port. The prototype also had an A/D for testing.

        For testing, we had it hooked up to a Star Wars game at the control yoke. Even without program cuing from the game it was still interesting.

        Mr. Eastwood and his two (then teenaged) sons were being shown around the labs and ....  well, I let them try it out. Remember, this was a prototype of a powerful, potentially dangerous, mechanical system designed by electronics geeks. Fortunately for all involved, no one got hurt and a good time was had by all.  

        The Firefox Sound section uses a Quad Pokey. The Quad Pokey was a hybrid circuit where the chips are bonded directly onto a substrate, usually ceramic, containing metal traces much like a PC Board. The traces are made by depositing metal, usually aluminum, using techniques similar to those used to make integrated circuits. This makes it possible to have narrower traces and spaces than are possible with standard PC Board etching methods. Then the chips are globbed. Nowadays,  it is called a MultiChip Module.

        I seem to remember there was some kind of problem with the original Quad Pokey that caused the Pokey on the end to prematurely age and pass away. I believe they fixed it by rerouting some traces.

        The Pokey was a full custom IC designed for the Atari 800/400 to reads pots and keys, which gave it its name POts and KEYs. There was some room left over so they put in some crude square wave sound generators as well as a UART.

        Missile Command was the first coin-op game to use the Pokey. Up until then, all of the sounds in the games were done in hardware which meant that the engineer developed the sounds. The Missile Command engineer didn't want to develop the sounds so he put in a Pokey. Since the Pokey is programmed by software it now became the Programmer's job to develop the sounds. Unfortunately, the Programmers didn't want to do the sounds, either. They wanted to concentrate on the game play, so they did the sounds last. By this time management would be screaming for the game so the programmers generally adapted sound that had already been used in other games. That is why most of the games ended up sounding the same.

        (Actually, the programmers didn't do sounds last. Self-Test was last.)

        Because Pokeys were bought in such enormous quantities by the Consumer Division they were really cheap so we were told to use them instead of the more costly hardware sounds. I had just done hardware sounds for Battlezone and was told to take them out and put in a Pokey. Because of all the work I had on the hardware sounds I was really pissed so I agreed to put in a Pokey if I could keep the hardware motor sound. Otherwise, they could do it themselves. They preferred that I do it, so I got to keep my motor.

        Star Wars originally had one Pokey and a TI speech synthesizer which were run by the Main Processor. But then Greg and Norm used up all available memory (64KBytes) and had no room for sounds. At this point in the project it was easier to design a new board and hang it on the existing board than it would have been to redesign the Main Board to use memory bank switching.

        That is how I got to be the first to use a separate processor in an in-house developed game. Because the board had to be at least a certain size I would have had a lot of space leftover. So I put in four Pokeys.

        Also, for the first time we hired someone to just do sounds (Earl Vickers). After we had one, everyone else wanted one, too. Eventually there were enough Sound Guys that they were put in their own group which, amazingly, was called the Sound Group.

        After I used a separate sound processor with four Pokeys, everyone wanted one, but they did not want a separate board, so work was started on the Quad Pokey: Four times the Pokey Without Four Times the Board Space. Star Wars was in production for several months. At one point I suggested we modify the Sound Board so it could accept a Quad Pokey or four Pokeys. I was told by Engineering Management, "No. We don't want people to know that a Quad Pokey is four Pokeys."

        The last game to use Pokey was the coin-op version of Tetris. By that time Tramiel wouldn't (or couldn't) sell us any Pokeys. Things got so desperate that, in order to finish the run, we were paying a bounty of $50 per Pokey.  

        Atari's last Laser Disc project  was called Playland which used a Laser Disc to give the player a first person view of an amusement park. The LD would also contain control codes for the seat.

        It used a Powered Moving Seat for which I designed a nice little (and cheap) motor controller.

        The game got as far as a Laser Disc containing a combination of computer generated graphics along with sequences video'd from a rather large model of a roller coaster that was built.

        By that time we had worked out the mechanics for the seat and we had a good motor controller (mine).

        Unfortunately, that's when the company went supernova. It contracted suddenly and then exploded, sending out chunks of matter and people, both large and small, into the Great Void.

        I have been asked about the cute little PCB artwork on one of the Firefox boards and the Star Wars sound board. The design on the artwork is the signature of the PC Board designer, Denny Simard. Denny was experienced in doing analog boards and was a member of our group, as opposed to being in the PCB Group. He was a nice guy and also looked *exactly* like Frank Zappa.

        Unfortunately, he only got to do a few boards before being caught in one of the first Great Layoffs. [Which became "Reduction-In-Force" and then "Downsizing"].

        I have also been asked why the Firefox discs are labeled "Firefox V" and were there four previous revisions that didn't make the cut? The answer is that I wasn't involved with the production of the disc, so I don't know, (The disc on my preproduction prototype is also a Firefox V.)

        We actually had a well-equipped video production facility at the time, located in the administration/marketing building.

        It is all long-gone.

        And so am I.
 

Jed Margolin
San Jose, CA
July 20, 2001

Doug's experience in repairing Firefox games can be found at:
www.dragons-lair-project.com/tech/docs/firefox