Data Processing Protocol System

This invention is related to a protocol system wherein each station connected to a data bus of a data processing system must calculate its time to transmit turn during which it may begin transmitting data on the bus. This data processing equipment may be used for building automation, monitoring and control for fire and security management, air-conditioning control, or the like.

Prior art data processing systems, for example those systems used for controlling or supervising the environment or fire and security points within a building, having a central panel for providing alarm or status indications of the various points within a building, originally provided a separate wire running from the panel to each supervised point. In today's large buildings, however, the cost of wiring in this manner can be prohibitive. Therefore, the prior art replaced these plurality of wires by a single transmission channel running from the central processing unit to the remote points being monitored.

The central processing unit is capable of scanning the status of the various remote points and controlling various operations at the remote points by addressing, in turn, each remote point and requiring each remote point to take some action. This action may be to report back its present status, such as temperaure, pressure or humidity, or to perform some operation such as starting or stopping a fan, opening a damper or the like. By using one or more of these multiplexed transmission channels, the cost of installation of these monitoring systems is greatly reduced.

In such systems, the central processing unit typically consisted of a computer connected to one or more such channels running to the various remote points in a building. The size of the computer which was used in the central processing unit depends upon the number of remote points to be monitored and controlled. Because of flexibility and convenience, it has become necessary to configure such systems in such a manner that they can be expanded and can be interconnected for complex, large or multiple building facilities.

The present system comprises a plurality of remote stations each connected to a data bus. Each remote station may have a microprocessor or other type of computer connected through various circuits to the data bus and also connected to peripheral devices such as displays, supervised and control points, or the like.

Since the remote stations connected to the data bus may be physically separated by several miles, data communications between them is in serial form to permit transmission over common and inexpensive mediums or buses, such as phone lines and triaxial cables. Serial data transmission on a "party line" requires a protocol to prevent more than one device from transmitting at a time and to identify the transmitting device. This communication protocol is accomplished through a disclosed Transmission Protocol residing in each remote station.

A bus controller is not necessary since each remote station is capable of transmitting and receiving data independently of a controller. However, a bus monitor is utilized to make sure that the remote stations are transmitting in their proper turns. Functionally, each remote station continuously monitors all "party line" bus traffic. Based on present and past transmissions, each module "knows" its turn or time slot to transmit and transmits only at that time.

A transceiver circuit is disclosed for use in a data processing system for converting between the non-return to zero signals of a data processing unit and the pulse-no pulse signals which are transmitted on a communication bus having a triggering circuit responsive to the leading edge of a non-return to zero signal to begin supplying triggering pulses of a predetermined frequency, a gate circuit for passing through the triggering pulses as long as the non-return to zero signal is at a predetermined level and a pulse-no pulse circuit responsive to the triggering pulses for supplying pulse-no pulses to the communication bus. The receiving portion of the transceiver circuit likewise converts the pulse-no pulse signals on the communication bus to non-return to zero signals for supplying to the data processing unit.

These and other features and advantages of the invention will become apparent when taken into consideration with the drawings in which:

FIG. 1 is a generalized block diagram of the present invention showing two remote stations connected to a common data bus;

FIG. 2 shows a typical message, comprising a plurality of words, which can be transmitted from one station to another;

FIG. 3 shows the typical construction for each word in the message of FIG. 2;

FIG. 4 shows the connecting lines between the various elements of one of the remote stations shown in FIG. 1;

FIGS. 5A-5D show in detail the interface circuit shown in FIG. 1 and 4;

FIGS. 6A-6E show the transceiver circuit shown in block form in FIGS. 1 and 4;

FIGS. 6F-6H show the timing diagrams for the transceiver circuit shown in 6A-6E;

FIG. 7 shows the general software routines which are used by the microprocessor of FIGS. 1 and 4 to control the protocol hardware;

FIGS. 8A-8C show the quiet interrupt routine;

FIGS. 9A-9D show the receive interrupt routine;

FIG. 10 shows the transmit interrupt routine; and,

FIG. 11 shows the routine used for forming the check word shown in FIG. 2.

In FIG. 1, Station I comprises microprocessor 10 connected to common data or communication bus 20 through interface circuit 11 and transceiver circuit 12. In addition, the microprocessor may also be connected to peripheral devices 13 and 14. These peripheral devices may take the form of CRT or printer displays, supervisory or control points in building automation or fire and security systems, or the like.

Station II is also connected to common data bus 20 and comprises microprocessor 30 connected to data bus 20 through interface circuit 31 and transceiver circuit 32. In addition, microprocessor 30 may be connected to peripheral devices 33 and 34.

Based on present and past transmissions, each station on the common communication bus "knows" its turn or sequential time slot during which it can transmit data, and transmits only at that precise turn or time slot. Thus, no two remote stations can talk on the bus at the same time unless a system error has occurred. If a remote station does not transmit when its turn to transmit occurs, a system error results. A remote station's individual time slot during which only it can transmit is calculated based on the sender's address of the remote station presently transmitting. Thus, a time slot does not depend upon time so much as it does upon priority. After each transmission is begun, each remote station recalculates its own time slot based upon the present sender's address and its own address. The number of total time slots for this system is fixed but does not have to equal the number of remote stations connected to the data bus. For example, in the present system the total number of time slots may be 127 and the maximum number of remote stations may be 99. Thus, a remote station may have assigned to it a real time slot (1-99) during which it can transmit any type of message and a pseudo time slot (100-127) during which it may transmit specified types of messages only.

Each station connected to the data bus contains its own precision oscillator which controls the station's Baud rate. Since the Baud rate for each station is the same, each station is synchronized with each other within one bit width.

After a station transmits and assuming it is not assigned to a pseudo time slot, it sets a countdown counter to 256 which is twice the 127 slots plus 2. The extra 2 counts results in a count of 4 being necessary to switch from time slot 127 to time slot 1 since there is no time slot 0. After station 48 transmits its address, station 50 calculates its time to transmit count by subtracting 48 from 50 and multiplying by 2, or 4. Each station also generates interrupts after predetermined lengths of time started after no data is on the data bus. If station 49 fails to transmit, station 50 will transmit after 4, [2(50-48)], interrupts. If station 49 transmits, station 50 now sets its counter at 2(50-49) or 2. After two quiet interrupts decrementing its counter to 0, station 50 may now transmit. Once station 50 transmits, it sets its counter to 256; but if it has a pseudo time slot, e.g. 101, it sets its counter at 2(50-101) or 102, calculations being done in the binary system. A station which has assigned to it both a real time slot and a pseudo time slot will have a real address and a pseudo address so that it may put into its countdown count a count which depends upon the address of the last transmitter, and its own real address or pseudo address depending upon whether its real time slot or pseudo time slot is coming up next.

FIG. 2 shows the message format for the present protocol system. Each message which is transmitted on the data bus must comprise at least two words, the control word and the sender's address. If more than two words are transmitted by a remote station, the message must then comprise a message length word which indicates the total number of words in a message, the data words and finally a check word. The check word must satisfy the algorithm shown in FIG. 11. As each word is transmitted, the algorithm of FIG. 11 is performed. After all words except the check word have been transmitted, the last or remaining calculated result from the algorithm is complemented and sent out as the check word. When the check word algorithm has been performed and is satisfied, the number or words which were used by the algorithm is compared to the number of words indicated in the message length word. If there is no agreement, a transmission error has occurred and appropriate action is taken. If there is agreement, an end of message has been detected and there has been no error.

Every station must transmit a message. If the station has no data to transmit, then it transmits an "I'm OK" message comprising the control word and its address. If a message is not transmitted by a station during its time slot, an error has occurred. A bus monitor is used in the system and it is the function of the bus monitor to detect these kinds of errors. Each remote station is capable of transmitting a global message which is a message directed to all other stations. Each station is also capable of transmitting direct messages to a specific station. The control word indicates the type of message which is being sent. In a two word "I'm OK" message, the two words must satisfy the check words algorithm even though a check word is not transmitted. Satisfying the check word algorithm is obtained by appropriately constructing the control word.

Each word in a message takes the format of FIG. 3. It is comprised of a start bit, eight data bits, a parity bit and a stop bit.

FIG. 4 shows a typical remote station such as Stations I and II of FIG. 1. In FIG. 4, the station comprises microprocessor 10 connected to interface 11 which in turn is connected to transceiver 12 and then to data bus 20. Microprocessor 10 and interface 11 are interconnected by eight data lines DOB-D7B. Interface 11 is also connected to microprocessing unit 10 by an address bus comprising eight address lines, A0B-A7B. Microprocessing unit 10 contains a system clock, not shown, which generates the transmit clock signal, XMIT CLK, to interfact 11 and transceiver 12 and PH2T and IOEN signals to interface 11. Also, microprocessing unit 10 generates a read-write signal, RDWR, to interface 11 and a RESET B signal, which is a reset signal generated upon power up, to both interface 11 and transceiver 12. Interface 11 generates interrupt signals IRQ and NMI to microprocessing unit 10. Data which is received from bus 20 by transceiver 12 is transmitted to interface 11 over receive data line, RCV DATA, which is then transmitted by interface 11 to microprocessing unit 10 over data bus lines D0B-D7B. Data to be transmitted from a station to another station on data bus 20 is transmitted by microprocessing unit 10 over data bus D0B-D7B to interface circuit 11 in parallel form and then by interface 11 to transceiver 12 over the MXIT DATA line in serial form and then on to bus 20. If redundant buses are used, the microprocessor through its address bus A0B-A7B will select through interface 11 the specific channel which it wishes to transmit data over. Consequently, a CHANNEL SELECT line runs from interface 11 to transceiver 12. When microprocessing unit 10 wishes to transmit data over the bus 20, it uses the address bus A0B-A7B to interface 11 and then the REQUEST TO SEND line from interface 11 to transceiver 12 to initiate the transceiver 12 to begin transmitting the data from interface 11 to bus 20.

As shown in FIG. 1, the microprocessor is connected not only to interface 11 but to peripheral devices 13 and 14. Thus, each device, i.e., interface 11 and peripherals 13 and 14, has a unique address assigned to it. When the microprocessing unit wishes to transmit data over bus 20, it addresses the interface 11 over the address bus A0B-A7B. Data between the microprocessing unit 10 and interface 11 is transmitted in parallel form over data bus D0B-D7B. Data is transmitted between interface 11 and transceiver 12 in serial form over either line RCV DATA or XMIT DATA. Data is transmitted over the data bus 20 in serial form.

Interface 11 is shown in more detail in FIGS. 5A-5D. FIGS. 5A-5D are interconnected at points designated by a letter number such as D7 or F1. In addition, the terminals which connect interface 11 to microprocessing unit 10 and to transceiver 12 use the symbology shown in FIG. 4.

The interface circuit has an address known to the microprocessor and may be addressed by the microprocessor over address lines A0B-A7B shown in FIG. 5A. These address lines are connected through buffer-inverters 41 to various inputs of NAND gate comparators 42 and 43. Specifically, each address line is connected to each NAND gate 42 and 43. In addition to buffer-inverter 41, address line A7B is connected through inverter 44 to both NAND gates 42 and 43. With respect to address line A3B, the output of buffer-inverter 41 is connected to NAND gate 43 directly and is also connected through inverter 45 to NAND gate 42. In addition to buffer-inverters 41, address lines A1B and A2B are connected through EXCLUSIVE OR gates 46 and 47 to each NAND gate 42 and 43. Each EXCLUSIVE OR gate 46 and 47 has a second input connected to a positive voltage bias. Address line A0B is not connected to NAND gates 42 and 43 but, instead, is directly connected to the RS terminal of an asynchronous communication interface adaptor (ACIA) 50 which may be of the type MC6850 manufactured by Motorola. The pin numbers of ACIA 50 have been shown to show how it is connected in the circuit. ACIA 50 is a register which is conditioned to send or receive data by being appropriately addressed by the microprocessing unit over address lines A0B-A7B. In addition, clock line IOEN is connected through a buffer-inverter 41 to both NAND gates 42 and 43. The arrangement of inverters 44 and 45 together with EXCLUSIVE OR gates 46 and 47 determine the particular address of interface 11 shown in FIGS. 5A-5D.

FIG. 5B shows the manner in which the data terminals D0-D7 of ACIA 50 can be connected to the data bus lines D0B-D7B. Data can be supplied to ACIA 50 over data bus lines D0B-D7B and then to data terminals D0-D7 of ACIA 50 for transmission over terminal F8 and then to the transceiver circuit 12 or may be received from transceiver circuit 12 at intput terminal F7 of ACIA 50 for supply over terminals D0-D7 to data bus lines D0B-D7B and then to microprocessing unit 10. Data bus lines D0B-D7B and data terminals D0-D7 of ACIA 50 are interconnected by tri-state buffers 51 and 52 which are bi-directional devices having the type number 8T26. These devices 51 and 52 are conditioned to allow data to pass therethrough in either direction dependent upon the state of AND gate 53 which receives a input at F1 from notted input OR gate 54 which generates an output at F1 whenever interface 11 receives an appropriate address signal from microprocessing unit 10. In addition, AND gate 53 must receive a signal from terminal F4 which receives the PH2T clock signal from microprocessing unit 10 and terminal F6 which receives a read-write signal over line RDWR from microprocessing unit 10. Thus, if data is to be transmitted from the microprocessor unit to the data bus 20, tri-state buffers 51 and 52 are conditioned to allow data to pass from data lines D0B-D7B to terminals D0-D7 of ACIA 50. At the same time, the microprocessing unit conditions ACIA 50 to receive this data and transmit it over terminal F8. If data is to be received from the transmission line and routed to the miroprocessing unit 10, the microprocessing unit 10 supplies the appropriate address to interface 11 which conditions ACIA 50 to receive the data over line F7 and transmit it to terminals D0-D7. At the same time, tri-state buffers 51 and 52 are conditioned by microprocessing unit 10 to allow the data at terminals D0-D7 of ACIA 50 to be supplied to data bus lines D0B-D7B.

FIG. 5B also shows a bank of switches S1-S7 which establish the address for the particular interface 11 shown in FIG. 4. A closed switch indicates a high level or "one" and an open switch indicates a low level or "zero." These switches are connected through a series of tri-state buffers 60 shown connected to all of the switches S1-S7 plus an additional tri-state buffer 60 shown in FIG. 5C connected to terminal D7 of ACIA 50 and to NMI flip-flop 61. The tri-state buffers 60 are connected to terminals D0-D7 of ACIA 50 and are multiplexed along with ACIA data lines D0-D7 to drive buffers 51 and 52 shown in FIG. 5B. When NMI flip-flop 61 is set, a 1 appears and is read at terminal D7 indicating that the device is requesting an NMI interrupt. The reading of this bit will automatically reset NMI flip-flop 61 as discussed hereinafter. Thus, the microprocessing unit will know when it receives an interrupt and, because of switches S1-S7, the identity of the device which has issued the interrupt.

Buffers 60 are conditioned to transmit the address and NMI interrupts to data terminals D0-D7 by NAND gate 62 shown in FIG. 5B. The output of NAND gate 62 is connected directly to buffers 60 shown in FIG. 5B and through terminal G2 to buffer 60 shown in FIG. 5C. In order for NAND gate 62 to condition buffers 60, NAND gate 62 receives inputs from terminal F2 which is an address decode terminal, terminal F4 which is the PH2T clock terminal and terminal F6 which is the RDWR input terminal.

On Line flip-flop 70 shown in FIG. 5D determines whether data can be transmitted between interface 11 and transceiver 12 over RCV DATA and XMIT DATA line. If On Line flip-flop 70 is in its reset state, its Q output is low which precludes the XMIT DATA NAND gate 71 and the RCV DATA NAND gate 72 from allowing the passage of data therethrough. With a zero to the input of NAND gate 72, its output is at a logic level 1 which conditions AND gate 73 to recirculate data from ACIA transmit terminal F8 through AND gate 73 and to universal asynchronous receiver transmiter 74 which may be a Harris 6402 device. Also, with On Line flip-flop 70 in its reset condition, its Q terminal is at a logic level one which conditions NAND gate 75 to allow the data to pass from transmit terminal F8 of ACIA 50 through NAND gate 75 to the input of NAND gate 76. Since one input of NAND gate 77 is connected to the Q terminal of On Line flip-flop 70, that input terminal will be a logic level 0 which dictates that its output is a logic level 1 which conditions NAND gate 76 to pass the data coming from ACIA output terminal F8 through NAND gate 75 to terminal F7 which is connected back to the input or data receive terminal F7 of ACIA 50. In this way, the software of the microprocessing unit can, at any time, disconnect the station from the data bus and run a diagnostic program to check the transmit and receive functions of the ACIA as well as the function of the Bit Time Interrupt Generator 80 to be discussed hereinafter. Terminal C of On Line flip-flop 70 is the reset terminal for this flip-flop as long as terminal S is connected to a high input which, as can be seen in FIG. 5D, it always is. Thus, with terminal C low, On Line flip-flop 70 is in a condition where its Q output is at a logic level 0 and its Q output is at a logic level 1 regardless of the D input and the clock input. When the C terminal goes high, the clock input will set the Q output terminal at whatever logic level happens to be on the D input terminal. On Line flip-flop 70 is reset at terminal C through NAND gate 78 which receives a reset signal from terminal G7 which, as shown in FIG. 5C, is connected to the output of an inverter from terminal G4, the RESET B input terminal shown in FIG. 5B. Thus upon power up, flip-flop 70 is reset. At other times it is allowed to be clocked at terminal F3 except when it receives an input at G8 from WATCHDOG TIMER 79 shown in FIG. 5C to be discussed hereinafter. Thus, the address decode terminal F3 clocks On Line flip-flop 70 to control the Q output terminal at a logic level dependent upon the D1 input terminal which is received from the microprocessing unit. When the microprocessing unit desires that interface 11 should be operated in a non-diagnostic mode such that information can be transferred between the bus 20 and microprocessing unit 10, a logic level 1 is supplied to the D input of On Line flip-flop 70 and then it is clocked by addressing code terminal F3 to force the Q output terminal to a logic level 1 which conditions NAND gate 71 to transmit data and NAND gate 72 to receive data.

On Line flip-flop may be device 74LS74 supplied by Fairchild. When the Q output of On Line flip-flop 70 is high, it conditions NAND gate 81 to supply a REQUEST TO SEND signal from terminal F9 ACIA 50 to transceiver 12.

Data is received on the RCV DATA line shown in FIG. 5D and is supplied through NAND gate 72 when the input of that NAND gate which is connected to the Q output of On Line flip-flop 70 is high. This data is supplied through AND gate 73 to the input to Universal Asynchronous Receiver Transmitter 74 and is clocked into that device by the clock terminal connected to G6 which, as shown in FIG. 5A, is the transmit clock line XMIT CLK. At the same time, the intput of NAND gate 75, which is connected to the Q terminal of On Line flip-flop 70, is at a logic level 0 which means its output is at a logic level 1 conditioning NAND gate 76 to supply data from NAND gates 72 and 77 to the input terminal F7 to ACIA 50. This data is clocked into the registers of ACIA 50 by the transmit clock line XMIT CLK. Similarly, when data is to be supplied to the transceiver board over terminal XMIT DATA, On Line flip-flop 70 conditions NAND gate 71 to supply this data from the output terminal F8 of ACIA 50 under control of the transmit clock line XMIT CLK connected to the ACIA 50.

The purpose of the Bit Time Interrupt Generator 80 is to generate NMI interrupt signals in a fixed repeating period when there is no transmission present on the data bus. Specifically, if, because of a failure at a station which is suppose to transmit during a particular time slot, interface 11 receives no message for a predetermined length of time, then all remote stations will receive interrupt signals from their Bit Time Interrupt Generators which are counted by the system software to determine the time slot when each station in its turn will transmit. In this way, the failure of one station to transmit does not terminate the transmission of data by all other stations.

Bit Time Interrupt Generator 80 comprises a divide by 16 counter 91 and a counter 92 both of which may be devices 74LS161. Generator 80 also comprises a Set NMI flip-flop 93 which may be of the same type as the flip-flop 70. Flip-flop 93 is connected such that its S terminal connected from NAND gate 94, connected as an inverter, is normally high such that the input from the XMIT CLK terminal G6 connected to terminal C has no effect on the Q and Q outputs. As soon as the input to inverter 94 goes high so that the S terminal flip-flop 93 goes low, the next XMIT CLK signal will change the state of the Q and Q outputs.

Counter 91 divides the XMIT CLK signal by 16 which essentially establishes the Baud rate for the system. Thus, a bit width is defined as 16 pulses of the transmit clock. Counter 92 divides the output from counter 91 by 8 to establish a time slice which is defined as 8-bit widths. Counters 91 and 92 are enabled to count clock pulses from the transmit clock terminal G6 through line 95 only when device 74 has received a complete data word. Thus, Bit Time Interrupt Generator 80 begins counting each time a word has been completely received by device 74 over the data bus 20 and from transceiver 12. AND gate 96 clears counters 91 and 92 when a new word is received through AND gate 73. Thus, as long as traffic exists on the data bus, AND gate 96 will prevent counters 91 and 92 and flip-flop 93 from generating an interrupt signal. When data is not on the data line, AND gate 96 will not clear counters 91 and 92 and, when counter 92 reaches a count of 8 (the predetermined length of time discussed above), flip-flop 93 will set upon receipt of the next clock signal from terminal G6 to switch its Q and Q outputs. When flip-flop 93 is set, counters 91 and 92 are cleared through AND gate 96 and a signal is supplied to terminal G5 which sets flip-flop 61 through NAND gate 97. When counter 92 is reset, the next clock pulse from terminal G6 will reset flip-flop 93. When NMI flip-flop 91 is set, an NMI interrupt is generated through notted input OR gate 98 of FIG. 5C and NAND gate 99. Line NMI is connected to the microprocessor. This interrupt signal, called a quiet interrupt because it was generated by Bit Time Interrupt Generator 80, is then used by the microprocessor in calculating its own time slot. Counters 91 and 92 will begin counting again and will supply further interrupts until traffic resumes on the data bus. Counters 91 and 92 may also be cleared by the RESET B inverted signal supplied through NAND gate 96 by terminal G7.

Watchdog Timer 79 is comprised of retriggerable one-shot multivibrators 100 and 101. Each device may be a 74LS123 with device 100 producing an output five seconds after it receives its input. Multivibrator 101 produces an output 500 nanoseconds after it receives an input from multivibrator 100. Multivibrator 100 is triggered through a combination of an input from terminal F3, which is a composite of the address decode circuit shown in FIG. 5A, the RDWR terminal and a PH2T clock signal, and a signal from terminal D2 of ACIA 50. Thus, each time the interface is properly addressed by the microprocessing software, multivibrator 100 is retriggered. Should a fault in the system occur such that communication between the microprocessing unit and the interface circuit has ceased, Watchdog Timer 79 times out and provides an output through NAND gate 102, inverter 103 and inverter 104 to the IRQ output which is an interrupt supplied to the microprocessing unit. In addition, an output is supplied by multivibrator 101 to terminal G8 which is connected through AND gate 78 to On Line flip-flop 70 of FIG. 5D for resetting this flip-flop and taking the interface apparatus off line.

The output from multivibrator 100 to NAND gate 102 can be masked by Watchdog Timer Mask flip-flop 105 which may be of the same type as On Line flip-flop 70. When the microprocessor wishes to prevent the Watchdog Timer from supplying an interrupt IRQ signal, it places a logic level 1 on the data bus D3B which is connected through a tri-state buffer to terminal D3 of flip-flop 105. A signal from terminal F3 will clock flip-flop 105 to drive the Q output low which precludes the output from multivibrator 100 from passing through NAND gate 102.

Channel Select flip-flop 106, which may be of the same type as On Line flip-flop 70, is configured such that a clock input will invert the D input on the Q output. Thus, the microprocessing unit can select which of two channels it desired to transmit data over by placing the appropriate signal on data bus line DOB which is connected through terminal DO to Channel Select flip-flop 106.

Finally, flip-flop 107, which may be of the same type as On Line flip-flop 70, is configured as an NMI mask flip-flop and can be utilized by the microprocessing unit for prohibiting NMI interrupt signals from being generated by the interface circuit to the microprocessing unit. The Q output from NMI mask flip-flop 107 is normally high which conditions NAND gate 99 to pass through to the NMI interrupt line any signal from NMI flip-flop 61. When an interrupt signal is generated by Bit Time Interrupt Generator 80 at terminal G5, it is connected through NAND gate 97 to set NMI flip-flop 61 at terminal S. This causes the Q terminal of NMI flip-flop 61 to go low which, because of the notted input terminal of OR gate 98, raises the output of OR gate 98 and produces an NMI interrupt on the NMI line. If the microprocessing unit decides to mask the NMI interrupt line, it provides an appropriate signal over data bus line D7B which is connected through terminal D7 to the input of NMI Mask flip-flop 107. Thus, upon the next clock signal, the Q output from flip-flop 107 goes low which prevents any signal passing through either NAND gate 97 or NAND gate 99.

When the microprocessing unit wishes to transmit data out on the data bus, it supplies data on data bus lines DOB-D7B which are entered into ACIA 50. When ACIA has received a complete word, it generates an IRQ output through notted input OR gate 98 to provide an NMI interrupt, called a transmit interrupt because of the transmit mode, informing the microprocessing unit that the ACIA is now in condition for transmitting data. The microprocessor unit will then address ACIA 50 and supply transmit clock signals to it to transmit the data out of the ACIA line at terminal F8.

When ACIA 50 has received data at its input terminal F7, an interrupt is generated at G1 indicating that a complete word has been received in ACIA 50. This interrupt, called a receive interrupt, is then connected through OR gate 98 to NMI interrupt line which is received by the microprocessing unit to then read the data stored in ACIA 50.

Therefore, three types of interrupts are generated over line NMI. A quiet interrupt which is generated by the Bit Time Interrupt Generator when no messages have been received by the interface for a predetermined length of time or count, a receive interrupt when data is being received and a transmit interrupt when data is to be transmitted from ACIA 50. Microprocessor 10 is capable of deciphering these three interrupts. Thus, if it is transmitting data and receives an interrupt, it knows that the interrupt is a transmit interrupt. To distinguish between a quiet interrupt generated by the Bit Time Interrupt Generator and a receive interrupt generated by ACIA 50, it scans the data bus lines DOB-D7B. If no data appears on the line, the interrupt is a quiet interrupt. If data does appear on the line, the interrupt is a receive interrupt. These interrupts and the ability of the microprocessing unit to distinguish between the types of interrupts it receives become important when considering the way in which the microprocessing unit determines its time slot for transmitting data.

The transceiver circuit which connects interface 11 to data bus 20 is shown in more detail in FIGS. 6A-6E. Again, the interconnection between these figures is shown by circles letter-number reference numerals such as H1 and K3. The connection between the transceiver circuit and the data bus is shown in FIG. 6B and the connection between transceiver circuit 12 and interface circuit 11 is shown by appropriate terminals in FIGS. 6A and 6E.

The primary functions performed by the transceiver circuit are to receive and transmit data in serial form between interface 11 and data bus 20 and to convert the non-return to zero level (NRZ-L) serial data received from the ACIA of the interface circuit to pulse-no pulse (PNP) serial data form which is used on the data line. FIG. 6G shows the difference between the form of the data transmitted between transceiver 12 and interface 11 and the data transmitted between transceiver 12 and data bus 20. The non-return to zero level signal is shown by the NRZ-L diagram of FIG. 6G. A high level indicates a data bit 0 and a low level indicates a data bit 1. This diagram shows a 1 start bit and then 8 data bits having various values. Data bits 1 and 2 are 0 bits, data bit 3 is a 1 data bit, data bit 4 is a 0 data bit, data bits 5 and 6 are 1 data bits, data bit 7 is a 0 data bit, and data bit 8 is a 1 data bit. When the transceiver circuit 12 is transmitting data from microprocessing unit 10 to data bus 20, it receives the NRZ-L data and converts it to PNP data. Thus, pulses are supplied to the data bus for the start bit and data bits 3, 5, 6 and 8. Conversely, as data bits are received on data bus 20 into transceiver 12, transceiver 12 converts the pulse-no pulse data bits into NRZ-L data information.

When the microprocessing unit 10 wishes to transmit data on data bus 20, it supplies data to the interface circuit as discussed above and also transmits a signal out through the interface circuit to be placed upon the channel select input line in FIG. 6A. This signal is supplied through a circuit 110 which comprises an inverter 111, a buffer-inverter 112 and a NAND circuit 113 connected as shown. A pair of straps 114 and 115 are also included in this circuit. If two data buses are used for redundancy, two transceivers are required. One of the straps 114 and 115 is disconnected in the other transceiver board. Thus, if the channel select signal is low, one transceiver board is selected and, if the channel select signal is high, the other transceiver board is selected. When the proper channel select input is received for the transceiver circuit shown in FIGS. 6A-6E, an input will be supplied to the D input of Transmit flip-flop 116 to enable it into operation.

When the data to be transmitted over the data bus has been accepted in ACIA 50, a REQUEST TO SEND signal is generated by the interface circuit and supplied to the transceiver to enable NAND gate 117. When the XMIT DATA line in FIG. 6A falls low due to the start bit as shown in FIG. 6G, Transmit flip-flop 116, which may be of the same type as On Line flip-flop 70, is clocked which drives its Q output terminal high and its Q output terminal low. When the Q terminal of Transmit flip-flop 116 goes high, Position Counter 118 and Character Counter 119, which may be devices 74LS161, are enabled. The operation of this circuit may be followed with the aid of the timing charts in FIG. 6F.

Position Counter 118 counts transmit clock pulses on line XMIT CLK. Position Counter 118 divides the transmit clock signal by 16 to establish the Baud rate. Therefore, for each 16 transmit clock pulse, an output is placed on its output line CT for clocking Character Counter 119. Once Position Counter 118 has been enabled, the next clock pulse will step its output Q0-Q3 to a 1,0,0,0 condition which, because of the buffer-inverters, will provide an output from NAND gate 120, which enables NAND gate 121. Thus, there is a 1/16 bit delay between the receive of the start bit and the transmission of a PNP bit as shown in FIG. 6G. Once NAND gate 121 has been enabled, a 0 XMIT DATA bit will set PNP flip-flop 122, which may be of the same type as On Line flip-flop 70, to drive output terminal K1 high. As shown in FIG. 6B, terminal K1 is connected to AND gate 123. When AND gate 123 receives a pulse from terminal K1, it transmits the pulse through Bus Line Drive Circuit 124 to supply the bit to pulse transformer 125 and then on data bus 20. Terminal K3 is connected to AND gate 123 and is connected at its other end to the receive portion of the transceiver circuit to inhibit the transmitter part of the transceiver circuit when data is being received. However, when data is not being received, terminal K3 is high allowing the bits on terminal K1 to be transmitted through AND gate 123.

At the same time, when the Q output of PNP flip-flop 122, which may be of the same type as counters 118 and 119, goes high, PNP Counter 126 of FIG. 6A is enabled and begins to count pulses from a 768 KHz supply. PNP Counter 126 divides the 786 KHz signal by 16 and supplies an output through PNP Reset Circuit 127 which is supplied through NAND gate 128 to reset PNP flip-flop 122. Thus, 20 microseconds after PNP flip-flop 122 has been set, it will be reset resulting in an output pulse supplied to data bus 20 of 20 microsecond duration.

After Position Counter 118 has reached a count of 16, it will supply a pulse to Character Counter 119 to step this counter. The next transmit pulse, which is a 0 as shown by FIG. 6G, will inhibit NAND gate 121 from supplying a set signal to PNP flip-flop 122. Therefore, flip-flop 122 will not supply a pulse to the bus drive 124 and no pulse is supplied to the data bus. Since data bit 2 is also a 0, no pulse is applied to the data bus. When data bit 3 is transmitted, the XMIT DATA line drops low which enables NAND gate 121 to supply the next pulse from Position Counter 118 to set PNP flip-flop 122 to provide another 20 microsecond pulse to the data bus. This operation continues until Character Counter 119 has indicated that a complete word is transmitted at which time it supplies an output through NAND gate 129 to the Transmit Flip-Flop Reset Circuit 130. This circuit resets the Transmit flip-flop 116 which clears counters 118 and 119 and the transceiver circuit waits for a new word to be transmitted from the microprocessor through the interface circuit.

Data can be received by the transceiver from either direction on data bus 20. As shown in FIG. 6B, a pickup or pulse transformer 139 is connected to data bus 20 and to terminals H7, H8 and H9. As mentioned hereinabove, when data is being received over the transmission line, terminal K3 to NAND gate 123 in FIG. 6B and to PNP flip-flop 122 of FIG. 6A will prevent any data from being transmitted by the transceiver circuit. Furthermore, the stop bit is always a level 1 such that, when Character Counter 119 reaches a count of 10, which includes the start bit, 8 data bits and the parity bit, it resets Transmit flip-flop 116. As shown in FIG. 6F, the 0 start bit sets the transmit flip-flop which enables the Position and Character Counters to begin counting based upon the transmit clock. The start bit also sets the PNP flip-flop which in turn enables the PNP Counter. When the output of the PNP Counter reaches a count of 16, it provides an output pulse which is used to reset the PNP flip-flop. The Transmit flip-flop is set during the transmission of the entire word so that the Position and Character Counters remain enabled. The next 0 data bit to be supplied from the interface circuit will again set the PNP flip-flop which will be reset through the PNP Counter and Reset Circuit 127. When the parity bit is received from the interface circuit, the PNP flip-flop is set and supplies a 20 microsecond pulse to the Bus Driver. This parity bit is the tenth character and at the end of this character, Position Counter 118 supplies an output over line CT to Character Counter 119 which resets the Transmit flip-flop through Reset Circuit 130.

Terminals H7, H8 and H9 are shown in FIG. 6C connected to circuit 141. Circuit 141 receives pulses from the data bus and distinguishes between negative and positive signals. Specifically, data being transmitted in one direction over data bus 20 is considered a positive signal and data transmitted over the other direction on the data bus 20 is considered a negative signal. Circuit 141 responds to each of these signals. The amplifiers in circuit 141 are biased so that on positive pulses the top most amplifier will switch and for negative pulses the bottom most amplifier will switch. Assuming that positive pulses are coming in through terminal H7, these pulses are supplied through NAND gate 142 to the input of NAND gate 143. NAND gate 143 also receives inputs from terminal H6 which is the Q output of Transmitted flip-flop 116. Thus, if Transmit flip-flop 116 is in a transmit mode, its Q output will be low which will block pulses from being supplied by terminal H7 through NAND gate 143. However, during a receive mode, terminal H6 will be high enabling NAND gate 143. Also, NAND gate 143 receives an input through terminal H5 from the Channel Select circuit which will designate this transceiver circuit to receive data on the line. The fourth input to NAND gate 143 is derived from AND gate 144 which has one input from the Q terminal of flip-flop 145, which is normally a high level to enable NAND gate 144, and a second input from terminal H3 which is a normally high RESET B signal. Thus, the fourth input to NAND gate 143 is normally high. Since all of the inputs to NAND gate 143 are normally high except for the input connected from NAND gate 142, it is in a condition to pass through data bits supplied by terminal H7. Likewise, NAND gate 146 will be in a condition to supply negative pulses from terminal H9 through it. Notted input OR gate 149 will produce a positive going pulse at terminal K6 when either a positive pulse is received at terminal H7 or a negative pulse is received at H9. Similarly, notted input OR gate 150 will produce a positive going pulse to enable counter 151 whenever flip-flop 145 or flip-flop 152 has been set.

When notted input OR gate 149 has received an input from either the positive pulse detection circuit or the negative pulse detection circuit, it will supply the bit to terminal K6 to enable Half-bit Counter 160, which may be a 74LS161 counter, in FIG. 6D, to enable NAND gate 161 and to supply an input to Delay Shift Register 162, which may be a 74LS164 register. The purpose of the circuit shown in FIG. 6D is to detect that a valid bit has been received by the pulse transformer receiver circuit. The requirement of this circuit is that the bit (or pulses received) must be at least a half bit wide. When the first bit at terminal K6 enables counter 160 and NAND gate 161 and provides an input to register 162, counter 160 and register 162 begin to count the 768 KHz clock. When counter 160 has counted 8 of these clock signals indicating half a PNP pulse, it produces an output at terminal K4 which clocks the valid data bit through flip-flop 152 or flip-flop 145 depending upon whether a positive or negative data bit had been received. When either one of these flip-flops switches, an output is produced by OR gate 150 to enable counter 151, which may be a 74LS161 counter. Also, when a positive bit has been received, flip-flop 152 will switch to inhibit NAND gate 146 through AND gate 170 to prevent negative going pulses from going through the circuit. This is a safeguard against noise being generated in the other direction and being accepted by the transceiver as a valid data bit. On the other hand, if a negative data bit had been received, then flip-flop 145 switches to inhibit NAND gate 143 through AND gate 144 from allowing positive pulses being supplied through it. Also, the D input of flip-flop 172 receives the output from OR gate 171. OR gate 171 will pass through either a positive or negative pulse such that, when the output at K4 clocks flip-flop 172, an output is provided at terminal K3 which is used in FIGS. 6A and 6B at flip-flop 122 and AND gate 123, respectively, to prevent the transmission of pulses on the data line.

If a valid data bit has been received, half-bit counter 160 will time out and provide an output at K4. This output essentially then takes two paths. The first path is to the clock input of Valid Bit flip-flop 173 which may be a 74LS74 flip-flop. At the same time, the bit input to delay shift register 162 is being shifted through the same clock which is driving Half-bit Counter 160 and provides an output at Q7 at the same time Half-bit Counter 160 provides its output to the clock input of Valid Bit flip-flop 173. The Q output of Valid Bit flip-flop 173 and the Q7 output of Delay Shift Register 162 go high at the same time which causes NAND gate 174 to supply a low signal to terminal K7. Terminal K7 is connected to the NRZ-L flip-flop 175, which may be a 74LS74 flip-flop, shown in FIG. 6E which changes its Q output to supply the data bit over line RCV DATA. At the same time that NRZ-L flip-flop 175's Q output goes high, Bit Counter 176, which may be a 74LS161 counter, is enabled to begin counting inverted transmit clock pulses at terminal K5. After it reaches a count of 8, it supplies an output over Q3 to NAND gate 177 for providing a signal to terminal K8 which sets flip-flop 178 for providing an output through AND gate 179 to reset Valid Bit flip-flop 173. After counter 176 of FIG. 6E reaches a count of 16, the equivalent of one bit width of an NRZ-L bit, it provides an output at TC for setting the flip-flop 181 which provides an output through NAND gate 180 to reset NRZ-L flip-flop 175. When flip-flops 178 and 181 receive the next transmit clock signal at terminal H2, they are reset. In this manner, a 20 microsecond PNP data bit on the data bus is converted to the NRZ-L bit as shown in FIG. 6G.

When Half-bit Counter 160 provides its output at terminal K4, this output is connected to the clock terminals of flip-flops 152, 172 and 145 of FIG. 6C. Thus, the positive signal at terminal H7 will be clocked through flip-flop 152 or the negative signal at terminal H9 will be clocked through flip-flop 145 and either signal will be clocked through flip-flop 172. When flip-flop 145 or flip-flop 152 has been switched, an output is provided through OR gate 150 to enable Inhibit Counter 151. Inhibit Counter 151 begins counting transmit clock pulses and, at the count of 8, will set flip-flop 183. When flip-flop 183 sets, it provides an output through OR gate 184 to reset flip-flops 152, 172 and 145.

If the pulse at K6 does not last for at least 8 counts of the 768 KHz signal, Half-bit Counter 160 will automatically reset at the notted MR terminal. Also, the loss of the signal at terminal K6 provides an output through NAND gate 161 to reset Delay Shift Register 162. Since no output will be provided at terminal K4, flip-flops 152, 172 and 145 are not switched and no signal is supplied by the Valid Bit flip-flop 173 to the NRZ-L flip-flop 175 to prevent the supply of a pulse on the RCV DATA line.

It is to be noted that PNP Counter 126 on the transmitting portion of the circuit shown in FIG. 6A is counting the 768 KHz pulses until a total of 16 is reached to insure that the PNP pulse on the data bus is 20 microseconds wide. Half-bit Counter 160 counts only 8 of the 768 KHz pulses to insure that the PNP pulse received is at least 10 microseconds long. The timing chart for the received pulses is shown in FIG. 6H. When a 20 microsecond PNP pulse is received from the data bus, Half-bit Counter 160 and Delay Shift Register 162 provide their outputs for only 10 microseconds which is the last 10 microseconds of the received pulse. When Valid Bit flip-flop 173 is set, the output from Half-bit counter 160 clocks either flip-flop 152 or flip-flop 145 to provide an output through OR gate 150 to enable Inhibit Counter 151. These two signals last for 52 microseconds due to the output from Inhibit Counter 151 resetting flip-flops 152 and 145. It is to be noted that the signal to the set terminal of NRZ-L flip-flop 175 can only last ten microseconds since Delay Shift Register 162 is reset after ten microseconds to disable NAND gate 174.

FIG. 7 shows the generalized flow chart for the software which controls microprocessor 10 and a program listing for this flow chart is included as an appendix hereto. According to the flow chart of FIG. 7, the microprocessor waits for an interrupt from the interface circuit. Once it does receive an interrupt, the microprocessor must determine what type of interrupt it has received. As previously discussed the miroprocessor can easily determine if the interrupt is a transmit interrupt if it is in the transmit mode. If it is not a transmit interrupt and a look at the data lines indicates that there is no received data, the interrupt must be a quiet interrupt. If there is data, the interrupt must be a receive interrupt.

Therefore, according to FIG. 7, if the interrupt is a quiet interrupt, i.e., an interrupt generated by the Bit Time Interrupt Generator of the interface circuit, the program proceeds to the flow chart shown in FIGS. 8A-8C. Upon entering the quiet interrupt routine, a test is made to establish if the previous message was properly completed. It does this by looking at the last word received to see if it has its full complement of bits and that there had been no error in transmission. If the last message was not properly finished, an error flag is set and the bad message is dropped from the storage buffer in the microprocessor. If the message was properly finished or after the bad message had been dropped from the storage buffer, a test is made to determine whether the secondary counter has been set. This counter is a memory location which is set to 6 when this station initiates a direct message and is expecting a response from another remote station, or if a direct message has appeared on the line but is not directed to this station. The secondary counter will be set to four if the direct message was for this station. Thus, this counter will wait for 6 interrupt signals to be generated before it takes action. In FIG. 8A, if the secondary counter has been set, it is reduced by 1 and a test is performed to see if the secondary counter has reached 0 yet. If it has not, we go to point A and wait for another interrupt. If it has reached 0, a test is performed to determine whether the transmit repeat request flag or response flag have been set. If neither flag has been set, the drop message flag is reset and the system waits for another interrupt. If the transmit repeat request flag or response flag has been set, the system proceeds to FIGS. 8B where the original setting of the secondary counter is checked. If the original setting was a 4, then a response is expected to be transmitted by this station. If a response is expected to be transmitted by this module, the program enters point D in FIG. 8C to begin transmission, which operation will be discussed hereinafter. If the secondary counter had originally been set to 6, a direct message had previously been transmitted by this remote station and it is awaiting an answer or this station has detected a direct message on the bus but directed to another station. Thus, if the secondary counter had originally been set to 6, and if the direct message had been repeated by this remote station and still no answer had been received back, the program proceeds to point A where it again waits for some type of interrupt. If the message had not been repeated, the secondary counter is set to 6, the repeat flag is set and the program enters FIG. 8C at point F to begin transmitting the direct message again. After the direct message is transmitted, and no message is received, the program again waits for 6 quiet interrupts and then proceed to point A to wait for some type of interrupt.

Upon receipt of a quiet interrupt, if the secondary counter had not been set, indicating that the secondary counter was equal to zero after the last message was tested to see if it was properly finished, the drop message flag is reset and the NMI counter is reduced by 1. The NMI counter is the transmit turn counter which is a memory location set to the number of time slots in the protocol system. As mentioned before, if the remote station does not have a pseudo time slot, the countdown counter is originally set to 256 after it has transmitted. Each time a subsequent station transmits, the remote station calculates a new count by subtracting the address of the subsequent station from its address and multiplying by 2. This new count is stored in the countdown counter. If no other station transmitted, the countdown counter would be decremented by one for each generated quiet interrupt. When the counter reaches 0, the remote station may transmit again. After available data is again transmitted, the NMI counter is again set to 256. After the NMI counter has been reduced by one, a test is made to determine if the counter has been reduced to zero. If the counter is at zero, this station's transmit time slot is now available. At this point, a test is made to determine whether the transmit flag has been reset. If it has not, the system proceeds to point A where it waits for another interrupt. If it has been reset, the transmission error flag is tested to see if it has been set. If it has, and if a repeat request has been received, the program proceeds to point A and waits for another interrupt. If the transmission error flag has been set and a repeat request has not been received, the repeat request is formed and the station proceeds to point C. The station also proceeds to point C if the error flag has not been set.

Upon arrival at point C in FIG. 8C, several conditions must be ascertained about the nature of the message to be transmitted. First, a test is made to determine whether this is a pseudo time slot. A pseudo time slot is one of the extra time slots discussed above. If this is a pseudo time slot but the next message to be sent is not a direct message, control is passed to point A where an interrupt is awaited. If the message to be sent next is not a direct message, then it must be a global message and global messages cannot be sent during pseudo time slots although direct messages can be sent in either. Therefore, the regular time slot of the remote station must be awaited in order to send this message. However, if there is a pseudo time slot and if the next message to be sent is a direct message or if this is not a pseudo time slot and the repeat flag has not been set, the pointers are set to send the next message. If the repeat flag has been set and this is not a pseudo time slot, then the pointer is set to send the last global message again since this is a non-pseudo time slot. Once the repeat bit in the control word has been set or the pointers have been set to send the next message, another test is made to determine if the message to be sent is a direct message and, if it is, the message buffer address pointer is saved in case of a repeat. Once this pointer has been saved or, looking at FIG. 8B, the secondary counter has been set to 6 and the message has not been repeated, the strapped address in switches S1-S7 of the interface circuit are stored into the message and the secondary counter is set to 6 indicating that this is a direct message and that we will wait six time slots for an answer. If the message to be sent is not a direct message or once the strapped address is stored into the message and the secondary counter is set to 6 or the program has come from point D in FIG. 8B indicating that this message is a response to a direct message received from another station, the hardware is set to the transmit mode, the first word is transmitted (the control word) and the station address number is inserted into the second word. Next, the block check calculation is made according to FIG. 11 for words 1 and 2. The program then proceeds to point A where it waits for another interrupt. Thus, when quiet interrupts are received and the secondary counter has not been set, indicating that a response is not expected from this station or a direct message which can be transmitted during the pseudo time slot is not to be sent, the NMI counter or time-to-transmit counter is decremented by 1 for each quiet interrupt produced. When this count has been reduced to zero, this station can transmit. If the secondary count has been set to a 4, a response from this station is expected to a previously sent direct message from another remote station. In this case, the response is transmitted. If the secondary counter had been set to 6 indicating that a direct message is to be sent, and a direct message has not been repeated, the system sets the secondary counter and sends this message. If the message has been repeated, the system proceeds to wait for the next interrupt of any type. If the NMI countdown counter (or the time-to-transmit counter) has reached zero but this is a pseudo time slot such that a global message cannot be sent, the system must now reset its NMI countdown and again wait for a time slot during which a global message can be transmitted. If this is a pseudo time slot, a direct message can be sent.

At point A the next interrupt must be investigated to determine what kind of interrupt it is. If this interrupt is not a quiet interrupt but it is a receive interrupt, the receive interrupt routine is entered and is shown in FIGS. 9A-9D. A receive interrupt is generated by the interface and interpreted as such by the microprocessor when transmission has been received over the data bus and is now available in the ACIA 50. The received word is first checked for a hardware error and, if there is a hardware error, the error flag is set and the drop-message flag is set and the program proceeds to point A to wait for another interrupt. If there is no hardware error but the drop-message flag has been set, the secondary counter, which was mentioned in the previous routine, is cleared and the system proceeds to point A to wait for another interrupt. In this way, a message containing any word error is rejected. After the message is ended, a quiet interrupt executes the quiet interrupt routine resetting the drop-message flag. If no hardware errors are detected, the received word is loaded into a character receive buffer. As each correct word is placed in the buffer, a receive word counter is incremented and a block check calculation is performed as shown in FIG. 11. This block check routine determines both whether there is an error and if the last word received is the end of message. Thus, an end-of-message test is performed. If the last word received is the last of the message, the routine proceeds to point E.

Since each word position in a message has a specific meaning, a means of identifying these positions is necessary. If the word received is not an end-of-message, a check is made to determine if it is the first word in a message. To do this, the word counter is examined. If it is the first word, another test is performed to determine if the message is an "I'm OK" message. If the control character word in the receive buffer indicates it is an "I'm OK" message, the routine is told to expect a message length of two characters. If it is an "I'm OK" message and the routine is told to expect a message length of two characters or if it is not an "I'm OK" message, an exit is made to point A where another interrupt is awaited. If this word was not the first word, a test is made to determine whether it is the second word. The second word position of a message is the sender's address. If the character is the second word, a calculation of the transmit turn time is performed, by subtracting the sender's address from this station's address and multiplying by 2, and saved and an exit is made to point A. If the word is the third word, which contains the message length, it is stored in the receive buffer and is also used for comparison to the word counter count at the end of the message.

Word four can be either the control extension word or a receive address word. A control extension indicates it is a global message. A receiver address word indicates it is a direct message. If it is a global message or a direct response message, an exit is made at A. If the message is a direct request message, the secondary counter is set to either a 4 indicating that it must respond after four quiet interrupts or a 6 for those stations which are not addressed. Thus, only the station which has been required to respond to the direct message on the data bus sets its secondary counter to 4. All other stations will set their secondary counters to 6. At the end of 4 time slices, this station will respond. None of the other stations will respond because their secondary counters have been set to 6. For each word received, an interrupt is generated which causes the program to enter the receive interrupt routine. Each word is checked and the block check is performed. When an end-of-message has been detected, the program proceeds to point E.

As shown in FIG. 9A, when a message has been received in its entirety, an end-of-message is detected and the program enters point E in FIG. 9C. FIGS. 9C and 9D show the end-of-message routine. The end-of-message routine first resets the word counter and block check result in the main receive routine and then examines the control word in the receive buffer to determine if the message is an "I'm OK." If the message is an "I'm OK" message and if this module is not a bus monitor, the "I'm OK" message is deleted from the receive buffer. If this station is a bus monitor, it saves the "I'm OK" message in order to update the bus monitor data file. The saved calculated transmit turn is now set into countdown counter. A test is made to determine if a repeat request message is needed due to an error flag or other conditions. If a repeat message is necessary, a repeat request is set up for transmission during the next time slot and the program proceeds to point A. If a repeat message is not needed, the program proceeds directly to point A.

If the message is not an "I'm OK" message, the control word in the receive buffer is examined to determine if it is a direct message. If it is, a further test is made to determine if the direct message is intended for this station as shown in FIG. 9D. If the direct message is intended for this station, it is next determined whether the message is a request for which an answer is necessary or a response to a previous request by this station. If it is a response message, the direct received-to-do flag is set and the module is tested to see if it is bus monitor in FIG. 9C. If the direct message is a request, a response message is prepared. The control extension word is checked for a start or stop transmitting directive. In either case, the appropriate transmit flag is set and then the direct received-to-do flag is set and the program is returned to point J in FIG. 9C.

If the message was not a direct message, then it must be a global message and is tested for a change of state. If a change of state has occurred, the global handler activity flag is set and the bus monitor test is performed. If the global message is a monitor directive such that there will be no change of state indication, the control extension word is tested to see if it is a code 1, 2, 3 or 4 message. A code 1 message is a message to switch bus loops, a code 2 message is a message to set a flag to repeat the last global message, a code 3 message is a message to switch loops and repeat the message, and a code 4 message is a message to wait for a repeat directive. From this point, the global handler activity flag is set and the routine proceeds to the bus monitor where the program eventually proceeds to point A to wait for the next interrupt. Thus, the NMI countdown counter is decremented either by a quiet interrupt or reset by receiving a senders address.

If the interrupt, according to FIG. 7, is neither a quiet interrupt nor a receive interrupt, it must be a transmit interrupt and the transmit interrupt routine in FIG. 10 is entered. Thus, once a message transmission begins, a hardware-generated interrupt is provided by the ACIA when it is ready to receive another word for transmission. The transmit interrupt causes execution of the transmit's second through last word. Immediately upon entering the transmit routine, a test is made of the transmit word counter to determine if all words have been sent. If they have not all been sent, the transmit word counter is decremented by 1, the next word is transmitted, and the block check calculation is updated. The system waits for the next transmit interrupt and this process is repeated.

Once the entire message is sent, i.e., all characters have been transmitted, the result of the last block check calculation is now complemented and then transmitted as the block check word, and the hardware is reset to the receive mode. If this station is a bus monitor, a check is performed to determine if the message was a bus switch directive. If it was, the bus switch flag is set prior to its exit to point A. If not, the program exits directly to point A to await the next interrupt.

The block check word routine, where a block check word is sent with each message having more than two words or where the control word and the sender's address in an "I'm OK" message must satisfy the routine, is shown in FIG. 11. The first word to be transmitted is obtained and rotated one place to the right. The next word is obtained and is added to the first word. A test is made to determine whether or not the sum of these two words results in all ones. If the result is all ones and the least significant bit of the first word before rotation was a one, the algorithm detects a two-word message. Assuming that all ones did not exist after the first two words, the previous result of the addition step is then rotated to the right one bit and the next word is obtained and added to the rotated addition of the first two words. A test is made to determine whether all ones exist. If not, this result is rotated to the right, the next word is obtained and added to it. When all ones are detected in the result, the number of words accumulated to that point is totaled. If the number of words transmitted when all ones exist is equal to the number of words designated for that particular transmission by the third word, an end-of-transmission has been detected. If all ones have been detected but the number of words accumulated to that point is not the same as the number of words designated in the message length third word, the next word is obtained and the process repeats until all of the words have been tested. If all of the words have been tested and the result is still not all ones, a system error has resulted.

In the special case of a two-word message where the two words are the control word and the address word, this check routine still must exist. A two-word message will satisfy this routine by first specifying that the most significant bit of the sender's address must be a zero. When the sender transmits his address, he first rotates the address word to the left by one bit and then complements the word. This result is sent as a first word or control word. By nature of the arithmetic, the least significant bit of the transmitted control word contains a 1. This bit is reserved to be a 1 only for a two-word message. When the check word algorithm is applied to this two-word message, the condition of all ones will exist after the second word is added to the first transmitted word with rotation.

If words 1, 2 and 3 had been transmitted, word 1 is rotated, then word 2 is rotated and added to rotated word 1. This result is added to rotated word 3. The final result is complemented and transmitted as the block check word. ##SPC1## ##SPC2## ##SPC3##