Today we have a second installment in our interview series by Nate Lockhart, host of The Memory Machine podcast and inaugural member of the Gaming Alexandria Patreon. Last time we brought you his interview with General Instrument luminary Gilbert Duncan Harrower and today as a follow up to that we have another in the series. We hope you enjoy!
This interview was initially published in April 2020 on The Geekiverse website, and is being republished here as it originally appeared.
In the early 70’s, the integrated circuit was still a very new technology, and its potential had yet to be fully tapped. One of the forerunners in IC development was General Instrument, and a key figure in GI was the engineering manager of the consumer products division, Steve Maine.
Early last year, I conducted an interview with Gilbert Duncan Harrower, who invented the Pong-on-a-chip that was a pervasive force in video games in the mid to late ’70s. What I learned through that interview was that General Instrument (the company for which Harrower worked) was a key player in those early days of home video games, being not only the home of the AY-3-8500 Pong chip, but also several other games on single chips, and the Intellivision game console.
Stephen Maine was another of General Instrument’s top engineers in those days, putting the popular tank combat game onto a single chip (later released as Coleco’s Combat!), and creating the impressive STIC chip, a very forward-thinking graphics processor used in the Mattel Intellivision. Also, as you’ll read, Stephen Maine had a hand in the nascent doings of a very successful computer animation studio!
This interview is a transcription of a phone interview, edited for clarity.
NL – I am very fascinated by how people started with electronics, especially in an era like the ’50s; electronics hobby kits mostly consisted of simple crystal radio sets. How did it happen for you?
SM – Well that’s exactly it. It’s a long complicated story. I was born in London, but I was brought up in Bristol in England, I spent a lot of my holidays back in London with an uncle who built model railways. He was very well known, built some incredible models and things. So early on I got a chance to experience electrical and mechanical engineering. Just simply because of the way the educational system worked in Britain, I ended up in a situation where it appeared my early future was to be an engineer.
So I studied electrical, mechanical and aeronautical engineering for two years, and went on to university and I got a degree in technology which was a unique thing at that time, I went to Birmingham College of Advanced Technology, so it was written on the wall for a variety of reasons, but like you said, I built crystal radios and the like… it wasn’t like I woke up one day and decided to be an electrical engineer. Of course, electronics back when I started was… the transistor hadn’t been around very long, so we were taught using thermionic valves. A very different era. I got involved in microelectronics very early, kind of by accident. So it was almost like destiny.
I also had a habit of taking things apart and putting them back together. I used to drive my parents crazy, I wanted to find what made it tick, you know? I couldn’t take the cat apart, so I did the next best thing! (laughs)
The first computer I ever worked on was while I was in my apprenticeship with Rolls-Royce aeroengines. It was a vacuum tube computer and the memory was acoustic mercury delay lines, there wasn’t any RAM as such.
NL – So how did you come to work with General Instrument? You say you started with Rolls-Royce?
SM – Well, my father worked for Rolls-Royce, on aeroengines. I don’t know quite how, but he was able when I was 15, 16 to… It wasn’t quite an internship, but actually… Worked at RR; it wasn’t called RR then. He got me the job there and I worked summers there. When I required a scholarship with an industrial company, RR provided the necessary backing for the Birmingham College of Advanced Technology. A University focused on Technology.
What happened actually, through the industrial trading periods, and during that academic periods of time, instead of just being a student I was being paid. So that supplemented my grants and so forth. And then I got a scholarship from General Electric company… they became a co-sponsor, and then I went back to Rolls-Royce for a while, so I spent a lot of time with them. I actually worked on the control systems for the experimental engines for the Concorde.
NL – So you really had your hands in all sorts of areas of pioneering electronics?
SM – There were actually no electronic engineers on staff… There were mechanical engineers and physicists, but… It might sound more grand than it is. But the fact is I have a pretty eclectic engineering background. When I was with General Electric, I worked with their telecommunications group for a period of time, so I was exposed to telecommunications, and the time they made all sorts of sub-components for International Computers and Tabulators (ICT) LTD, which was a British computer company… That’s where I first got involved in television, was with the General Electric company. That was back when they were still black and white!
NL – Sinclair had the pocket TV, that was a big deal…
SM – That’s where I started, I started with them, and then I actually went through a string of companies in a very short period of time. I worked for a company called Racal Communications when they were a start-up I worked for their research group, and then from there is where I got my first experience with designing integrated circuits, went from there to work for Smith (something) Aviation Industries as a chief designer for their microelectronics group, and from there to a Hughes Aircraft joint venture with EMI in the UK for microelectronics… it was all through a very short period of time, and I rose up very quickly, fortunately.
When I worked for Hughes, we installed one of the first MOS processors in Britain, we decided that we tried to find markets for integrated circuits, using CMOS processes, we developed a lot of relationships; we worked on some of the very first electronic watches and calculators, a whole bunch of things. Hughes Aircraft decided that they didn’t want to be in that business, so several of us negotiated with Hughes that we could take the technology that we were working on that we knew wasn’t proprietary and form our own business, and through that we got funding from General Instrument. So that’s what brought me to GI and video games. Duncan [Harrower] and I first met at GI headquarters in Hicksville, NY.
NL – So then shortly thereafter, you and Duncan moved to Long Island. A very exciting time to be in that area.
SM – Yes, we moved there in 1976, right around the Bicentennial.
NL – So you say you were interested in television, that must have translated very nicely to your work at GI. The chips that Duncan Harrower pointed out were the AY-3-8710, which was the Tank game…
SM – Yeah, that was the one… That chip came about since I had a meeting with Nolan Bushnell in Silicon Valley the day that Time Warner acquired Atari. We had lunch. But we had spent a couple of hours at their plant in the Sunnyvale area, and they had Tank games there, and I asked how much it cost to build one. I said, “That’s insane! We could put all that junk on a piece of silicon for about five dollars.” And that’s where the idea came from to do the Tank chip (without breaking any trademarks or copyrights). The trouble was there wasn’t any true animation in video games at that time, not what we would think of as animation. So, there’s two things we tried to do with the Tank game was provide some kind of realistic animation, and realistic sound.
The second thing was to make it compatible with any TV set that was in the home in the US. It was quite a challenge; there were proprietary designs with Sony’s Trinitron, there was Sylvania, all sorts of brands. They were all different. Some were vacuum tubes, some were mixtures of vacuum tubes and transistors… very few of those TVs worked the same way twice. NTSC stands for Never The Same Color twice. When color was implemented in the games, we couldn’t get consistent results.
One of the reasons was it didn’t generate a true NTSC raster for the projection of the video, so we decided to put on a piece of silicon all those things required to make video animation. We weren’t going to commit to a piece of silicon out of the box. So we built these enormous prototypes that were four foot long, they were huge, bigger than the electronics in arcades at the time since we had to emulate what we were going to put into silicon. We learned in that process, that we generated lots and lots of visual artifacts, they were just unacceptable. So we dropped the interlace, but we still maintained all the correct timing so we could actually do it in color.
So that was the birth of both animation and forward momentum that got us into programmable video games.
NL – I see that online you can find the original Gimini catalog that has a lot of these TV Game circuits. Towards the end of the document, there’s a programmable game circuit. I think perhaps it was used for the Coleco Telstar Arcade?
SM – Well, unfortunately, not all the catalogs are online. When you look at it, it’s not very clear what exactly those devices do. The problem at that point in time is that there was a huge demand for video games, and there was a huge demand for engineering resources to design them. We were in limited capacity to design, so we were looking into a programmable design of some sort, and we took what was common in all these builds and builds we wanted to make so that we would have a two-chip solution, one with the game material and the other with the common material used to make the console work.
So that was the idea, and that first chip we worked on was called TIC – the Television Interface Chip. It took care of all the TV generation and video; it took care of all the characters or images your game might want. And we realized very quickly that it wasn’t flexible enough. AY-3-8800, AY-3-8890, I can’t remember the code, but we called it the TIC. But we realized very quickly it didn’t do what we wanted, so then we developed the STIC, or the Standard Television Interface Chip. It was the core graphics processor for the Mattel Intellivision.
NL – To go back to the Tank game when you talking about graphics in the early arcades, none of them were microprocessor based. It was all discrete logic. In the case of the original Atari Pong, it was 66 different, discrete, off-the-shelf logic chips. In fact, so off-the-shelf that you can still buy the chips and build the game yourself today! In fact, someone mentioned to me after the Duncan Harrower review that they had successfully built the Ay-3-8500 chip to work in an FPGA. Before we go on to the STIC, did you work on the AY-3-8610 (Superstar)?
SM – No, the guy who did most of those chips was a guy named Don Butler, in fact, he ended up becoming the engineering manager there [at GI]. He, like me, in fact transitioned from GI to become original members of Microchip Technology. Anyway, he’s the guy who did the next generation of Ball & Paddle and most of those other devices. Duncan and I around about ’78 were going to do something called “Intelligent Television,” and our original design was for something much more flexible, specifically for the terms of game-space.
We had no desire for it be a personal entertainment terminal, Mattel sort of drove it, it was never designed for that purpose. You could use as a sub-TRS-80 computer, but it was intended to be a graphics sub-processor that would work with anybody’s microprocessor. The idea is that the STIC would handle all the graphics management: graphics collision, animation, overlap, etc, to offload the main processor. It wasn’t intended to be so proprietary.
NL – It is a very robust chip – it can handle collision detection, mirroring, stretching, hardware sprite priority… no other system had anything like that at the time.
SM – When we decided on the first prototype system, we had the CP1600 processor…
NL – And was there difficulty with that processor? It was a 16-bit processor during a time when such things were rare…
SM – The CP1600’s origin… it was developed for Honeywell control systems. And we packaged so it could work as a consumer version; the Honeywell version was very expensive and consumed much more power. We had all sorts of problems with the device to be honest, in terms of actually getting to where we needed it to be as we were using plastic packages for the first time. It took a while to straighten everything power related out. We chose that device for no other reason except that we had it. We had 8-bit processors too, but they wouldn’t have given us the performance we needed.
If I may segway, there’s a company called Amico that’s working on 21st century version of the Intellivision. If you see the promo for it, you’ll see it’s a sort of deja vu. But it’s looking to be an incredible machine.
NL – It might surprise you to know that there are people who are still working with the original Intellivision, creating new games and tinkering with the hardware, getting it to do new things. Someone right now is working on down-porting Castlevania from the NES!
What were the greatest challenges with the STIC chip?
SM – Honestly, the biggest challenge was keeping senior management interested enough to keep funding us. GI actually weren’t pioneers in that they pioneered new markets, they were just constantly working towards new product development. They either worked towards existing markets or were hired by other companies to come up with something specific. STIC was nearly killed off a couple of times, but finally, Mattel was put on the hook and nothing was changed. There weren’t technical challenges. Obviously [working with] anything technical yields problems that can strain your technical limits. The lack of inexpensive memory was a big hurdle to achieve what we really wanted to do. We started chip development in ’77. Memory, RAM was very, very expensive.
NL – The idea I get from the people I talk to is that the video game market was insatiable at the time, and executives were very pushy… pushy and nervous to get things out to market as quickly as could be done.
SM – Well there was hardly such a thing as video game market. There was a sort of a market generated by the Ball & Paddle games, but that was sort of a disposable commodity. But to go the next step and put a multi LSI system together, that was a big commitment, because that required enormous amounts of time from engineering, and you needed partners, you couldn’t do it on your own.
NL – Could you tell me who developed the AY-3-8914, the sound chip of the Intellivision?
SM – Oh that was a variation of the AY-3-8910. Yes, I can, that was Steve Burstein. That device came about as a last resort. Our focus was on the graphics processing. What we really wanted in the system was our speech chip, which was a forerunner to the SP0256 which could also do music and other types of sound, but it wasn’t far enough along the road to make it possible, so it was like the last contingency, what could we do? So we all had our ideas on how to do it, so we pooled our ideas into a young guy who then ran with it. I don’t have his contact, but I do have his brother-in-law’s contact, he [Eric Berman] was the engineering manager for the STIC.
NL – The AY-3-8910 was used frequently in other games and devices.
SM – Yeah, well we weren’t too smart, we never patented it. Yamaha took the basic concept and used a different implementation technology and they ended up using that in all sorts of devices. The concept was sound, but GI wasn’t quick enough to patent it. They had an imitator who ended up doing a better job at the end of the day. So it was a last-minute thing to make sure the system had sound.
NL – Which is fine, I actually enjoy the sound of the Intellivision quite a lot, certainly better than the Atari 2600.
SM – Have you ever seen a working system with Intellivoice?
NL – Indeed, I have one.
SM – Did you ever play the B-17 Bomber game? That was a great game. You’re very fortunate to have one; I only have the pieces of one and nothing in it works. The person who did the detailed speech is still around, I haven’t spoken to him in about a year, but I do have his contact information. There’s also Ken Naiff who implemented the Tank chip, and Phil McLaughlin who did the speech chip.
NL – Did you and your colleagues have anything to do with the infamous Intellivision Keyboard Component or the Entertainment Computer System? If so, what was your involvement?
GIM supported Mattel on its keyboard component efforts in its conception stage and consulted during its development Both companies had a different perspective on the product’s implementation. GIM also had internal dissension concerning the additional product development required by GIM’s internal engineering priorities. Mattel’s response was to introduce a second microprocessor not manufactured by GIM.
NL – Is there anything that we might not know about in your career that we should know about?
SM – Well, I’ll tell you a funny story. When GI decided that they were not going to invest any more in sophisticated video games, Duncan and I went to GI corporate and persuaded them to fund a small business which would develop the next generation in video graphics processing technology. And we started developing low-cost simulators for the services, we actually had working systems. GI decided under the new management they would divest any of these skunk-work type businesses and they wanted to get their money out. So I was asked to raise some more money or sell the business.
We decided to sell the business to this computer graphics lab, but before we did that I met with Steve Jobs a couple times when he was with NeXT, and he thought our work was a bunch of junk. Blah blah blah. He was not very complimentary, to say the least. Well, anyway we ended up selling to Computer Graphic Labs. Computer Graphics Labs ended up being the core business of Pixar. But the interesting thing is that I only found that out only recently. Most of the graphics patents that Apple has cite Duncan and my patent. So we were right all along!
NL – What was the project called?
SM – The business was called Image Management Systems, but the product we called AGILE, which stood for Animated Graphic Images using List Execution, with sort of an early AI concept. If you look at the patent you see that it uses list processing with a means of matching graphical items. CGL was the commercial arm of NY Institute of Technology. But those guys all went on to CGL, they were all key guys. One of them went on to Microsoft. So after CGL I went back to be one of the founders of Microchip Technology. I had multiple roles; I was VP of Marketing for a while but it was really VP of Business Development so I decided to look for broader horizons. They needed partners. Everybody needs partners.
NL – So what do you think of video games today? Do you still play them? Are you interested in them at all?
SM – Well my grandsons are, I’m not.
NL – Duncan had mentioned that he was surprised that no one has come up with a decent car simulator yet.
SM – No it’s funny actually, very few simulators are user friendly or are done all that well. I’m not sure why it seems like after what we’ve seen in [the last couple of years], it should be pretty easy to put that together. Because you need much more than tactical feedback, you need the visuals which you can do through VR. I saw some back in the late ’80s early ’90s, some amazing simulators, probably for military or aerospace but not the larger marketplace.
NL – I remember some machines in the early 90’s, these behemoths in the mall. It was $2.00 for five minutes.
SM – I’ll tell you what, the video game point of view, once they became pretty much “shoot-em up” games, very few of them were video games that appeared to address uh, role involvement or sports things which were done back in the Intellivision days like you could never get that sort of interaction so that you could not only play a game but also teach yourself the sport, to use a racket [for instance]. It never seemed to go beyond feed-forward information, not much feed-back. The technology is there to do it, it’s gyroscopes, all those things, really inexpensive, touch sensors and pressure sensors are very inexpensive.
I moved to LA seven years ago, and I had a friend who was a VP at Intel, and he was a pretty wealthy guy, and he decided he wanted to invest in a company down near San Diego who created new treatments in sleep apnea. To cut a long story short, I started consulting for them full time but in the process, I actually invented something and got a patent granted this year, so I know a great deal about tactics and motions, which I never did back in the video game days. A lot of these mechanical engineers don’t know software, so my plan actually was to set them up, which I did with a team of engineers, but in the process, there was a challenge that came up, and I thought I could solve it, and I did! So there’s a few miles left in this old horse.
One reason I say tactile, this treatment is a collar, it fits around the mandible on the neck, it’s silicon rubber-based, and it has a micro/miniature pump which was invented in the UK very recently, and what it does it evacuates the air in space, this keratin pressure, and that in fact keeps the airway open while you’re sleeping and you can talk, eat and drink while you’re wearing it. Because the collar is flexible, moving around as you move around in bed, and you nod your head, and you move your body, the pressure is being changed all the time by the human person, not the control system. So the control system to control and keep it steady at a known pressure is quite a challenge.
The collar has four barometric pressure sensors which are very sensitive. And it’s got an accelerometer and a gyroscope. So it monitors the person to maintain the pressure. But when I looked at it and started doing it, I realized “gee, this technology to make a truly interactive video game is there, and nobody seems to be using it!” And it’s so cheap it’s like they’re giving them away! You realize there’s a barometer in your phone? And it’s accurate to a couple of feet in altitude too, you can put it on the table and on the floor and measure the change in barometric pressure, it’s very very sensitive.
NL – And our phones can emulate an Intellivision too!
SM – Oh it can do that standing on its head.
NL – It’s been an absolute pleasure speaking with you.
SM – Well I will be interested to see what you extract from that diatribe!
