Sol-20

Lee Felsenstein was one of the sysops of Community Memory, the first public bulletin board system. Community Memory opened in 1973, running on a SDS 940 mainframe that was accessed through a Teletype Model 33, essentially a computer printer and keyboard, in a record store in Berkeley, California. The cost of running the system was untenable; the teletype normally cost $1,500 (their first example was donated from Tymshare as junk), the modem another $300, and time on the SDS was expensive – in 1968, Tymshare charged $13 per hour (equivalent to $109 in 2022).[4] Even the reams of paper output from the terminal were too expensive to be practical and the system jammed all the time.[5] The replacement of the Model 33 with a Hazeltine glass terminal helped, but it required constant repairs.[6][lower-alpha 3]

Since 1973, Felsenstein had been looking for ways to lower the cost.[7] One of his earliest designs in the computer field was the Pennywhistle modem,[8] a 300 bits per second acoustic coupler that was 1/3 the cost of commercial models. When he saw Don Lancaster's TV Typewriter on the cover of the September 1973 Popular Electronics, he began adapting its circuitry as the basis for a design he called the Tom Swift Terminal. The terminal was deliberately designed to allow it to be easily repaired.[9] Combined with the Pennywhistle, users would have a cost-effective way to access Community Memory.[10]

In January 1975, Felsenstein saw a post on Community Memory by Bob Marsh asking if anyone would like to share a garage.[lower-alpha 4] Marsh was designing a fancy wood-cased digital clock and needed space to work on it.[lower-alpha 5] Felsenstein had previously met Marsh at school and agreed to split the $175 rent on a garage in Berkeley. Shortly after, Community Memory shut down for the last time,[13] having burned out the relationship with its primary funding source, Project One, as well the energy of its founding members.[5]

As supplied, the Altair 8800 could only be programmed via front-panel switches and lights. A cost-effective terminal was lacking.

January 1975 was also the month that the Altair 8800 appeared on the front page of Popular Electronics, sparking off intense interest among the engineers of the rapidly growing Silicon Valley. Shortly thereafter, on 5 March 1975, Gordon French and Fred Moore held the first meeting of what would become the Homebrew Computer Club. Felsenstein took Marsh to one of the meetings,[14][lower-alpha 6] Marsh saw an opportunity supplying add-on cards for the Altair, and in April, he formed Processor Technology with his friend Gary Ingram.[14]

The new company's first product was a 4 kB DRAM memory card for the Altair. A similar card was already available from the Altair's designers, MITS, but it was almost impossible to get working properly.[16] Marsh began offering Felsenstein contracts to draw schematics or write manuals for the products they planned to introduce. Felsenstein was still working on the terminal as well, and in July, Marsh offered to pay him to develop the video portion.[7] This was essentially a version of the terminal where the data would be supplied by the main memory of the Altair rather than a serial port.[17]

The result was the VDM-1, the first graphics card. The VDM-1 could display 16 lines of 64 characters per line,[lower-alpha 7] and included the complete ASCII character set with upper- and lower-case characters and a number of graphics characters like arrows and basic math symbols. An Altair equipped with a VDM-1 for output and Processor Technology's 3P+S card running a keyboard for input removed the need for a terminal, yet cost less than dedicated smart terminal products like the Hazeltine.[19][lower-alpha 8]

Before the VDM-1 was launched in late 1975, the only way to program the Altair was through its front-panel switches and LED lamps, or by purchasing a serial card and using a terminal of some sort. This was typically a Model 33, which still cost $1,500 if available. Normally the teletypes were not available – Teletype Corporation typically sold them only to large commercial customers, which led to a thriving market for broken-down machines that could be repaired and sold into the microcomputer market.[21] Ed Roberts, who had developed the Altair, eventually arranged a deal with Teletype to supply refurbished Model 33s to MITS customers who had bought an Altair.[21]

Les Solomon, whose Popular Electronics magazine launched the Altair, felt a low-cost smart terminal would be highly desirable in the rapidly expanding microcomputer market. In December 1975, Solomon traveled to Phoenix to meet with Don Lancaster to ask about using his TV Typewriter as a video display in a terminal. Lancaster seemed interested, so Solomon took him to Albuquerque to meet Roberts.[21] The two immediately began arguing when Lancaster criticized the design of the Altair and suggested changes to better support expansion cards, demands that Roberts flatly refused. Any hopes of a partnership disappeared.[22]

Solomon then traveled to California and approached Marsh with the same idea, stating that if they could produce the design within 30 days, he would put it on the cover of the magazine. Marsh once again hired Felsenstein to design the system.[22] As Felsenstein later noted:

In the process leading to the birth of the Sol, Leslie Solomon performed the act of the male. I still don't know who solicited whom.[23]

Felsenstein initially wanted to build a terminal following the model of his earlier Tom Swift design, using discrete electronics.[18] Marsh, in parallel, sketched out a version using the Intel 8080. It quickly became apparent the difference in cost would only be about $10, and from then on the original dedicated terminal concept was dropped.[22][7] Over time the plans changed, and at some point, Marsh told Felsenstein "We want you to design a computer around the VDM display."[24]

Initially, the idea was to sell a kit system, as was common in the industry at that time. The kit concept would make it through to the release, at which time it was known as the Sol-PC. As the design process continued, at some point the decision was made to offer the system in complete form, with all the parts needed for a complete system.[lower-alpha 9]

Felsenstein originally thought he was only needed for the initial design, but as the physical layout began it was clear that the layout artist they had hired would not be able to do it on his own. Marsh had a woodworker friend build a large light table and Felsenstein and the layout artist began using it to design the printed circuit board for the motherboard.[25] While Felsenstein worked on the design, Marsh continually came up with new ideas that he demanded to be included. This led to creeping featuritis problems and the final design was not delivered until about two months of "frantic" work.[26]

The final product consisted of a single motherboard with the 8080, a simplified version of the VDM-1, serial input/output, and 1k of SRAM for the screen buffer.[1] A ROM, the "personality module", would include the terminal driver or other code which would begin running as soon as the machine was turned on. The module was designed so it could be removed or inserted without accessing the interior of the machine.[26]

Marsh, meanwhile, was working on the physical design. He demanded from the start that it use walnut sides; while working on the digital clock project he had learned from his woodworker friend that they could get parts for practically nothing if they were small enough to be made from off-cuts. Beyond that requirement, anything was fair. The deadline for the magazine had been pushed back, but there was still little time to finalize the layout before it needed to be photographed. Marsh decided that the machine should have a cassette deck, so they mocked up a machine with a keyboard on the left and cassette player on the right.[25]

The first motherboard arrived 45 days after the project started, and the first cases and power supplies about 15 days after that. By this point it was clear the system was a usable microcomputer on its own, but "the decision was made to soft-pedal the fact until the last possible moment. Once published, all the fuss possible was to be made about its general-purpose nature; but until it actually saw print, it was to be treated first as a terminal."[26]

As the machine increasingly expanded in power, Felsenstein suggested the name "Sol", because they were including "the wisdom of Solomon" in the system.[7] Les Solomon would later quip that "if it worked, they'll say Sol means 'sun' in Spanish. If it don't work, they're gonna blame it on the Jewish guys."[27] Stan Veit later joked to Solomon that they named it after him in another way, "the LES Intelligent Terminal".[28]

The original mock-up of the Sol shows a slimmer case and a cassette deck where the numeric keypad would ultimately be placed.

In February 1976, the first machine, a kludged-up box of parts, was readied and flown to New York to show Solomon. As he pointed out the features, Solomon asked what was stopping anyone from putting a BASIC on the personality ROM. Felsenstein, who had been told to avoid referring to it as a computer, simply replied "beats me".[24] When they powered it up the machine would not work, displaying unreadably fuzzy images. Marsh and Felsenstein then flew to Boston to visit the offices of the newly started Byte magazine. While there, Felsenstein had time to discover the problem was a tiny bit of broken wire that got stuck under a chip, shorting out two of the video lines. They returned to Solomon's house to demonstrate the working unit.[24]

Due to publication timelines, it did not appear in the magazine until the July 1976 issue, where it was described as "high-quality intelligent terminal". The cover image showed the mockup version, packaged in a slim case.[29] By the time the article appeared, the design had changed; the new design had a distinct "step" behind the keyboard that rose up over the expansion chassis and power supply at the back of the case. A bent piece of sheet steel formed most of the case, capped on the left and right by the wooden panels Marsh demanded.[25]

The new design was first shown at the Midwest Area Computer Club conference in June 1976. The machine was not ready for sales at this point, but they did a brisk business selling their existing expansion card line.[30] This was followed by the Personal Computing '76 (PC'76) show in late August in the dilapidated Shelburne Hotel in Atlantic City. The order book was officially opened[3] and Sol was a huge hit at this show.[31]

Soon after, Marsh was invited to demonstrate the Sol on NBC's The Tomorrow Show. They used a game by Steve Dompier called "Target" to show off the system's capabilities.[32] The show's host, Tom Snyder, ended up playing the game right through the commercial breaks, and they had to force him to give up the machine in order to finish the show.[33]

The Sol was initially offered in three versions.[34] The base motherboard was offered as the Sol-PC, available as a kit for $575, or fully assembled and tested for $745. The Sol-10 added a case, keyboard and power supply, was $895 in kit form and $1,295 assembled. Finally the Sol-20 added a keyboard with numeric keypad, and a larger power supply to feed the five expansion slots and a fan to cool them, for $995 as a kit or $1,495 assembled. Advertising of the time referred to the Sol-20 as "The first complete small computer under $1,000".[35] Most systems would require additional pieces, which they bundled as the "Sol Systems"; the Sol System I consisted of a Sol-20, an 8k RAM card, a PT-872 monitor and the RQ-413 Cassette Recorder, for $2,129.[36]

In keeping with the hacker ethic, the company also offered to send out copies of the schematic for the motherboard for the cost of postage, later estimating that somewhere between 40,000 and 50,000 copies were sent. Few, if any, Sol-10s were sold,[lower-alpha 10] and the company focused on the Sol-20. The first machines shipped in December 1976. These were also available for third-party sales, and this began the formation of a dealer network among some of the earliest computer stores. By 1977, Processor Technology had a reputation for quality and was among the best-selling computers in the world.[31]

By this time, S-100 machines were beginning to make inroads into business markets. Processor Technology invited all of their dealers to a meeting in Emeryville, California, outside Berkeley, to introduce their Helios floppy disk drive for $1,199, along with their PTDOS system to work with it. They also promised larger memory cards and a color video card. Additionally, dealers could now order 30 days net, as opposed to cash-on-delivery, although to do so they had to put in orders at least once a quarter.[31]

These plans quickly fell apart. The Helios was initially based on a new mechanism from Diablo Data Systems. Diablo had been purchased by Xerox in 1972, and shortly after the Helios was announced, Xerox canceled development of the floppy line. Processor Technology selected the new Persci 270 in its place. The 270 had two drive bays operated by a single drive and voice coil head positioner, which meant a two-drive system was only slightly more complex than a single drive. This was released as the Helios II, at $1,895 for the kit or $2,295 assembled. Processor Technology moved to a much larger factory in Pleasanton, California.[31]

It was at about this point that Radio Shack introduced the TRS-80. Like the Sol, it was a complete all-in-one machine but came with its own monitor and sold for about half the price. Moreover, it was available at hundreds of Radio Shack stores across North America. Sales of the Sol plummeted. Meanwhile, the company failed to introduce any of the other new products it mentioned, notably the color graphics card. When the Apple II appeared with color graphics, it quickly became a best seller.[31]

To add to their woes, Processor Technology had contracted North Star Computers to write a new version of the BASIC for the Sol machines. North Star then began selling the resulting North Star BASIC to other vendors as well. Processor Technology sued North Star, claiming the contract had been exclusive. The suit dragged on, hurting both companies before Processor Technology ultimately lost. To add to the injury, North Star then released a new 5.25-inch drive for the system that sold at half the cost of the Helios. A patch that allowed CP/M to run on the new drives killed off any interest in alternatives like PTDOS, and new business applications like WordStar and Electric Pencil soon cemented CP/M as the standard operating system for all S-100 machines.[31]

Processor Technology continued selling the Helios system and refused to consider replacing PTDOS with CP/M. Helios proved to be highly unreliable and resulted in a lawsuit by those owners that had purchased them. Meanwhile, the company introduced one of its few new products during this period, 32 and 64 kB memory cards based on dynamic RAM which was much denser than the older SRAMs. These began failing at an alarming rate, overwhelming the company's ability to repair them.[31]

These problems caused the company to go bankrupt, and the company was eventually liquidated on 14 May 1979.[31] Talks of producing a successor to the Sol-20 between Felsenstein and a group of investors including Adam Osborne fell through after Felsenstein enumerated all the improvements that would need to be made to make it competitive in the burgeoning early 1980s home computer market. However, in 1980, he collaborated with a Swedish businessman named Mats Ingemanson and brought to market the Micro Expander, seen as the spiritual successor to the Sol-20.[37]

A Sol-20 with the rear cover removed.

Sol-20 motherboard. Note the expansion chassis in the center, and the Personality Module ROM cartridge on the extreme right.

Two PM-5204, two PM-2708, one Golemics, and one Micro Complex dual personality modules for Sol-20

Looking at the Sol-20 from the front, where the operator would sit, the keyboard was in a typical location with the main QWERTY-style layout on the left and the numeric keypad on the right. The wooden sides of the case were close on either side of the keyboard, potentially interfering with the operator's hands.[38]

On the rear right of the case (as seen from the front), directly to the rear of the numeric keypad, was the power supply, which also provided a fan to cool the circuitry. The main motherboard sat to the left of the power supply, spanning about 2/3 of the case's width. The motherboard extended forward under the keyboard all the way to the front of the case.[38]

Cassette, parallel and serial ports extended off the back of the motherboard into holes in the case. Directly below the fan, was a UHF connector that produced composite video output. This could be connected to a monitor, or with a bit of work, a conventional television. The processor was near the back of the machine, with the memory and video circuits at the front. This required the video output to be routed to the back of the machine with a coaxial cable running across the top of the card.[39]

Originally, expansion was going to be handled through an external cage that connected to the main console using two 50-pin ribbon cables. The original Altair bus design lacked signal ground pins for each of its data lines, a decision that had been made in order to reduce pin count and allow it to fit into 100-pin connectors they found in surplus. This led to noisy signals as they all shared a common ground, a topic of considerable derision by many users. When the bus was extended into a ribbon cable, the resulting signals were too noisy to be useful, and Marsh demanded that there be additional ground pins spread across the cable to reduce this noise.[26]

The Sol solved this problem by supporting only one of the two data busses at a time, allowing input or output and switching between them by signaling with the DBIN pin on the 8080. Since only one bus was being used at a time, they could share a set of eight pins, which allowed the eight formerly dedicated to the second bus to be used as ground lines instead. Ultimately, the idea of using an external chassis was dropped. By this time the decision to use the additional lines for grounds had been made, which had the desirable side-effect of making the board easier to design.[26]

The same 50-pin concept was instead implemented in an internal expansion chassis, the Sol-BPB. This extended vertically upwards from roughly the center of the motherboard. It had five horizontal connectors, and a metal framework on either side mechanically supported the expansion cards. The chassis also had another edge connector at the top, but it is unclear whether this could be used for further expansion.[40] The BPB retained the DBIN signaling and ground pins of the early design and this quickly became a de facto standard for S-100 cards.[26]

This change to the bus design was contentious, as it meant cards for the Altair did not work in the Sol without some adjustments. Felsenstein noted, "I take the position that Bob made me do it, and he takes the position that history will absolve him."[26]

Three "personality modules" were released with the original systems.[41] CONSOL provided a simple terminal emulator function, along with a small number of additional commands to load and run programs from tape using TLOAD. SOLOS added names to the files on the cassette, the TSAVE command for saving data to the tape into a named file, and TCAT to print out the details of a named program. TXEC loaded and executed a named program in one step. SOLED included block-mode editing, used on some mainframe systems, but it is not clear if this was actually available.[42]

One commonly used software for the Sol-20 was the BASIC/5 language. This was able to run in even a minimal machine with a 4 KB expansion, but in order to fit it had only single-precision floating point numbers and lacked string variables. An Extended BASIC that ran in 8 KB added strings and other functions. Processor Technology also sold a wide variety of other programs, including many games, on cassette format for the Sol, or on punch tape for other S-100 machines.[43]

• Crosby, Kip (February 1996). "Computers for Their Own Sake: From the Dompier Music to the 1980 Computer Faire". The Analytical Engine. The Computer History Association of California. 3 (2): 8–20. ISSN 1071-6351 – via the Internet Archive.
• Anderson, David (December 1976). "Product Review: Processor Technology VDM-1" (PDF). Byte. pp. 36–39.
• Barbour, Dennis (April 1978). "User's report: The SOL-20" (PDF). Byte. pp. 126–130.
• Hogan, Thom (April 27, 1981). "Sol Evolves". InfoWorld. CW Communications. 3 (8): 1, 34 – via Google Books.
• Felsenstein, Lee (July 1977). "Sol: the inside story". ROM Magazine. pp. 60–65.
• Felsenstein, Lee (7 May 2008). "Oral History of Lee Felsenstein" (PDF). Computer History Museum.
• "The Small Computer Catalog" (PDF). Processor Technology. 1977.
• "Sol System Price List" (PDF). Processor Technology. March 1977.
• Levy, Steven (2010). Hackers: Heroes of the Computer Revolution – 25th Anniversary Edition. O'Reilly Media. ISBN 9781449393748.
• Sol Systems Manual (PDF). Processor Technology. 1977.
• "Now You Can Build a High-Quality Intelligent Terminal". Popular Electronics. July 1976.
• Swaine, Michael; Freiberger, Paul (2014). Fire in the Valley: The Birth and Death of the Personal Computer. Pragmatic Bookshelf. ISBN 9781680503524.
• Veit, Stan (1993). Stan Veit's History of the Personal Computer. WorldComm. ISBN 9781566640305. Alt URL
• Younger, J. Kelley, ed. (January 1986). "Brave New Reality". PC World. Vol. 4, no. 1. PC World Communications. p. 45. ISSN 0737-8939.