Advent of Computing - Episode 62 - What's With The Wedge? Part 1

Episode Date: August 8, 2021

Early home microcomputers have a very distinctive shape to them. From the Apple II and the ZX Spectrum, to the Commodore 64 and the Amiga, wedged were the order of the day. I've always wondered why that shape became so popular, and where it came from. Today we start a deep dive into that question, slowly tracing the origins of the first wedge shaped computer. Selected Sources: http://www.leefelsenstein.com/wp-content/uploads/2013/01/TST_scan_150.pdf The Tom Swift Terminal, or a Convivial Cybernetic Device https://archive.org/details/levy-s-hackers-heroes-computer-revolution "Hackers", by Levy http://www.s100computers.com/Hardware%20Manuals/Processor%20Technology/VDM-1%20Manual.pdf VDM-1 manual

Quick, without checking, tell me what a computer looks like. It's just a box, right? Well, at least some fashion of a box. Desktop computers are, well, they're basically big rectangular boxes that sit on or under a desk. Servers can come in a lot of sizes and dimensions, but they all boil down to a big sheet steel box. The same goes for laptops. At least when you close them, you get a flat box. I think this discovery is important, so let's give it a name.

I'm going to call it the Digital Box Hypothesis, and it states that all computers reside inside boxes with 90-degree angles. I think you'll find that my theory holds true in the historical context as well. ENIAC, well, that's just a room full of really big boxes. IBM might have made fancy front plates for their mainframes, but underneath, what is it? It's just a big metal box. Even the original Macintosh, with its fancy built-in screen screen adheres pretty closely to my hypothesis. Sure, some corners are rounded, but they still meet at solid 90 degree angles. It's a tidy little theory, and it's one that would definitely make my life a lot easier if it was true. But, dear listener,

the history of computers is never as tidy as I want it to be. The digital box hypothesis has a big glaring hole in it, and it all comes down to wedges. When microcomputers burst into the home in the 1970s and early 80s, some came in boxes, but others came housed in molded plastic wedges. The computer itself was a printed circuit board crammed into the case, with a keyboard mounted on the top at a slight angle. We have the Apple II, Commodore 64, TRS-80, and many more computers that all came in this one very specific form factor. To me, the big question has always been, why?

Welcome back to Advent of Computing. I'm your host, Sean Haas, and this is Episode 63, What's with the Wedge, Part 1. Today, I'm taking up a bit of an ambitious project, although it may be hard to see why it's ambitious at first. Early home microcomputers predominantly came in the shape of, well, a wedge. You'd get a box with an inclined top, an integrated keyboard, and plugs to hook into a TV or maybe a monitor around the back. There were, of course, exceptions, there always are, but in general, the latter part of the 70s and the early part of the 80s saw plastic wedges dropped onto desks across the world. And it wasn't just one manufacturer doing this,

it was widespread in the industry. Apple, Texas Instruments, Commodore, Atari, RadioShack, Acorn, Sinclair, the list goes on. Almost anyone in the game at the time was making these wedges. Now, this form factor has always struck me as an odd choice, partly because I grew up with PCs, so I'm just used to computers following my ironclad digital box hypothesis. But the manufacturing logistics just seem annoying for wedge computers. Every system had to have a custom molded case. Parts had to be built to fit in smaller areas. The form factor dictates a lot of choices for the computer's overall design. So I've always wondered why this shape was everywhere for such a long period of

time. Well, it turns out that answering that question digs up some really deep roots. This episode, we're going to look at the first home computer to adhere to this wedge design. It also turns out to be one of the first home computers in general. It's a story that's tied up with the history of the Altair 8800, but also popular electronics, the Electronics Magazine, cheap computer terminals, virtualized communities, and hand-soldered modems. There's a lot of stuff to this story, a lot more than I expected, and I think it gives me a really good opportunity to address some of the early days of home computing, at least one aspect of it that I haven't yet covered on the show. So to make sure this all fits, the topic's going to be

broken up into two episodes. Let's just say that August is home computing month here on Advent of Computing. So without further preamble, let's jump in. Today, we're starting our slow approach to the Sol20 terminal computer. And on the way, we're going to hit a whole lot of background and try to understand not just how the first wedge computer was built, but why it would take anything like that form factor to begin with. It's hard to overstate how important magazines were to the home computing revolution. Websites have kind of taken their place in the modern era,

but not really in the same way or capacity. One of the core magazines in this early period of home computing was Popular Electronics, but it's not the only one in play. Popular electronics catered specifically to, for lack of better terms, nerds with soldering irons and some kind of interest in the field. Crucially, it was very much targeted at hobbyists. Professionals had their own industry mags or conference proceedings. Tinkerers and hobbyists who were looking for something new, exciting, or just plain fun had popular electronics. The real core offering of this magazine, as well as other electronics hobbyist magazines in the same niche, came down to projects. These types of articles were a mix of prose describing the project, detailed instructions, part listings, sometimes schematics, and on occasion, even masks for etching your own circuit boards printed right there in the magazine.

If the full-on masks weren't printed on the page itself, then you'd be provided with details on how to order those. Each new issue was an easy way to dive into a new adventure. Now, Popular Electronics wasn't the only show in town, and in the early 70s, the magazine faced a series of existential threats. In 1972, a merger and restructuring inside the parent company shook things up quite a bit. Changes can be a good thing, but it also tends to ruffle feathers. During this year, some of Popular Electronics' best contributors got a little bit ruffled, and they decided to go over to other magazines.

One of their competitors that gained from the defection was Radio Electronics. All these magazines have really similar names. Many started to feel that Popular Electronics was falling behind in quality. And I guess this is where we need to take a second to talk about how this kind of publishing works. This will help make the story make sense, so just stick with me here. I swear this is still a computer podcast. Most magazines don't have a whole lot of staff writers. If you find yourself reading an article in Popular Electronics, then chances were it was pinned by someone outside the company on a contract basis. These outside contractors are either contacted directly by the magazine, or more often, they

send in pitches for articles. The upside here is it's a lot cheaper, and you have a lot more flexibility. Authors aren't full-time employees, so there's just less overhead in general. The company just has to have editors on staff. But the flexibility component is kind of a double-edged sword. It lets magazines pull from a wider array of sources, which adds good variety, but it also means that authors aren't necessarily beholden to any one publication. As we see with popular electronics, authors can just jump ship, or they can get poached by another magazine that might pay them better.

Or, conversely, if a certain author is in hot demand, then they can end up writing for competing magazines. Publishers end up in this kind of strange position where they have to keep good relations with authors in order to publish good content otherwise they'll run into issues now business models aside let's get back to the matter at hand in september 1973 radio electronics printed a huge hit that year they published plans for the so-called TV Typewriter. It was designed and written by Don Lancaster. From my research, I'm fairly certain that Lancaster wasn't one of

popular electronics defectors, but this article did add to the pile of blows against the older magazine. More broadly speaking, it helped push the hobby electronics world closer to the verge of computing. The TV typewriter is, all things considered, a bit of a silly device. It does exactly what the name suggests. It's a small box with a keyboard that you plug into a television, and as you type, letters appear on the screen. It's not that useful. The Radio Electronics article provided all the details needed to build one of these machines, from parts to simple schematics. Lancaster also provided a full explanation of how the device worked, and how to expand it. The TV typewriter had no processor. It didn't even have real memory, but it did expose a data bus.

So in theory, one of these little devices could be hooked up to another data source besides a keyboard. Say, a computer or maybe a modem. In theory, you could build your own terminal that worked over a normal TV and could connect to even a mainframe. The main point here is that the TV typewriter wasn't a computer, but we can see these hobby project articles approaching computing. We can also catch a glimpse of the new wedge form factor. Lancaster's design didn't explicitly describe how to build a case for the typewriter. That was left as an exercise to the reader, so to speak. But the magazine's cover showed an illustration of a rectangular box

with a keyboard mounted on its top face. We don't have the inclined keyboard, and we don't have a microprocessor, but we're starting to get there. Lancaster's neat little device, and more specifically the article, were a big hit amongst the hobbyist crowd. For $2, you could mail out for the full PCB masks and complete instructions to make your own board. Add in $120 of parts, and you're up and running with a typewriter. With the expansion capability built in, the TV typewriter was a fascinating new platform to build off of. It sparked the imagination, and it provided a really good return for radio electronics.

Spurred on by this project and others, hobbyists were getting hungry for computer-related work. Popular electronics had to do something to stay on the leading edge. They just couldn't afford to keep slipping up. Of course, a contributing factor to all this was the development of the microprocessor. Intel released the 4004 in 1971, the 8008 in 1972, and then they really hit their first practical chip, the 8080, in April of 1974. So microprocessors were just starting to become available as popular electronics was looking for a way to one-up the TV typewriter. Near the end of 1974, one of the MAG's editors, Les Solomon, found the hit that he was trying to find.

He had heard of a little machine called the Altair 8800 that was being developed by a company called MITS down in New Mexico. Solomon already had a working relationship with the crew at MITS. A handful of employees had written for Popular Electronics in the past, and MITS had even sold kits for a couple of their products through the magazine. These were mainly hand calculators. When Solomon heard that they were planning an 8080-based computer, he jumped at the opportunity. It would be just the cover story that the magazine was after. The article ran in the January 1975 issue of Popular Electronics, and it was an unqualified success. I know that I use the Altair 8800 as a mile marker a lot, but that's with good reason. This was the first time that a consumer could order a fully

assembled computer. All you had to do was send a check down to Albuquerque, New Mexico, then wait to join the computer revolution as soon as shipping was complete. Just the impact of the article alone shaped the modern computing scene in innumerable ways. As one fun example, Microsoft was started after Bill Gates saw the cover story. He and Paul Allen realized that, hey, maybe home computers were about to be really big, and the rest is history. But while the Altair 8800 was, by the strictest definition, a microcomputer, it still wasn't that far removed from a fun weekend project.

It wasn't quite practical, at least initially. I've ragged on the cover story about the machine before, and definitely there's a lot of embellishment going on in the article. The Altair shipped with a whopping 256 bytes of built-in RAM. That's right, bytes. Not megabytes, not kilobytes, but just a few hundred bytes. It also had really limited input-output capabilities. The assembled units came as a sheet steel box with an array of lights and switches on its front plate. To use a stock Altair, you had to master flipping switches and reading binary numbers off lines of LEDs. But it was a start. To get any

serious use out of an Altair, you had to upgrade it, and luckily, MITS had a solution for sale, at least. The computer used a backplane design. This was a central 100-pin bus that printed circuit boards could slot into. This 100- pin bus was, well, as simple as possible. Each pin was just wired up in parallel, so any expansion card that you plugged in was effectively wired directly into every other card. Even the microprocessor was housed on one of these expansion cards, so the backplane was literally just a circuit board with vertical lines on it. Like the TV typewriter before it, this also made the Altair a wonderful platform for tinkering. It also opened up a whole new market niche just for these

expansion cards for the computer. So what did an actually useful build of an Altair look like? What if you wanted to use it like a serious computer with a screen and keyboard and actual software? Well, that was possible, but it was not a cheap proposition. The general loadout for a usable Altair included a RAM expansion card, a serial input-output card, and, the most expensive component, a whole separate terminal. The big expenditure here was, of course, buying a whole separate terminal. A tricked-out Altair was not the cheapest thing in the world, maybe running around $800 or so, maybe a little more if you wanted more RAM. A cheap computer terminal could cost easily over $1,000. And this is all in 1975 money.

At the time, hobbyist terminals just weren't a thing. There weren't hobbyists that needed computer terminals until the Altair came around. The upside was that this introduced another opportunity for popular electronics. Soon after Altair started shipping, Solomon went on the hunt for a terminal to pair with the computer. It's easy to see how that would make a fantastic follow-up to the cover story, especially if he could beat competitors to the punch. This next part is chronicled pretty well in the book Hackers, Heroes of the Computer Revolution, written by Levy. But the gist of the situation is that Solomon wasn't going to find an

easy win. His first attempt was to try and construct a dream team to build a new dream computer. In his mind, that would turn out to be either a cheap terminal that would go with an Altair, or some all-in-one computer with a video display and a keyboard. One side of the partnership, at least in Solomon's mind, had to be Ed Roberts, the founder of MITS and designer of the 8800. From Levy, quote, Solomon went to Phoenix to visit Don Lancaster, inventor of the TV typewriter, and convinced him to drive down to Albuquerque to meet Ed Roberts. Maybe the two giants might combine on a terminal project. As Solomon later

described it, the meeting was, quote, bang clash, a clash of egos. Don refused to change his design to match Ed's computer because he said Ed's was inefficient. Ed said, no way, I can't redesign it. They immediately decided to kill each other on the spot and I separated them, end quote. So, you know, maybe not the best outcome. Solomon had to keep moving forward. He had to find something to put on the cover. As an editor for an electronics magazine, he was never short on contacts, so he just had to move down the list. He had recently heard from a small outfit called Processor Technology out in Berkeley, California.

They had been selling RAM expansion cards for the Altair and, more recently, were flaunting a new video output board. Maybe with a little push, a few suggestions, Processor Technology could build the machine that Solomon was after. So who is this shady Processor Technology? The exact roster shifted around a bit, but we have two central players that matter for today's episode. Bob Marsh and Lee Felsenstein. They worked out of a rented garage in Berkeley, California, and they wanted to build computers, or at least computer components. We have here the classic story of the home computer revolution. But there's a bit of a twist here. Marsh and Felsenstein both fit the image of early hackers.

Both were cast in the mold of 70s Bay Area counterculture. Both were fascinated by electronics, and both loved to tinker with new things. For the twist, we need to jump back a few years. Before anyone had ever heard about TV typewriters, and before MITS ever built a computer, Felsenstein was involved in a little project called Community Memory. In short, it was a project designed to create a new digital commons. If you want the longer version, then the latest bonus episode up on Patreon covers the project, so I guess this is a plug to go sign up and get some extra shows. Anyway, the how of community memory is really interesting.

A group of computer enthusiasts working with grant money and donated hardware set up publicly accessible terminals around the Bay Area. The first terminal was an old teletype installed outside a record shop in Berkeley, with more soon to follow. Any passerby could plop down in front of a keyboard and access an early bulletin board system. They could read and write posts totally anonymously. The idea being that community memory would be a resource to help foster real-world communities. It was a way to bring people together with the aid of a computer. It would also come to serve as a means for really free expression in a safe space.

The physical access part, that is, the terminals, became a bit of a recurring issue for the project. As I already covered, in 1975 terminals weren't really cheap, so in the even earlier 70s terminals were less cheap. And community memory wasn't really a wealthy project, so to speak. wasn't really a wealthy project, so to speak. A complicating factor was that these terminals weren't going to be used in a lab or some controlled setting. They were meant for public use, by folk with a wide range of experience levels. This was a known problem even as the first terminal was set up. A volunteer from the project had to be present to guide new users in how to use the machine. They'd have to fix paper jams and refill the feed. You just couldn't leave a teletype alone in the wild for long. It would break down. CRT terminals did help, somewhat.

Community Memory got their hands on an early one soon after the project started. You no longer had to deal with paper jams, but in exchange, you got another point of failure. CRT tubes suffered from burn-in. If an image was displayed on the tube for too long, it could leave a permanent mark. Tubes could also break. I don't know how common that was for early terminals, but the danger was at least present. Especially in a public setting. And that's not to mention all the electronic doodads that made the terminal tick. Those could all go bad. A power irregularity could cause capacitors or semiconductors to fry,

or maybe a bad part was installed at the factory. The point is, you get maintenance costs on top of an already expensive device, and publicly installed terminals stood the chance of receiving a lot more rough wear and tear than machines held in a lab or some other institutional setting. Felsenstein was in charge of hardware at community memory, so that meant he was in charge of finding a way around the terminal issues. But there was a big wall to any progress. No terminal on the market matched what Felsenstein needed. He was looking for a machine that was relatively cheap, durable, and easy to repair. It also had to, you know, function as a terminal. It needed to receive signals over a phone line, display data, accept keyboard inputs, and then send signals back out over the same phone

line. Low on options, Felsenstein decided that he kind of had to bite the bullet and start in on his own system. It was a really big undertaking, so he decided to go in stages. The modem would be a big first step. That's the part of the terminal that converts between data and tones on a phone line. At the time, Community Memory was running with these modems that cost around $300. They were donated, but if they wanted to expand to more terminals and more installations, they'd need to eventually buy more modems, and they just didn't have $300 to spare for each installation. Felsenstein set to work making a cost-reduced model, and this is where we can really see the hacker ethic at work. His first step was to cut some corners.

Community memory was only ever going to be running data at 300 baud. That just means that there was a fixed rate at which data was received and transmitted. Contemporary modems could be configured to handle a very wide range of baud rates. But that wasn't necessary. Felsenstein just didn't need that. Scrapping that one feature cleared out a lot of complications. The other trick that we pulled is a little more complicated, so here's the simplified version. It all comes down to how modems interpret incoming data. Traditionally, this is done by holding a reference voltage. To determine if you have a 1 or a 0, all inbound signals have to be compared to that reference value. The circuit that handled that was

complicated and a little tricky to make. Felsenstein happened to know that there is a sneaky way around the issue. He had some experience working with serial communications from his day job. He had been working as an occasional consultant for an electronics firm, so he knew a bit of the inside baseball. A normal data connection on a phone line isn't straight data. There's some packing going on that's used to maintain signal integrity. A one value is pulsed between each bit of actual data on the line. So if you send out 000, then that gets changed and transmitted as 01010. That intermixed 1 also happens to be at the same logic level needed for a reference voltage. Felsenstein decided that, hey, that was good enough for him. So he hijacked the normally

useless part of the signal. It comes at regular intervals, so Felsenstein just had to use this pulse to charge up a capacitor. That functioned as a reference voltage that was tuned to the input signal, and it would be topped off by each new pulse. It would stop working if you didn't receive any data or any pulses, but that would break down on any modem. Felsenstein called this new modem the penny whistle, you know, since it was cheap to produce and it made sound. Anyway, a side benefit to this signal-derived reference voltage was that the penny whistle was resilient to bad connections. An interesting byproduct is that you could use the same modem that you connected to a

phone line to read and write data to and from audio cassettes. It's all just audio. Cassettes tended to drift and flutter a little more than a phone line, but it's still audio. The overall result is a cheap, simple, and surprisingly versatile device. Simplicity was key because it made the Penny Whistle easy to repair if broken in the field. So that was one huge piece of the puzzle sorted. It was also perhaps the easiest part, sadly. Up next was the actual terminal itself, some kind of system that could handle, display, keyboard inputs, and connect to that modem.

This was a much bigger undertaking. Felsenstein knew that he would need some kind of help. The next chunk of the story is a little confusing for me, mainly because I've seen it reported a number of different ways. The most cogent telling I can come up with is from a few interviews with Felsenstein himself, and it goes something like this. Felsenstein knew Bob Marsh from back in college. The two reconnected sometime around 1973. Felsenstein was posting on

Community Memory about his plans to devise a new terminal, and he was looking for collaborators to help out with the scheme. A meeting to discuss the topic was eventually held, and Marsh was in attendance, where he made two crucial contributions to our story. First, Marsh suggested that Felsenstein look into adapting the TV typewriter for his own uses. Marsh had just constructed one of these devices and was more than happy to show off his handiwork. And second, Marsh asked if Felsenstein would be interested in going halfsies on rent for a garage. That may sound absurd, but at the time, it was a really good offer. Felsenstein and Marsh started renting a garage together and used it as a shared workspace. This would be the start of a very long period of collaboration. Settled in the garage,

Felsenstein got the chance to tinker with Marsh's TV typewriter. Now, ideally, theoretically, this would solve the entire terminal issue. Lancaster's device could be wired up to receive data from a modem, and you could probably make it send serial data from the keyboard to the modem. Its video signal worked on a standard TV, you know, the kind you can pick up at a store or find at a second-hand shop. And best of all, the TV typewriter was cheap and relatively simple to construct. It was at least more simple and a lot cheaper than any other option. But Felsenstein ran into a dirty little secret tucked away inside the typewriter. To quote,

little secret tucked away inside the typewriter. To quote, the TV typewriter was not entirely usable as a terminal because it was a paged display system. When you got to the end of one page and the last character went up on the screen, you had only 1 30th of a second before the screen blanked and you saw the next character. So you were actually likely to be outrun by it. End quote. It all came down to memory, or rather, lack of memory. The TV typewriter, or any digital device that outputs an image, needs somewhere to store data used to construct that image. A television won't hold an image you send over to it. You need to refresh

it at a set frequency. For the TV typewriter, you only needed to store a screen's worth of text characters. It's not a whole lot of data. But when Lancaster was designing the machine, he ran into the issue of cost. At the time, random access memory chips were still relatively expensive, cost. At the time, random access memory chips were still relatively expensive, so he went for the cheaper option and used shift registers. These chips function as sequential memory. You can put in some value, send a pulse, and it gets stored in a long list of numbers. On the other side, you can tell the chip to pop out a number from the end of the list. In 1973, a handful of shift registers cost less than even a single RAM chip. So Lancaster designed the device around this sequential memory storage. The simple fact is it works, but only in a very limited capacity.

There's no scrolling, no tricks, just text displayed on a TV screen. With that realization, Felsenstein was left to cut a totally new path. The way he saw it, the pieces were there. Everything just needed to be connected up somehow. You take the penny whistle modem, the TV output and keyboard input of Lancaster's typewriter, and a little memory and logic to hold everything together,

and you should have the cheap and robust terminal of everyone's dreams. Felsenstein didn't really have the funds on hand to make that happen, but he could do something almost as good. In late 1974, he laid down plans, eventually compiling them into a manifesto called the Tom Swift Terminal, or a convivial cybernetic device. Now, I'm not calling this a manifesto just to sound cool. I think that's the most accurate term for this document. Just to get a taste for the prose, here's a short passage. We believe that even cybernetic electronics, held by some to be the epitome of mystifying technology,

will yield to the application of convivial design. We have heard the industry jokes about the maintenance specialists who are shipped with large computer systems and know that not only are such jokes very nearly true, but that such subordination of man to machine signifies a potentially disastrous tendency of technological development. In other words, we're dealing with a manifesto of personal computing, even if Felsenstein isn't explicitly calling his terminal a personal computer. On the surface, this may seem like a lot of feature creep for a terminal project, but I think it's totally in line with Felsenstein's overall goals.

He wasn't just trying to make a cheap and repairable terminal for fun. It was in support of community memory. The project was all about bringing people together, democratizing data in an ethical way. Here we can see Felsenstein looking for an ethical way to democratize computing. So sure, on the other side you get a cheap terminal, but wrapped up into that you get an accessible computing platform. So where does this whole computing part come in? So far I've only talked about the terminal side of the equation.

For the full picture, we need to get into the Tom Swift terminal's actual design specification. The core of the terminal was a 40-pin bus, with each pin wired in parallel. Actual components, that's the smarts of the terminal, were plugged into the shared bus. Sound maybe a little bit familiar? This is before the Altair 8800 graced the January 1975 edition of Popular Electronics. We have, independently of the Altair, a personal computer being designed to use a very similar modular approach. Now, when I first read this, I was a little bit surprised. It seems like a strange case of parallel development. After doing some more digging and connecting a few dots, I've changed my mind a little. This isn't just some surprising technical coincidence. I think it's

a window into the earliest days of personal computing. I want to take a beat to look at this modular-ness in a few ways. The closest connection is, perhaps no surprise, the TV typewriter. You see, I told you at the start this story is a bit of a tangled deep web. Lancaster's system also uses a parallel bus of pins to connect multiple cards together. It's a little harder to see than in the Altair case, but the TV typewriter does follow this formula. The device initially is built up from three circuit boards, each with a set of pins on their bottom and sockets on their top. They stack in this piggyback fashion, pins held in sockets, forming a communications bus. The key here is that every board has this pin and socket bus.

Initially, this was just three core boards, but expansion cards could be constructed that plugged into that same bus. So, we have here a direct precursor to the microcomputer, with a bus structure very similar to later arts. On the small side of things, a main bus just makes practical sense. Expandability was a big plus, but it also added to the hack value of the system. If you have exposed pins and empty slots, then some nerd with a soldering iron is going to find a way to use them. I know, I always try to find a way to use any exposed header pins.

Just this simple design choice opened up a world of possibilities. The folk designing these computers, be it Lancaster, Felsenstein, or Roberts, were all nerds with soldering irons, so it should come as no surprise that they would cater to their people. This design also saved money since you could build a system using smaller printed circuit boards. The cost of etching, cutting, or printing circuits scales as a factor of size after all. Over on the larger side of computing, we get into something just as interesting. Felsenstein credits DEC specifically for the idea of a bus-based computer.

This comes up explicitly in his Tom Swift Terminal article. I currently don't have enough expertise to get into the finer details of the DEC PDP-11 bus architecture, but I think this part does bear mentioning. The bus-based design wasn't just popular in the hobbyist space, it was being used in the industry as well. Far from some weird idiosyncrasy of these small machines, the parallel development of a bus in both the Tom Swift design and the Altair 8800 speaks to a trend. Now, going deeper into the Tom Swift paper gets us to the machine's actual guts. Those are the cards that plug into that magic bus. This part is equally as simple.

Felsenstein described a memory card, an I.O. card, and a display card. Notice the distinct lack of a microprocessor here. I'll get back to that. The memory card is nothing more than some RAM chips soldered onto a fiberglass board. The basic configuration in the paper would come with just enough memory to hold two 32 by 16 character screens of text. That would allow for scrolling, thus beating Lancaster to that feature. The input-output card is also really simple. It handles input from a keyboard and a serial connection that hooks into an external modem, or really any external serial

device. Just as a point of order, the keyboard was described as a separate unit, so we aren't into the wedge just yet. Keyboard signals are routed out to the serial interface, and incoming serial data is pushed into memory. Of course, appropriate shifting happens to make scrolling function as expected and to put bits in the right locations. The final piece was the display card, and this was the most complicated component. It handled reading in a chunk of memory, converting it from characters to bitmaps, and then generating an analog TV signal. The card would also handle control characters. Since we're in the realm of digital to analog, this required the most complex circuit

in the Tom Swift terminal. If that wasn't enough, Felsenstein also describes the display card as having multiple video modes. The standard would be all text all the time, but a second mode would interpret memory as a bitmap. This would allow for either easy debugging or, on the cooler side, primitive bitmap graphics. That's the base system, but remember that the Tom Swift is an expandable terminal. Felsenstein included a list of possibilities in his manifesto, the most pedestrian being memory expansion cards. He also describes a microprocessor expansion card,

that is, a card that would take the Tom Swift terminal and make it into the Tom Swift computer. But while that's an exciting option to see, a processor was never central to Felsenstein's design. As he would explain in an interview with the Computer History Museum, quote, the Tom Swift terminal was designed not to rely on microprocessors. I came up with the idea that in a computer, everything is peripheral to memory, which may be true. Once again, this comes down to a mix of practicality and functionality. RAM chips had been getting cheaper, but microprocessors were still really expensive and hard to get a hold of. Around the same time, Gary Kildall, the future creator of CPM, was going through all kinds of contortions

just to get his hands on a couple Intel chips. Basing the system around memory also allowed for a surprisingly practical design approach. Data traversing the I-O card was just pushed directly into memory at the proper location. Then data in some specific range of memory was pulled into the display card to be rendered on screen. This is more succinctly called memory mapped I.O. It's a feature that we still rely on today. It allows for on-the-fly expandability without changing the existing design. And with the addition of a microprocessor and a little software, any new expansion could very easily be supported. The bottom line is Felsenstein had designed a darn good terminal, and really a darn good computer. But that wouldn't be enough to make Tom Swift a reality.

Lee ended up selling copies of his manifesto to friends and some interested parties. There was some buzz and discussion of the machine, but that never got very far. It at least didn't get big enough to drum up funding. Plus, there were some extenuating circumstances to be aware of. This catches us back up to January 1975 and the Altair 8800. Felsenstein and Marsh first met the Altair in the garage of a mutual friend named Gordon French. At the time, they were resoundingly unimpressed. French was running a small outfit called the People's Computer Company, primarily a newsletter-based group

centered around computer accessibility. He had convinced Mitts to send him an Altair for review on the newsletter. The unit arrived sometime before March 1975, so the group in French's garage were some of the first people to get their hands on the hot new machine. But still, all parties sounded pretty unimpressed. The ragtag group started passing around the Altair trying to figure out what to do with it. Felsenstein recalled, quote, I took it over to Ephraim's place and asked him what to make of it. Frankly, he considered it useless and in a way he was right, since there was nothing to it but switches and lights. He kept it as a sculpture in his living room, on the same table with his guinea pig cage, with its lights flashing to keep the guinea pigs company, end quote. Now, if the Altair is this big,

shining milestone, then why were early adopters so blasé about the machine? I think by this point in the episode, we're pretty well equipped to answer that question. Felsenstein had just designed a computer that was eerily similar to the Altair. Inside the 8800 was a 100-pin parallel bus with a handful of sockets, a memory card, and a processor card. That was it. The specifics differed, but we can see how the Altair was just a few steps removed from the Tom Swift terminal. The stock Altair 8800s were also, to put it lightly, really, really underpowered. Note the lack of any fancy I.O. options. There's no serial, no keyboard, and definitely no TV or display output. The standard memory card came with a whole lot of sockets, but only had a few RAM chips, just enough for 256

bytes worth of memory. Just to reiterate, that's a ridiculously small amount of storage. It's simply not enough to fit any kind of practical software. Felsenstein also had technical quibbles with the Altair, specifically its bus design. The machine from MITS used a 100-pin wide bus. Felsenstein's Tom Swift design called for only 40 pins. The Altair's bus, eventually called the S100 bus, took a pretty ham-fisted approach to expandability. Pins from the computer's 8080 processor were mapped almost directly onto the bus after going through some supporting chips.

These broke out separate lines for data in and data out signals to and from the processor. To Felsenstein, this looked like wasteful design. You can't read and write data at the same time. That's just how computers work. The Tom Swift documents called for a single set of data pins with some semaphore lines dedicated to telling whether the pins were for data in or data out. In general, that's just a more clean way to do things. The point here is that Felsenstein and his colleagues just felt like the Altair wasn't

anything special, at least not technologically. What was special was the fact that the Altair existed as a consumer product. In all things considered, it was a pretty cheap one. A kit cost just $439. That had all the parts needed to build your own computer. A pre-assembled unit came in at just over $620. Inflation jacks that up a lot, but it's still in the realm of a reasonable product. It's less than a used car.

What's really cool here is the system came with an Intel 8080. On its own, that chip cost over $300. But MITS, as a company, worked out a deal to get them in bulk for $75 each. MITS was making computing accessible even if the computer itself wasn't all that useful yet. This had a side effect. It got a lot more electronics hobbyists interested in computers. The group Felsenstein was involved with started to grow. On March 5th, 1975, a meeting was held to make things a little more official. Hosted again by Gordon French and Fred Moore, this ragtag team called itself the Homebrew Computer Club. The Altair 8800 played a key role

in bringing these people together, but it also drove them deeper into the field. Felsenstein described the prevailing feeling in these early meetings like this, quote, Maybe as a group, we knew as much as these MITS guys, and possibly the Altair wasn't even an this. Quote, to everybody as the first newsletter. I talked about computer memory in the Tom Swift terminal. Wozniak talked about the breakout game he had done and discussed the video terminal he was working on. End quote. So, the Homebrew Computer Club has become a kind of legend in the history of home computing. Wozniak himself is probably the most well-known member today, but a handful of other early computer companies would spawn from these meetings. Exposure to the Altair 8800 was a flashpoint, and the growing community kept the fire stoked. Felsenstein and Marsh would pretty quickly jump into this new niche. Shortly

after the Altair was released, MITS started offering 4K RAM expansion boards. These weren't well designed, and they had manufacturing flaws. It seems like the consensus is that MITS was overextending themselves. Basically, it was a non-starter. Marsh decided that, hey, if he knew as much about computers as these MITS guys, then he could probably make his own RAM boards. So he did. By April of 1975, Marsh started shipping 4K cards out of the garage he shared with Felsenstein. Lee wrote the manuals that accompanied the boards. To facilitate things, Marsh and another friend named Gary Ingram formed Processor Technology. This was the first company to sell Altair-compatible cards. At this point, Felsenstein was on the periphery of processor technology. Marsh and Ingram were

the actual insiders, but that didn't last long. Quoting again from Felsenstein at the History Museum, quote, Early on in processor technology, Bob Marsh said to me, We will give you the opportunity to design the Tom Swift terminal, but you have to do it our way. And that became the video display module, the VDM-1. End quote. Now, our way had a few meanings. For one, this realization of the Tom Swift terminal wouldn't be its own machine. Instead, it would be a card that slotted into the Altair. It would also have to be marketable, since processor technology, as small and niche as it was, did exist as a for-profit company. By this time, waiting around had actually solved the majority

of Felsenstein's design problems. Memory was handled now. There was simple I.O. available for the Altair, it even had a microprocessor. So the last thing was just video. But that's not to say that designing the VDM-1 was a simple task. We're dealing with a circuit that takes digital inputs and creates an analog TV signal. As I said, a good rule of thumb is that crossing the digital analog barrier is gross. Felsenstein readily admitted that he was a little out of his depth. He had worked up a spec for how a video signal should be handled, but that was a far cry from actually building the circuit. That wouldn't stop him for very long, and he readily accepted Marsh's offer. So, what does Felsenstein come up with?

I think before we get into the product, we need to look at exactly what he was working with. That all comes down to the Altair, and the S100 bus, really. Everything had to come across those 100 pins, so that forms the main design constraint. Luckily, it had just about everything a hacker could want. You have a big chunk of pins on the bus that were dedicated to power. This is a really weird part of these old computers. The bus provided plus 8, plus 16, and minus 16 volts. It's gross, and I don't like it. Today we're a lot more used to seeing plus 5 for running microchips

and maybe a plus or minus 12 for some motors. But back in the day, early microprocessors had strange power requirements. The Intel 8080 needed plus 5, minus 5, and plus 12 to operate. Serial, as in real old-school RS-232 standardized serial, required plus and minus 12 volts. The strange bus voltages were there so that each card could regulate things down to whatever they needed. The plus side, at least the one that I have stuck in my head for whatever reason, is that a savvy designer had access to a wide range of power options. To me, that seems just convenient for the analog part of the VDM-1, but that might just be me reading too much into things.

The rest of the S100 bus was fully digital. You got data in, data out, address lines, and some semaphore from the processor card. Once again, we can see how it's kind of ham-fisted. I personally like it. Felsenstein thought it was kind of dumb. Anyway, you basically have all the pins from the 8080 demultiplexed and then splayed out onto a connector. The upside for Felsenstein was that this gave him access to everything in the computer's memory. More importantly, he could read from memory without involving the processor at all. In theory, the processor didn't even need to be connected to get the job done. This was exactly how Felsenstein wanted his Tom Swift terminal to operate, so VDM1 really did

become a realization of that earlier idea. The real heart of the operation for the VDM-1 all came down to memory mapping, but this is accomplished in a pretty clever way. And I kind of fibbed a little bit here. The VDM-1 doesn't actually read from the Altair's other memory cards. Instead, the display card contains its own memory. That memory is exposed to the rest of the computer over the S100 bus. To the alterant any software running on it, it just looks like any other chunk of RAM chips. But any data pushed into that special region gets turned into a TV signal. I told you, the VDM-1 didn't need a processor to function, it just sat on the bus and did its own thing. That makes it sound simple, but the whole turned-into-a-TV-signal part is where the

hard work is hiding. The VDM1's manual provides full details on how to construct the card and its theory of operation. It's over 100 pages long, so yeah, we're dealing with a complicated device. The basic data flow works something like this. On a fixed clock cycle, the VDM1 reads from internal memory to get the next line of characters to render. That's represented in memory as numbers, which simply won't work. Those numbers have to be fed through a character generator. That's a chip that can map numeric values into bitmapped font tiles.

Those tiles were further broken down into scanlines, then converted into analog signals that a TV can understand. The entire process has to be timed perfectly to line up with a TV's refresh rate. The VDM-1 had a number of clock crystals on board just to negotiate this timing conundrum. The bottom line is that we're dealing with a lot of stuff. There's a lot of stuff going on here. More specifically, this couldn't have been done in software. Timing had to be exact, which meant that every fraction of a second, some kind of operation had to be performed. which meant that every fraction of a second, some kind of operation had to be performed.

If you tried to do this all using the Altair's processor, then you wouldn't have any cycles left for anything else. That made the VDM-1 an invaluable tool. It did something that, on its own, an Altair could not. Another interesting quirk that doesn't impact the story, but I just have to mention, is that the VDM-1 had its own onboard ROM. However, this didn't contain any code. Instead, the board's ROM held its font. The character generator needed some kind of source to pull characters from, you know. The interesting byproduct of this is that you could, at least in theory, change the VDM-1's font. When development was completed, Felsenstein had

himself a pretty capable little cart. The VDM-1 output crisp text at 16 by 64 characters per screen, and it could even handle scrolling. It could technically work on any store-bought TV, but you had to make some modifications to the television's tuner. The final board was crammed with chips. It was probably one of the more complex additions for the Altair at the time. At $199 for a kit, processor technology was soon inundated with orders. The board was really well-received because it was doing a job that nothing else could. For $199, plus a TV and some time soldering and a keyboard interface, you could run an Altair as a fully functioning computer. No terminal needed. The VDM-1 offered

the final puzzle piece to break that dependence on terminals. Before the end of 1975, the Altair could function as a standalone unit. It still wasn't cheap. This kind of configuration would be in the neighborhood of a thousand dollars, maybe more, but it was possible. Perhaps most interestingly is that the Altair with a VDM1 and serial card could function as a smart terminal in its own right. In that sense, we're seeing the full realization of Felsenstein's convivial cybernetic device. It looked like he had finally succeeded in his initial task. But the VDM-1 wasn't an ending. It was really just a start, a foot in the door. And this brings us back to our prolonged digression. Solomon of Popular Electronics

was on the lookout for a computer terminal. He contacted Bob Marsh and made an offer. Get me an intelligent terminal within 30 days, and I'll put it on the cover of Popular Electronics. Marsh talked it all over with Felsenstein, who thought it was definitely possible. The pieces were there, after all. They just needed to be somehow put together. Alright, that brings us to the end of part one on the story of the wedge.

At the ending here, there's a theme that I really want to highlight. Today, we've looked at just one aspect of the larger story of the development of personal computing. The story of a bunch of electronics nerds in their garages cobbling together computers is really just the home stretch if you think about it. A Wozniak or a Felsenstein was put in a position to build a personal computer thanks to a really slow dance that had been going on for decades. We can see shades of this everywhere. Felsenstein directly cites DEC as part of the technical inspiration for his Tom Swift terminal design.

DEC itself had been working on making smaller and cheaper computers for years. Community Memory, the project that started Felsenstein on this whole quest, was out on its own quest for ways to make computing accessible to normal people, and that happened well before the microprocessor really enters the picture in a meaningful way. What I'm getting at is this. It's often easy to conflate the rise of home microcomputers with the birth of personal computing. The simple fact is that the story isn't that simple. That's the reason I've been taking such a slow approach in this episode, and hopefully it hasn't been too

agonizing. Next episode, we're going to pick up exactly where we left off. Processor technology has just a month to create a prototype for popular electronics. The terminal they built would get a little bit of feature creep, becoming a fully-fledged computer. Along the way, we'll see a familiar shape take form. We'll also take a look at how the wedge spread from patient zero

to a larger population. Until then, thanks for listening to Advent of Computing. And hey, if you like the show, there are now a few ways you can support it.

If you know someone else who would like the show or is interested in the history of computers, then why not take a minute to share it with them? You can rate and review on Apple Podcasts as well. And if you want to be a super fan, you can support the show directly through Advent of Computing merch or signing up as a patron on Patreon. Patrons get early access to episodes, polls for the direction of the show, and bonus content. You can find links to everything on my website, adventofcomputing.com If you have any comments or suggestions for a future episode, then go ahead and shoot me a tweet

I'm at Advent of Comp on Twitter And as always, have a great rest of your day