LC8080 Manual

Free Web Hosting Provider - Web Hosting - E-commerce - High Speed Internet - Free Web Page

LC8080 Single Board Computer

User Manual

Introduction

The LC8080 Single Board Computer is a contemporary recreation of a 1970's prototype 8080 computer, utilizing the components and technologies available at the time. Certain components are actually of 70's vintage; this varies from unit to unit. The unit's primary purpose is to serve as an artistic showcase for the 8080A processor and the prototyping techniques of the 1970's. As the 8080 is the grandfather of the 8088 in the original IBM PC, and the heart of most of the first practical "personal" computers, it certainly ranks as one of the most important designs in computing and electronics history.

Hardware Features:

8080a Processor, various manufacturers, running at 1.78 Mhz (16 Mhz crystal)or AMD 9080 improved, "higher speed" version of the 8080 operating at 2.22 Mhz with 20 Mhz crystal (depending on unit)

Highest quality available components are utilized (e.g., expensive gold-inlaid machined pin wire-wrap sockets, fiberglass tinned prototyping board)

Data and Address Lines fully buffered - no "marginal" signals - usually found only on full-sized systems. Extensive bypassing.

2K of ultraviolet-erasable programmable read only memory (UVEPROM)

1K of random access memory (RAM)

Two 4-character 5x7 matrix LED display, of a quality usually seen only in medical/avionics applications

Oversized linear regulator, heat-sinked, and inverter for special 8080/8224 voltages

Oversized UL-approved AC adapter

Power and error LED's

Processor status LED's (status LED's depending on model), generated by 8212.

Proper, manufacturer-recommended crystal controlled 8224 monolithic two-phase clock generator (unlike many 8080-based computers of the 70's, notably the original Altair)

Reset switch

"True to period" design with no technology or chip unavailable in the '70's used (examples: no smart LCD displays, no green LED's, no 64K UVEPROM of flash,no monolithic capacitors, etc.).

Software Features

Power on self-tests for CPU, RAM, and ROM, display with diagnostic LED.

Message customizable at the "factory".

Receiving Inspection

A great deal of work and shipment testing went into the packaging. Use scissors or a cutter to remove the AC adapter and board from the box. Inspect for damage. It is fantastically unlikely, but possible, that a chip could come loose during shipment. See that the chips are not sticking out from sockets. If this should happen, email me immediately. You should save the container for future

Installation

Keep in mind that this is not exactly a consumer product. There is obviously voltage present on the board, and some of the components do get hot because they are designed to do so. No voltage is higher than +12 or -5, but nonetheless under bizarre conditions a shock or fire hazard could exist, especially if the unit is dropped or falls while unattended. Touching points on the board with metal (like that you might have on your finger, a ring) will
probably cause serious damage to the board and may possibly heat up the metal enough to cause a burn. A secure mounting is EXTREMELY
important both for the safety of the unit and your own safety. Make sure that the "wall wart" cord is placed in such a manner that it cannot be tripped over.

It is a good idea to make certain that the unit will not be in direct sunlight, for heat and other reasons, and out of the reach of bystanders.

Static Electricity

Integrated circuits are far less susceptible to static electricity when installed in a circuit. However, consider the following:

IC's can be "hurt" by surprisingly low static voltages that you could never feel when touching a doorknob, for example. These effects on mean-time-before-failure (MTBF) are cumulative. You might knock a chip with a MTBF of decades down to a few years after several good "hits", even if it is not obviously damaged. The amount of "shock" required to do this is far less than a shock that you could feel.

The best policy is to avoid touching the unit's circuit board whether or not it is on; if it is to be in a public place, it should definitely not be accessible to touching. Remember, however, that the unit generates considerable heat, so ventilation is important.

UVEPROM Label

This label serves to identify the program contained in the UVEPROM and also to cover the chip's window. The window is made of a special glass that passes the UV rays that erase the chip. A special high-intensity (and dangerous) UV lamp is used to accomplish this; however, office lighting and especially sunlight will erase the chip at a much slower rate, but will nonetheless erase it. There are 16,000 bits on the chip; one of them changing from a 0 to a 1 will cause the unit to fail.

Even under perfect conditions, the storage cells in UVEPROM's will eventually discharge, but this is thought to take decades under ordinary conditions with an opaque label covering the chip's window. For this reason, it is strongly recommended that you acquire the CD-ROM after executing a non-disclosure agreement (NDA). Sometime in the distant future, you or your heirs might need to have the UVEPROM reprogrammed!

Therefore, do not remove the UVEPROM label.

Self-Test and Operation

Upon power-up, the unit will go into a self-test mode. Note: if you see no display, press the "reset" switch. The power-on reset circuit does not work under all conditions. You should see:

CPU PASS
RAM PASS
ROM PASS

Then, for one second or so, all LED's on the display should be lit. You should see no "holes". The diagnostic LED will flash, once.

Should there be a failure, the diagnostic LED (is a slightly different location depending on unit) will flash with the code for the error. The reason this is not displayed on the alphanumeric display is because a failure would definitely make this unlikely to function correctly. The codes are:

Two short flashes, long pause-CPU failure
Three short flashes, long pause-ROM failure
Four short flashes, long pause-RAM failure

See the Theory of Operation for more details.

Read This: Heat is NORMAL.

Certain components on the LC8080 get hot. Keep in mind that this board utilizes 1970's NMOS and PMOS technologies, which were not nearly as power-efficient as the modern CMOS that you would find in your PC. Furthermore, all modern computers use "switching power supplies", which only pass the voltage and current required to operate the electronics. This board uses a brute force linear regulator, which dissipates excess power as heat. After testing, I have determined that under room conditions, the board will never get hot enough to burn persons or property. However, the uninformed might be startled or upset about the heat output. Here are the "hot points"

1. LM323K Linear Regulator

This IC (it is an integrated circuit, not a transistor) is basically an integrated circuit built around a large pass transistor. Approximately 9 volts go into the unit from the wall wart, and 5 volts is delivered to the board's circuitry. The rest is dissipated as heat. Higher voltages will be dissipated as more heat, which might overheat the 323 and cause it to shut down. This is why it is extremely important that the provided AC adapter only be utilized to operate this unit.

2. 8224 Clock Generator

This chip produces a high-voltage, high frequency clock, and does dissipate quite a bit of power. It was designed to run "hot", however.

3. 8080 Processor

This processor is PMOS, and does consume quite a bit of power. This is normal. Consider this; the chip has 6,400 transistors on a 6-micron process. If you made a CMOS Pentium II with a 0.18 micron process and millions of transistors this way, it would probably take a welding transformer to power it and it would operate for only a few milliseconds before shattering from the heat!

4. HP Displays

These displays give up efficiency for brightness, although in their time were considered very efficient, and advertised as daylight-readable - very important in military applications. They do dissipate quite a bit of heat. However, their ceramic and glass construction is designed to tolerate the heat.

General Design Theory

(A more detailed version exists on the CD-ROM, available to the purchaser after execution of an NDA)

Power Supplies

The "wall wart" provides approximately 9.3 volts of power, which is fed to the LM323K linear pass regulator, where is is dropped to +5 volts. The "black box" beneath is a high-frequency DC-DC converter (yes, these did exist in the 70's-every HP calculator built from '72 to '80 had one!) creates +15 and -15 volts from the +9.3 volt input. The +15 is dropped to +12 by another linear regulator, which requires no heatsink due to low power requirements. -15v is dropped to -5 by a resistor and Zener diode. The current requirement is very small. Bulk filtering via electrolytic caps is utilized on most outputs and the 323 input; bypass capacitors are used throughout the board, generally on a 1 to 1 basis with each IC, which is quite conservative. Obviously, wire-wrap techniques are generally "noisier" than PCB designs, and this was deemed a requirement. Oscilloscope displays show that it was worthwhile.The rear of the board uses +5 and ground busses.

+5 is used by every IC on the board.

+12 is used by the 8212 and the 8080.

-5 is used as the "drain" voltage for the PMOS-based 8080.

Clock/Reset

The 8224 chip creates the special two-phase, high voltage clock that drives the internal circuits of the 8080. The two chips share several sync signals for reset and clock. The reset switch and power-on reset are connected to the 8212. The power on reset consists of a resistor and capacitor. A crystal regulates the frequency of the clock(s).

CPU and Buffers

The CPU's data, address, and control signals have very limited drive capability (as did most large-scale IC's of that era). Except in very small
designs, it is necessary to "buffer" and amplify these signals. The LS241 chips do this for the outgoing addresses; the 245 is bidirectional, and
switches back and forth under processor control between outgoing and incoming data.

8212 Status Signals/LED's

This chip is not required for operation, but is used to decode certain processor signals and make them available to the engineer via oscilloscope displays or LED's. Or, it is there because a computer of the 70's HAD to have blinking lights, take your pick. The LED's are flashing thousands to hundreds of thousands of times a second during operation. Not all units are built with the same display. Typically they are as follows:

LEDS (in order from top to bottom)
1 write output (the RAM is being written to)
2 stack operation (the processor is in a subroutine that required a stack push
or pop)
3 M1 ( a memory enable state)
4 memory read phase (memory, ROM or RAM, is being read)
(5) halt. The code in the LC8080 never uses a halt, input, or output
instruction, so this light is never lit in normal operation except possibly in
a reset state.

Address Decoder

The 74LS138 decoder decodes the highest three address lines to produce chip selects for the RAM, ROM, and display/373. Having 65K of address space, this produces a memory map on 8K boundaries. The ROM must exist at address 0, because this is the reset vector of the 8080.

TTL Logic

The LS04 and '00 invert certain signals and produce the chip select latches for the 373 "output port" which actually exists at a memory address and is not addressed by I/O instructions.

RAM

The RAM is 1K by 4 bits wide; therefore two together yield 1K by 8 bits. By today's standards, they are terribly slow, about 500 nanoseconds, but fast enough for the 8080. Not all of the RAM is utilized by the provided program, but it is all tested during the RAM test.

ROM

The ROM is UV-programmed in a special computer driven "EPROM blower". Naturally, it is read only, and "permanently" contains the program that operates the displays.

74LS373 "output port" and HP Displays

Note again that I/O instructions are NOT used on this board; the LS373 is memory mapped. This is because the board has no capability of decoding I/O instructions, and also because I/O fell out of favor generally, because memory-mapped "peripherals" are much more universal. Many microprocessors contemporary to the 8080 had no true I/O instructions.

The LS373 has 8 bits of output.

The first five turn the transistors that power the displays on and off. You might notice if you glance at the display quickly that the entire display is never lit, it is "scanned". This was, and is, a common technique. If the CPU had individual control of the displays, 7x5x8 wires, bits from ports, and transistors would be required.

Two bits are used for display "clock" and data. Data "out" from the shift register of display 1 goes to the data in of the second display. Display 1 is on the right, so the displays scroll right-to-left.

The display is refreshed as follows:

The display is turned off.
The CPU looks up the character to be displayed at a character position.
It looks up the column to be displayed.
It looks up the bit pattern for that column of that character in a table.
It clocks the bit pattern into the display for that column, serially.
It repeats this for all eight characters. The bits "snake" from one character
to the next.
The column is turned on; each character has only one column lit.
The cycle is repeated for all five columns, many times a second.

I deliberately slowed down the display so that you could visually see how it is done if you try. However, it is certainly possible to do this fast enough to be completely invisible, even with a processor as "slow" as the 8080.

Interrupts

Interrupts are not used on this board, and are not decoded. The interrupt line is permanently pulled "high".

Error Display

The last bit is used for the error display. The LED connected to this bit is not usually lit. Note that the power on self tests cannot always display the
cause of a problem, because the problem itself may render this impossible. However, the error LED is more likely to work than the full alphanumeric display, and this is why the error LED exists.

Firmware

The firmware was written in assembler, and is not very elegant, mostly because I am fairly rusty in coding for this chip. After assembly, it is downloaded to an EPROM emulator from a PC and tested. When the program is shown to work, a UVEPROM programmer is used to "burn" the chip.

MTBF and Endurance

How long, exactly, will this unit operate before failing? This is difficult to say. Although the wire wrap looks delicate, keep in mind that the technique actually produces tiny "gas tight" welds from pressure on each corner of the pin, and there are several wraps. Barring physical damage, overheating, voltage spikes, and other transients, here is what can go wrong:

The large tube type capacitors on the unit (not the small disks) contain a fluid that can dry out over time, especially with heat. This can cause the power supply to "ripple", causing the computer to misinterpret data and crash. In extreme cases, the chips can be damaged.

The transformer will eventually open, and will have to be replaced. This should be easy to acquire in the future, but note that it must have a capacity of at least 1.2A and a voltage of no less than 7.3 nor more than 9.5 volts.

The LM323K does get hot, and will someday fail. It contains shutdown mechanisms that hopefully will prevent it from putting to much voltage to the computer. It is a common and inexpensive part and has been in production for 20 years or more. Hopefully, it will be available in the future.

The chips packaged in ceramic (which, until recently , were the only type the military would buy or permit to be used) are far less vulnerable to this phenomenon. Note that the high-value, most irreplaceable chips in the LC8080 are ceramic.

UVEPROMS are not "permanent" storage, although they are used this way often. They contain transistor cells that trap a charge. These charges, even without UV light, will eventually leak away. The time is measured, theoretically, in decades. I have not seen this happen although I have EPROMS that were programmed over twenty years ago. Nonetheless, it will eventually happen. This is why I provide the code on a high-quality CD-ROM, which, under good conditions, is thought to be stable for 50-100 years.

I mention these things as an aid in repairing the board for future generations of geeks!

Limited Warranty

The customer agrees that this is a special purpose, non-consumer device with special care requirements beyond typical consumer electronics. The device is sold as a curiosity and artwork, and not as a computing device.

The unit is covered by a limited warranty for one year for failure of a component or faulty labor. Exceptions to the limited warranty include, but are
not limited to, the following:

1. Physical damage, e.g., dropping the unit or bending wire wrap pins on the rear.

2. Disassembly of the board from the mounting.

3. Removal or modification of components

4. Static damage

5. Damage caused by improper line voltage or lightning

6. Improper mounting and/or ventilation

Repair or replacement of components will be accomplished with like components; however, given the rarity of the components used, exact physical replacement availability is not guaranteed (although a "safety stock" has been put aside for such eventualities). However, all reasonable efforts will be made to keep the unit as close to that originally sold as possible.

You acquire this unit with the understanding that my total liability is limited up to the purchase price of the product.

Bottom line is, I'll do what I can to reasonably accommodate any problems if you take reasonable care of the unit. However, I'm not exactly wealthy, so don't bother to get the ghost of Melvin Belli after me!

Software (Firmware) License:

You are licensed to use the software contained in the unit ("firmware"). You are not authorized to copy, modify, disassemble, reverse engineer, decompile, or transmit the firmware to other parties, or store it in any form on any information retrieval system, or to permit others to do so.

A copy of the source code, object code, simplified schematic sufficient for repair of the unit, memory map, and other details can be obtained on CD-ROM from the designer for $5.00 after execution by the customer of a signed non-disclosure agreement. This is recommended, as I hope that the 20 or so LC8080's that I build will be working far into the future, whereas, I'm not a young man, and I might not be working by then!

A Brief History of Intel and the 8080

This is a summary of documents available on the internet and recollections from my memory.

Intel, short for Intelligent Electronics, was started in 1969 to produce semiconductor memory. At first, it was named "Moore-Noyce", but Bob Noyce and Gordon Moore saw that more noise was a Bad Thing in the electronics arena. Intel enjoyed modest success in its narrow field of business.

Although there are those who would argue this point, most would agree that Intel developed the first microprocessor, the 4004. It was a four-bit
chip, released in 1972, designed for use in a calculator. Although Intel developed it exclusively for a customer, because of some bad luck on the customer's part and good luck for Intel, Intel retained the rights to the design and all of the technology they had developed in order to produce it.

Four bits was not enough to easily deal with character based data, and the 4004 was missing critical functions (like interrupts) that would keep it from being used in a "real" computer cost effectively. So Intel designed the 8008. Still slow, the 8008 had interrupts, a stack, and 8 bits. The 8008 achieved some success, although the speed was a disappointment to Intel - the original 8008 failed to meet its design goals. The 8008 was also a true horror to design a computer with - Intel put it in an 18-pin package, not nearly enough pins to deal with all of the signals a computer needed for input and output. These signals had to be recreated by external circuitry - circuitry that cost board space, power, testing time, money, and so on. This circuitry involved some real digital voodoo, and this, combined with Intel's horrible documentation at the time, made the chip very difficult to use in a computer, much less to write software for. (Incidentally, I expect to come out with an 8008 design similar to the LC8080 shortly, the LC8008.) Later, the 8008-1, a selected 8008, met the original design speed goals of the chip - still very, very slow. A very few computers that could be thought of as "personal" were designed, built, and sold with the 8008 CPU, but these were not very popular due to their high expense and very limited capabilities. Software was scarce and the talent to write it was even more so.

So, the 8080 was designed, and released in 1974 at a list price of $360.00, just for the chip! In 1974, you could get a drivable used car for that kind of money. An engineering fraternity at my university "scammed" Intel by creating a paper corporation, requesting samples, and getting a box of cosmetically defective 8080's for free! I myself remember paying $25.00 for an NEC 8080 in about 1976 after I destroyed my original chip with a slip of a voltmeter lead. I was making about $3.00 an hour at the time. Intel did a much better job of supporting the 8080 than with the 8008, and the 8080, while not nearly as easy to design with as today's micros, was much simpler to work with than the 8008.

The 8080 was soon replaced by the 8080a, which was electrically improved, although there were no new instructions and it wasn't any faster. Virtually all 8080 designs commercially produced in quantity were of the 8080a variety. The famous MITS Altair 8800 used this chip, and when combined with Bill Gates' BASIC, which he sold through MITS, the machine was actually practical.

The rest, as they say, is history.

Intel had "second sources" signed up to produce the chip, which was, and is, commonplace in the semiconductor business, although it seems insane at first thought to give your competitors the know-how to produce your product. No matter how good or economical the chip was, many companies would be fearful of using the 8080 in a commercial design unless there were "second sources", for the following reasons:

-Price Competition

Obviously, if there is more than one source, there will be competition.

-Availability

What if people at Intel went on strike? What if there was a problem at their factory - like an earthquake (a real possibility)? With no second source, a product using the 8080 might not be shippable, with disastrous business consequences for the company embedding it into a product. What if Intel "put the squeeze" on customers who had no choice but to pay up?

-Distribution

Intel licensed the 8080 design to many companies, such as National and NEC (Nippon Electric, Japan). Not all these companies had offices or the ability to sell in all parts of the world, and Intel certainly did not have the sales force at the time to cover all possible sales anyway.

-Production Capacity

Even at full tilt, Intel certainly could not have produced all of the chips required by the marketplace at that time, and would not have wanted to commit its production lines to producing only one kind of chip anyway.

Oddly, the Soviets cloned the 8080 chip, although this was not licensed or authorized. Soviet chip production quality was notoriously bad; batches of chips would be sent out with "errata" sheets, such as, "don't use the RLC instruction - it doesn't work". But the chips were in tremendous demand and production capacity was very small - so Soviet software designers would have to modify their code to accommodate certain batches of chips!

Improvements

Many second source chips were simply exact copies of the Intel design, possibly from the original masks. NEC made a variant that executed one commonly used instruction type faster; this was good for speed but it was also bad; firmware might need to be changed in critical timing loops if this chip was to replace the original Intel part. Another NEC variant made the clock work at 5 volts, doing away with the +12 power supply requirement and the expensive 8224 chip.

AMD made the 9080, reverse engineered and not licensed, which was designed to be clockable at a faster speed than the 8080a (although Intel also made faster 8080's like the -A-1, -A-2, and -B variants). Incidentally, the Intel-AMD technology agreements made in the 70's haunted both companies till the 1990's in various lawsuits. AMD, of course, makes Intel software-compatible CPU's to this day, without any help from Intel.

An important designer at Intel, Fredrico Faggin, left Intel to start Zilog, a company that produced the Z-80. This chip was faster than the 8080, both in clock speed and in execution times per clock cycle per instruction, instruction-compatible (but not pin-compatible), and included support for dynamic RAM and complex, powerful new instructions - although in my opinion they weren't that useful in terms of speed, they did help to make code more compact to write. It was also aggressively priced. Furthermore, it required only +5 volts to operate. Many late 70's computers used this chip instead of the 8080; many "Altair" type S-100 machines had their 8080 boards replaced with Z-80's, and this is one of the reasons that working 8080's are now rare. Zilog, too, still exists today; after a promising start, they failed time and again to produce new products as promised.. This actually burned an employer of mine in the early 1980's. They could have been where Intel is now in the business world had they followed through with more timely successes.

Intel released the 8085 a few years after the 8080; it was slightly faster, required only +5 volts to run, and had some new instructions useful in serial communication. This chip also contained its own clock generator, making it cheaper in terms of power, board space, and chip cost, although it still wasn't as fast as the Z-80. This marked the end of the 8080 era for new designs in the US, although I recall that several "trainer" computers utilized it until the mid-1980's. I don't believe that the 8080 microprocessor has been made by anybody in years; in fact, modern production facilities might not be able to produce it even if there was a market for it.