Understanding Microprocessors in PinballOriginally published on April 22nd, 1996 in the Usenet newsgroup: Rec.Games.Pinball
(revised January 2003)
The mere mention the words "dead CPU board" or "microprocessor" is enough to have some people running for cover. For those not familar with this area of technology, repairing anything with a microprocessor can seem can seem a daunting and frightening task. If you are one of those afflicted with this fear then I hope this article will go some way to alleviate those fears.
In this article we will be cover 8-bit microprocessors from early, pre 1982, solid state pinball machines. The Rockwell 6502 and Motorola 6800, 6802 and 6808 were the processors of the day.
Microprocessor Usage Table
All the microprocessor listed above have the same basic architecture. They are have three busses: A 16-bit 'address' bus, an 8-bit 'data' bus and a 'control' bus. All the processors use NMOS (N-channel Metal Oxide Silicon) technology.
The 6802 is identical to the 6808 with the exception that the 02 contains 128 bytes of internal RAM. To use this internal RAM, the RE (RAM Enable) pin of the processor needs to be held high. If you wish to test your processor you can place a 6808 into a socket previously occupied by a 6802 with no ill effects. However, to do this you must ensure RE is connected to zero volts and a 6810 RAM IC is placed into the socket provided.
The Address Bus
All 16 lines of the address bus (A0-A15) are omni-directional. Quite simply, this means that data on this bus flows in only one direction, -- out of the processor. These lines aer used to 'address' devices on the bus': RAM, ROM, PIAs etc. The number of address line the microprocessor outputs determines the size of the area it can address, for all those listed above, this is 65336 bytes or ' 64kb'. In digital binary terms, one thousand equals 1024 and not 1000.
The Data Bus
Data flows between the microprocessor and the devices connected to it (RAM, ROM...) via this 8-line bidirectional bus. The microprocessor accepts data (in) from the bus during a 'read' cycle and writes data (out) on the bus during a 'write' cycle. The size of the data bus actually denotes the size specified for the microprocessor. All the above are designated as '8-bit' microprocessors. They are however, all capable of writing 16-bits of data. This is achieved by writing two 8-bit bytes one after the other. The bytes are divided into 'high' and 'low' bytes. The low byte would contain bits 0-7 and the upper bit 8-15.
The Control Bus
There are various signals here associated with each microprocessor but the specific signals we are interested in are listed below. (Note that an asterisk (*) next to the signal name indicates that the signal is active when low/zero volts):
Not really part of the control bus but the most important pin outside the power and ground pins. This input recieves a TTL clock pulse from an external crystal oscillator, in most cases. Early Bally CPUs did not use a crystal, instead a multivibrator circuit was used. The clock is extremely important as all bus timing events and processing is derived directly from this signal.
Used as you would expect, to 'reset' the microprocessor. This pin is an input and in all cases the line must be held low for a predetermined time period as power is applied. This allows the processor to stabilise the internal circuits correctly as the power reaches the correct level. Being an input, the processor cannot assert reset to other devices on the bus. The reset line is also common among other ICs such as the 6820/1 PIA (Peripheral Interface Adaptor).
This control signal is output by the microprocessor whenever data flows. If the microprocessor is performing a 'read' cycle, it will assert this pin high to inform devices on the bus that it requires to read data from the address specified by the address lines. If the microprocessor wants to write data, it asserts this line low.
Interrupts: IRQ* and NMI*
Interrupts are exceptional conditions. They serve to interrupt the microprocessors from it's current task (for example, running the main program). The microprocessor then jumps to a unique location where it runs a program or 'service routine' associated with the interrupt type. When the routine has been run and is complete, the microprocessor returns to the main program and continues where it left off.
There are two different types of interrupt: IRQ refers to 'Interrupt ReQuest' and NMI refers to 'Non-Maskable Interrupt'. IRQ is a masked interrupt. Simply put, this means that when an IRQ interrupt has been recognised by the microprocessor, it will not allow another IRQ (it will ignore further IRQ requests) until it has finished running the current IRQ service routine. NMI is a non-maskable interrupt (hence it's name) and another NMI will be recognised as valid and acted upon even if the processor is already running an NMI service routine. This effectively interrupts the current interrupt to run another.
VMA: Valid Memory Address
Anytime the microprocessor placed an address on the bus, this signal is asserted high. It's use depends on the hardware design engineer but commonly it is used to aid the selection of devices on the bus (memory address decode logic, bus drivers etc.).
How It All WorksWhen any of the microprocessors mentioned above power-up or receive reset they go through a boot-up routine know as 'bootstrapping'. The microprocessor outputs a fixed address after the reset line goes high which selects the location or 'reset vector' which in turn my contain the first two bytes that form the start address of the main program. The reset vector always reflects and EPROM or masked ROM location on the pinball CPU boards we are refering to. This automatic address function is built into the microprocessor as part of it's internal achitecture and cannot be altered. The microprocesor also asserts read/write high to indicate it is performing a read cycle on the bus. The microprocessor loads the contents of the reset vector into an internal counter known as the 'program counter'. It then treats the contents of the program counter as the address of the main program and makes a jump to that address to execute the first instruction. The program is now effectively running.
After power-up, the Gottlieb System 80 6502 microprocessor issues the address $FFFC onto the address bus and asserts the read/write line high to signal a read cycle on the bus. Consequently, address $FFFC gives up it's contents and places them onto the data bus. The 6502 then reads the byte of data of the address bus and stores it as the low byte (bits 0-7) of the start address in the program counter. The 6502 then issues $FFFD onto the address bus and reads this byte on the data bus. It then places this byte into the program counter as the high byte of the start address (bits 8-15). So, if the contents of $FFFC was $00 and the contents of $FFFD was $80, the program counter would loaded with $8000 and the processor would jump to that address for it's first instruction. Remember that in the case of the 6502, the low byte is asked for first ($FFFC, bits 0-7)) followed by the high byte ($FFFD bits 8-15). Motorola flipped this and fetched the high byte first followed by the low byte. The address of the reset vector is also different but this is of no real importance unless you are actually programming the systems.
By the way, the program counter is simply a 16-bit counter that keeps track of the program address.
An area of RAM needs to be reserved for something called the 'stack'. The stack is needed by the microprocessor for storing information, in particular, address's. Whenever the processor is interrupted, it needs to remember what address it was at before the interruption occurred. It does this by placing the current address plus onto the stack. Plus one because it will finish the current instruction and will want to return to the next instruction to carry on. It retrieves this address when the interrupt is over so that it may carry on as before. The processor also saves the contents of it's internal registers onto the stack before processing interrupts.