SCELBI 2002 (Under Construction)

Switch Bank
Switch Detail (note springs)
8008 Motherboard

Scelbi-8B

This is my tribute to the SCELBI Computer, arguably the first US microcomputer you could reasonably call personal or affordable, available to the "common" geek. I believe that 2003 will be the 30'th anniversary of the SCELBI.

SCELBI (Scientific, Electrical, Biological) Consulting was a pioneer of computing. The SCELBI 8-B/H was available in kit or partial kit form. Fully loaded, the Intel 8008-based machine was quite expensive because of the high cost of 1101 256x1 bit RAM chips. Strangely, it did not have a built in power supply - this was external to the computer itself and connected by cable.

SCELBAL

SCELBI also wrote a considerable amount of software for the machine and developed significant know-how. Apparently after stopping the production of the SCELBI computer (after selling about 200 at a loss of $500 each, according to sources) they released programming books and a BASIC-like language for the 8008, SCELBAL. It was available in a book, in heavily commented source form as well as modifiable object code, at a price of only $49.95. This was incredibly inexpensive at the time for any BASIC, much less one with source code and floating point capability. The code was also assembleable on an 8080 computer, although it didn't take advantage of the enhanced features of the 8080. This book and the guide 8008 programming sold very well and SCELBI decided that software publishing was an easier way to make money (which is certainly true even today).

It should be noted that Bill Gates himself stated that it would be impossible to write a BASIC for the 8008.

Why didn't SCELBAL become popular? Firstly, by the time it was released, the 8008 was somewhat obsolete; no new commercial or hobby computers were being designed with the 8008. From a system point of view, the far superior 8080 actually cost less than the 8008 by 1976. Secondly, it should be noted that SCELBAL required 8K to run reasonably well (remember, the 8008 took a lot of instructions to do anything). 8K in 1976 was a lot for a hobbyist to afford. SCELBAL, as far as I know, was never actually released in machine-readable form; one of the reasons for this was that before the Altair 8800, I/O functions were not standardized and any implementation of SCELBAL would require some modification, requiring considerable know-how by an average hobbyist. Imagine having to write and link in a device driver into Excel every time you upgraded your video board or changed your keyboard!

A "stock" SCELBI, fully loaded, could hold only 4K of RAM. UVEPROM was very, very expensive; ROM was not much cheaper, and took a great capital expenditure purchase them with minimum quantities of 1000 or so.

The most important factor (in my opinion, anyway) in SCELBAL's lack of success was the availability of Bill Gates' 4K BASIC for the 8080. Although it was originally sold only with the Altair 8800, and later made available at a high $200+ price, in object form only. It was heavily pirated and easily modified. The laws at the time were nebulous about computer software; I knew someone at the computer club that I belonged to in 1976 who provided copies of "Randy's 4K BASIC" for $5.00 on paper tape, claiming that having a different name patched into the code, it was not the same work, and, therefore, not copyrighted. This may or may not have been true given copyright law at the time. MITS BASIC only ran on an 8080, but by 1976, this didn't matter. $5.00 for a machine-readable (although pirated) and superior BASIC vs $50.00 for a larger, slower BASIC you had to type in (using your very, very limited computer) and modify yourself - the decision was pretty clear to the always-cash-strapped hobbyist. Keep in mind that in 1976, even $50.00 was a lot of money to a young person.

The president of SCELBI Consulting, Nat Wadsworth suffered two heart attacks during the production of the SCELBI computer, and SCELBI seems to have disappeared around 1979 or so. His book on programming the 8008 was largely written from a hospital bed.

Mr. Wadsworth later went on to become successful as an author of computer programming books in the 80's . I believe that he died of a heart attack in the early 1990's. It is worth pondering what would have happened if SCELBI had just a little more financial and business resources and/or Mr. Wadsworth's health would have been better. God knows, given their location (Boston), they would have had limitless "geek" support from MIT. Maybe the center of computing technology would never have shifted to the West Coast.

This computer was constructed in a 3H rack mount, with a switching power supply. The board uses "period" components and technologies, notably an original C8008, Intel C2708 1K UVEPROMS, and 7K of 2114's. A ceramic/gold AY-5-1013 provides UART functions, which are buffered by a 1488/1489 combination.

The SCELBI 2002's hand wire wrapped board is a combination of the original SCELBI design and the Martin Research "Mike" 8008 computer, with a few touches of my own. I did this mostly because the original SCELBI design was quite complex, not because of parts availability. I have the SCELBI 8-H prints, and no TTL components unobtainable in 2002 were used in the SCELBI design. The only "rare" parts need to build a SCELBI would be (obviously) the 8008 itself, and also the 1101 RAM. I feel that using 2102's in an 8008 design is acceptable (the original MARK-8 had a third party board using 2102's, and it would be reasonable that a low-budget homebrewer would use 2114's in the late 1970's with an 8008). The amount of 1101 RAM I would have required would have cost at at least $1000 today (if I could find the chips) and would require considerable wiring work, board space, and power (7K would require 224 1101's, possibly drawing over 20 amps at 5 volts).

Note that the more famous Mark-8 8008 design is far "messier" than the SCELBI, and it uses unobtainable logic chips to boot.

I have implemented only two I/O serial ports, one bit-banged and one UART; seven parallel input and 3 output bits are available but not brought out to a connector. The original SCELBI had a tremendous amount of I/O using expensive connectors (IMHO, more than anyone would reasonably need from such a modest computer).

The front panel almost exactly mimics the feel and operation of the original SCELBI front panel. The paddle switches (now very, very hard to find) were a problem, particularly momentary switches, which are just not made much anymore. See Figure X to see how a momentary switch was constructed from a typical "normal" paddle switch. Note that the springs are not visible from the front of the unit. I wanted this machine to be "first class" in appearance as well as operation and I would not settle for metal toggle switches from Radio Shack.

The low-brightness T-1 LED front panel lamps are a superset of those found on the original SCELBI-8B's front panel; All T states are displayed, the bus states are decoded, and flags are shown, in addition to the H L registers and the memory contents (the two LED's at the upper left show that the +5 and -9 power supplies are present). All wiring between the front panel and the motherboard is divided between 34 and 26 pin ribbon cables.


The UVEPROM's contain SCELBAL, which I had reconstructed, made ROMMable, and made a UART I/O modification to. However, the present firmware functions internally exactly (although is not assembled exactly the same) as the original SCELBAL software, operating at a mind-numbing 110 baud, bit-banged in and out of an I/O port.

The operation of the SCELBI front panel is far more tedious in use even than the Altair 8800's. It utilized the 8008's capability of "jamming" an instruction or data on the bus during a wait state after an interrupt.


Note that you can do anything with this front panel; examine any register; or input from or output to any port, using the 8008 itself to perform the instruction. It does take, however, a lot of work to do it!


Incredibly, there is a note on USENET about a guy who loaded SCELBAL (approximately 7K) this way! Obviously, this man has a great deal of patience and/or was some kind of obsessive nut!

Fortunately, I have set up my machine to require only a toggled in JMP 000 024 to the SCELBAL interpreter. I could have set up SCELBAL to run immediately from ROM upon reset, but I thought that to be TOO easy, violating the "spirit" of the machine.

This may be the only 8008 computer in existence that is capable of running SCELBAL (including the very few real SCELBI computers in collections). If not, I'd like to hear from you. Or, for that matter, I'd like to hear from any 8008 based computer owners.

I have written an emulator for a SCELBI and for SCELBAL to run in, available here.