US 20010003844 A1
Programmable logic array devices are programmed from programming devices in networks that facilitate programming any number of such logic devices with programs of any size or complexity. The source of programming data and control may be a microprocessor or one or more serial EPROMs, one EPROM being equipped with a clock circuit. Several parallel data streams may be used to speed up the programming operation. A clock circuit with a programmably variable speed may be provided to facilitate programming logic devices with different speed characteristics. The programming protocol may include an acknowledgment from the logic device(s) to the programming data source after each programming data transmission so that the source can automatically transmit programming data at the speed at which the logic device is able to accept that data.
1. An apparatus to program a plurality of programmable logic devices (PLDS) located on respective integrated circuits, comprising:
a source configured to generate programming data, said source being external to said integrated circuits;
a second source to generate a clock signal, said second source being external to said integrated circuits;
a first PLD device configured to receive said programming data and said clock signal, said first PLD device further configured to use said clock signal to clock said programming data to program said first PLD device to a first logic state, said first PLD device further including a done circuit which is configured to generate a done signal when said first PLD device is programmed to said first logic state; and
a second PLD device configured to receive said done signal, and in response thereto, is further configured to receive said clock signal and said programming data to program said second PLD device to a second logic state.
2. The apparatus of
3. The apparatus of
4. The apparatus of
5. The apparatus of
6. The apparatus of
7. The apparatus of
8. The method of programming a programmable logic integrated circuit device from separate programming circuitry which includes programming data memory and a single clock signal circuit that is programmable to produce a clock signal having any of a plurality of frequencies, each frequency being suitable for use as a programming clock signal frequency by different ones of said programmable logic integrated circuit device, said method comprising:
determining a one of said frequencies that is suitable for use as a programming clock signal frequency by said programmable logic integrated circuit device;
programming said single clock signal circuit to produce said clock signal with said one of said frequencies; and
applying programming data from said programming data memory to said programmable logic integrated circuit device in synchronism with said clock signal produced by said clock signal circuit at said one of said frequencies.
9. The method defined in
using said clock signal produced by said single clock signal circuit to retrieve programming data from said programming data memory at said one of said frequencies.
10. The method defined in
applying said clock signal produced by said single clock signal circuit to said programmable logic integrated circuit device as a programming clock signal.
11. The method defined in
applying said clock signal produced by said single clock signal circuit to said programmable logic integrated circuit in parallel with said programming data.
12. The method defined in
programming said shunt circuits to effectively remove from said series a number of said delay elements appropriate to cause said clock signal to have said one of said frequencies.
13. Programmable logic device apparatus comprising:
a programmable logic integrated circuit device configured to be programmed by successive programming data signals applied to said device in synchronism with a programming clock signal having one of a plurality of different frequencies at which different such programmable logic integrated circuit devices can be similarly programmed by said same programming data; and
programming circuitry separate from said device and including a programming data memory and a single clock circuit, said single clock circuit being programmable to produce a programming clock signal having any one of said plurality of different frequencies, said single clock circuit being configured to apply programming data from said programming data memory to said device as programming data signals synchronized with said programming clock signal produced by said single clock circuit with said one of said frequencies.
14. The apparatus defined in
15. The apparatus defined in
a first connection between said programming circuitry and said device configured to apply said programming clock signal produced by said single clock circuit to said device.
16. The apparatus defined in
a second connection between said programming circuit and said device configured to apply said programming data signals to said device in parallel with said programming clock signal applied to said device via said first connection.
17. The apparatus defined in
a plurality of delay elements connected in series;
a plurality of programmable shunt circuits, each of which is capable of effectively removing an associated subset of said delay elements from said series by shunting said delay elements in said associated subset, said frequency of said programming clock signal being determined by the number of delay elements that are selectively present in said series.
 This is a continuation of application Ser. No. 08/851,250, filed May 5, 1997, which is a continuation application Ser. No. 08/747,194 filed Nov. 12, 1996, now is U.S. Pat. No. 5,680,061, which is a continuation of application Ser. No. 08/658,537, filed Jun. 5, 1996, now abandoned.
 This invention relates to programmable logic array devices, and more particularly to techniques for programming such devices.
 Illustrative programmable logic array devices requiring programming are shown in Cliff U.S. Pat. No. 5,237,219 and Cliff et al. U.S. Pat. No. 5,434,514. Typically, such devices are “programmed” in order to set them up to thereafter perform desired logic functions. In other words, the programming determines what logic functions the device will perform. The present invention is particularly of interest in connection with programming programmable logic array devices whose programming memory elements are volatile and reprogrammable. For example, such devices typically require reprogramming each time their power supplies are turned on (from having been off). Such devices may also require reprogramming whenever it is desired to change the logic functions they perform, which may occur during certain normal uses of the devices. Because such programming (or reprogramming) may have to be performed relatively frequently, and because the logic devices are generally not usable during programming, it is important to have rapid and efficient programming techniques.
 Programmable logic array devices are often designed to be “general purpose” devices. In other words, the programmable logic device is made without any particular end use in mind. It is intended that the customer will use the number of such devices that is appropriate to the customer's application, and that the customer will program those devices in the manner required to enable them to perform the logic required in the customer's application. Because the size and complexity of various customer applications may vary considerably, it would be desirable to have programming techniques that are modular and lend themselves to programming different numbers of devices with programs of different sizes.
 In view of the foregoing, it is an object of this invention to provide improved techniques for programming programmable logic array devices.
 It is another object of this invention to provide more rapid techniques for programming programmable logic array devices.
 It is still another object of this invention to provide programmable logic array device programming techniques which lend themselves to programming any number of such devices with programs of any size or complexity.
 These and other objects of the invention are accomplished in accordance with the principles of the invention by providing programmable logic array devices which can be programmed one after another in any number from programming devices such as serial erasable programmable read only memories (“serial EPROMs”). Any number of such programming devices can be connected to operate serially. Thus any number of logic devices can be programmed from any number of programming devices, making the programming technique highly modular and capable of performing programming tasks of any size and complexity. The logic devices may be equipped with programming register configurations that allow the logic device to receive several programming data streams in parallel, thereby speeding up the transfer of programming data from the programming device(s) to the logic device(s). A programming device may be equipped with a clock signal generating circuit whose operating speed is programmably variable, thereby enabling the programming device(s) to be used to program logic device(s) having different clock rate requirements. Various communications protocols may be used between the programming devices and the logic devices.
 Further features of the invention, its nature and various advantages will be more apparent from the accompanying drawings and the following detailed description of the preferred embodiments.
FIG. 1 is a simplified schematic block diagram of portions of an illustrative programmable logic array device requiring programming in accordance with this invention.
FIG. 2 is a simplified schematic block diagram of a network of logic devices (each of which can be of the type shown in FIG. 1) and programming data and control source devices in accordance with a first illustrative embodiment of this invention.
FIG. 3 is a simplified block diagram of portions of an alternative programmable logic device in accordance with this invention.
FIG. 4 shows how the network of FIG. 2 can be modified in accordance with this invention to program devices of the type shown in FIG. 3.
FIG. 5 shows illustrative signals in networks of the types shown in FIGS. 2 and 4.
FIG. 6 shows other illustrative signals in networks of the types shown in FIGS. 2 and 4.
FIG. 7 shows more illustrative signals in networks of the types shown in FIGS. 2 and 4.
FIG. 8 shows still more illustrative signals in networks of the types shown in FIGS. 2 and 4.
FIG. 9 is a simplified schematic block diagram of a circuit which can be used on one of the programming devices in FIG. 2 or FIG. 4 in accordance with this invention.
FIG. 10 is a simplified schematic block diagram similar to FIG. 2 or FIG. 4 showing an alternate signalling scheme for programming data in accordance with this invention.
FIG. 11 shows illustrative signals in networks of the type shown in FIG. 10.
 Although the invention is equally applicable to programming other types of programmable logic array devices, the invention will be fully understood from the following explanation of its application to programming programmable logic array devices of the general type shown in FIG. 1 (which depicts a structure like that shown in Cliff U.S. Pat. No. 5,237,219 and Cliff et al. U.S. Pat. No. 5,434,514, both of which are incorporated by reference herein). Programmable logic array device 10, which is preferably a single integrated circuit, includes a shift register 20 having a plurality of serially connected shift register stages 20A-20 m. (The letter “m” is used in FIG. 1 as a general index limit which can have any desired value.) Programming data supplied to device 10 via lead DATA from an external programming data source is shifted into shift register 20 from left to right as viewed in FIG. 1 by clock pulses applied to lead DCLK. In accordance with the present invention, the DCLK pulses also come from a source external to device 10. When register 20 is fully loaded, a signal supplied to the BCLK lead loads all the stages 30A-30 m of register 30 in parallel from register 20. The BCLK signal may be generated by device 10 itself based on counting the DCLK pulses and producing a BCLK pulse after each m DCLK pulses have been received. A count of m DCLK pulses indicates that register 20 is full and ready to be dumped to register 30. Dumping register 20 to register 30 makes it possible for register 20 to immediately begin shifting in more programming data, while the data in register 30 is going in parallel into the programmable registers 40 of device 10 as will now be described.
 Each stage of register 30 feeds data to an associated chain of programmable registers 40. For example, register stage 30A feeds the chain of registers that includes stages 40Ala through 40And (where again the letters “d” and “n” are used as index limits which can have any desired values). These register chains may be so-called “first-in-first-out” or “FIFO” chains which progressively fill with data from the bottom (e.g., stage 40And) to the top (e.g., stage 40Ala). In other words, the first programming data bit supplied to a chain from register 30 passes down through all the stages of the chain to be stored in the bottom-most stage. Device 10 then cuts off the bottom-most stage so that the next programming data bit supplied to the chain from register 30 is stored in the next to bottom-most stage of the chain, which is then cut off from the stages above. This process continues until all the stages of registers 40 have been programmed. More detail regarding this type of FIFO chain programming will be found in above-mentioned Cliff U.S. Pat. No. 5,237,219.
 Each stage of each register 40 controls some aspect of the programmable logic 50 of device 10. For example, register stages 40Ala-40Ald control various portions of the programmable logic in logic array block 50A1, while register stages 40A2 a-40A2 d control various portions of the programmable logic in logic array block 50A2. It will be understood that each of logic array blocks 50 is capable of performing any of several logic functions, depending in part on how it is controlled by the programming signals stored in the associated register 40 stages.
 In addition to the elements described above, device 10 typically includes a network of conductors (not shown) for interconnecting logic array blocks 50 with one another and with input and output pins (also not shown) of device 10. An illustrative arrangement of such other elements is shown in Cliff et al. U.S. Pat. No. 5,260,611, which is also hereby incorporated by reference herein.
 An illustrative network in accordance with this invention for applying programming data to one or more programmable logic array devices 10 a, 10 b, etc. (each of which can be like device 10 in FIG. 1) is shown in FIG. 2. Each of devices 10 a, 10 b, etc. is typically a separate integrated circuit. Each of devices 100 a, 100 b, etc. is also typically a separate integrated circuit. For example, each of devices 100 may be a serial erasable programmable read only memory (“serial EPROM”). Device 100 a is the main device of this kind. Device 100 b is an auxiliary device which is included only if device 100 a does not have enough capacity to store all the programming data needed to program all of the connected devices 10. As suggested by the dots on the left, additional auxiliary devices 100 may be included if needed to provide still more programming data storage capacity.
 When power is first applied to devices 10 and 100, each of those devices pulls down on node N1 via its nSTATUS or nERR terminal until it is ready to operate. When each of devices 100 is no longer pulling down on node N1, each of those devices monitors (via its nERR terminal) the level of the signal at node N1. When all of the devices connected to node N1 are ready to operate and no device is pulling down on that node, the electrical potential of that node rises to VCC. This indicates to devices 100 that programming of devices 10 can begin.
 Each of devices 10 also pulls down on node N2 via its CONDONE terminal until that device is fully programmed. As long as node N2 is low, device 100 a is enabled to operate via its nCE input terminal. Each of devices 10 can detect when it is fully programmed, for example, by counting the number of DCLK pulses it has received since it was enabled via its nCE input terminal.
 The nCONFIG signal is a reset type signal which can be used to initiate a re-programming of devices 10. For example, if programming of devices 10 were controlled by a microprocessor with the ability to select different programming data at different times (e.g., to change the logic functions performed by devices 10), the microprocessor could apply an appropriate nCONFIG signal to devices 10 whenever re-programming is desired. Among the effects of an nCONFIG request for re-programming of devices 10 is that each of devices 10 again pulls down on node N2 via its CONDONE terminal. This can be used to signal the programming data source that devices 10 are ready to begin receiving new programming data. Other effects of an nCONFIG request for re-programming are (1) readying each device 10 to again begin counting DCLK pulses from a reset starting value, and (2) restoring the nCEO output signal of each device 10 to its initial unprogrammed value.
 Device 10 a is enabled to accept programming data at all times because its nCE input terminal is tied to ground. Until each device 10 is fully programmed, that device applies to the nCE input terminal of the next device in the series of devices 10 an input signal that prevents the next device from accepting programming data. Devices 10 are therefore programmed one after another in order, beginning with device 10 a.
 When device 100 a is enabled by node N1 being high while node N2 is low, device 100 a begins issuing clock signals on its DCLK output lead, as well as issuing programming data bits (synchronized with the DCLK pulses) on its DATA output lead. These data and clock signals are respectively applied to the DATA and DCLK input terminals of all of devices 10. At first, however, only device 10 a responds to these signals because only device 10 a has a chip enabling signal applied to its nCE input terminal. Thus only device 10 a operates as described above in connection with FIG. 1 to take in the programming data and make use of that data for programming itself.
 Once device 10 a is fully programmed, it cannot respond to any more programming data even though more such data and DCLK pulses may be applied to it. As soon as device 10 a is fully programmed, it applies a chip enabling signal to the nCE input terminal of device 10 b. This enables device 10 b to begin to take in the programming data applied to its DATA input terminal at the DCLK rate. This begins the programming of device 10 b. When device 10 b is fully programmed, it produces an nCEO output signal suitable for enabling the next device 10 to begin accepting programming data. (The possible presence of such further devices 10 is indicated by the dots extending to the right in FIG. 2.) The process of successively programming devices 10 continues until all of those devices have been fully programmed. Node N2 then rises to VCC, thereby disabling device 100 a and any other devices 100 in the network. For example, when thus disabled, device 100 a stops issuing DCLK signals and otherwise goes into a state in which it consumes little or no power. (Via the nCEO-nCE connection chain between devices 100, any other device(s) 100 in the network are similarly placed in a low or no power state when node N2 rises to VCC.)
 It should be noted that all of devices 10 also monitor (via their CONDONE terminals) the level of the node N2 signal. When node N2 rises to VCC, each of devices 10 responds by preparing to begin normal operation as a logic device. This may include such conventional operations as resetting various clocks and counters, releasing the reset on various registers, and enabling output drivers.
 If more data is required to program devices 10 than can be produced by one device 100, then device 100 a is supplemented by additional devices such as 100 b. As long as device 100 a is applying data to the data bus of the network, device 10 a applies to the nCE input terminal of device 100 b a high signal which disables device 100 b. (Each device 100 also applies such a high signal to the adjacent device 100 as long as the signal applied to its nCE input terminal is high.) Device 100 b also receives the DCLK output signal of device 100 a, but device 100 b cannot and does not respond to that signal until it is enabled by a chip enabling signal applied to its nCE input terminal.
 When device 10 a has applied the last of its data to its DATA output terminal, it changes the state of the signal applied to the nCE terminal of device 100 b. This enables device 100 b to begin responding to the applied DCLK signal, which device 100 a continues to produce at the same rate. Device 100 b then begins to output its data via its DATA output terminal at the DCLK rate. The data from device 100 b therefore becomes a continuation of the data stream from device 100 a and programming of devices 10 accordingly continues on the basis of that data.
 If even more programming data is required than can be held by devices 100 a and 100 b, the series of devices 100 can be extended to as many as are required to hold all the necessary data. Device 100 b applies a chip enabling signal to the nCE terminal of the next device 100 after it has output all of its data. The next device 100 is thereby enabled to respond to continued DCLK pulses from device 100 a and to begin outputting its data via its DATA output terminal.
 It will be apparent from the foregoing that there is no required correlation between the relative sizes of devices 10 and devices 100, although it is preferred for each transition from one device 100 to the next to occur at the end of a “frame” of data. (A “frame” of data is the data required to fill register 20. There may be a small delay in the start-up of each successive device 100. To cope with this, each device 100 initially outputs a few dummy data values (e.g., a series of binary ones) which are ignored by the device 10 being programmed. To facilitate ignoring such dummy data it preferably occurs between frames of data rather than in the midst of a frame of data. Thus it is preferred that transitions between devices 100 occur between frames of real programming data.) Except for the possible minor constraint explained in the immediately preceding parenthetical, the transitions between deriving programming data from successive devices 100 can occur at any times relative to the transitions between programming successive devices 10. For example, programming data may stop coming from device 100 a and start coming from device 100 b at the end of a frame of data halfway through the programming of device 10 b. The programming networks of this invention are therefore highly modular and flexible with regard to device sizes. Devices 10 of any size(s) can be used with devices 100 of any size(s), again bearing in mind the preference for transitions from one device 100 to the next device 100 at the end of a frame of date.
FIG. 3 shows an alternative embodiment 10′ of programmable logic array device 10 which can be programmed more rapidly than device 10. In device 10′ shift register 20′ has several data input terminals D0 through DN spaced equally along its length. For example, if shift register 20′ has 100 stages (from stage 0 at the left to stage 99 at the right), and if N=9, then data input terminal D0 is at stage 0, terminal D1 is at stage 10, terminal D2 is at stage 20, and so on through input terminal D9 at stage 90. Register 20′ receives data in parallel at its several data input terminals and shifts that data to the right at the DCLK rate. (Data is not shifted from the left into shift register stages having inputs D0-DN. Thus shift register 20′ may alternatively be N+1 separate shift registers, each having a respective one of inputs D0-DN.) Accordingly, the time required to fill register 20′ from its several data input terminals is only 1/(N+1) the time required to fill register 20 in FIG. 1 from its single data input terminal. In other respects device 10′ can be identical to device 10. Thus each time device 10′ detects (e.g., by counting DCLK pulses that have been received) that register 20′ contains data that is all new since the last BCLK pulse, device 10′ applies a BCLK pulse to register 30. As in device 10, this causes register 30 to accept in parallel all the data contained in register 20′. Register 20′ is thereby freed to begin accepting new data via its D0-DN input terminals, while the data in register 30 is used to program the main portion 40/50 of device 10′ as described above in connection with FIG. 1.
FIG. 4 shows how the network of FIG. 2 can be modified for programmable logic array devices 10′ of the type shown in FIG. 3. Instead of one data input terminal as in FIG. 2, each device 10 a′, 10 b′, etc. in FIG. 4 has N+1 data input terminals. Similarly, each device 100 a′, 100 b′, etc. in FIG. 4 has N+1 data output terminals rather than one such terminal as in FIG. 2. (Alternatively, each of devices 100′ could be N+1 serial devices arranged in parallel.) Thus one of devices 100′ outputs N+1 programming data bits in parallel during each DCLK pulse interval, and one of devices 10′ inputs those data bits during that interval. The data bus in FIG. 4 is therefore N+1 conductors wide, rather than being a single conductor as in FIG. 2. In all other respects the network of FIG. 4 may be constructed and may operate exactly as described above in connection with FIG. 2.
 A typical signal sequence in FIG. 2 or FIG. 4 when only one device 100 a or 100 a′is needed to program device(s) 10 or 10′ is shown in FIG. 5. (The nCE and nCEO signals shown in FIG. 5 are those associated with device 100 a or 100 a′.) At 120 all of devices 10 and 100 a or 10′ and 100 a′ have signalled that they are ready to begin the programming process. The level of the signal at node N1 therefore rises to VCC. Device 100 a or 100 a′ responds by beginning to produce synchronized DCLK and DATA output signals. Each DATA signal pulse in FIG. 1 represents either a bit of data (in the case of networks of the type shown in FIG. 2) or a word of data (in the case of networks of the type shown in FIG. 4).
 Assuming that n bits or words of data are required to fully program device(s) 10 or 10′, when device 100 a or 100 a′ outputs the last bit or word, device(s) 10 or 10′ detect that they are filled and allow the signal at node N2 to rise to VCC as shown at 122 in FIG. 5. Device 100 a or 100 a′ then produces a few more (e.g., 16) DCLK pulses. If no error conditions are detected during those further DCLK pulses, the programming process has been completed successfully and device 100 a or 100 a′ switches to the low or no power mode described above. (Examples of error conditions are discussed below in connection with FIGS. 7 and 8.)
FIG. 6 illustrates a typical signalling sequence in FIG. 2 or FIG. 4 when two or more devices 100 or 100′ are required to produce the data needed to program the device(s) 10 or 10′ in the network. In FIG. 6 the upper signals nCE, DCLK, DATA, and nCEO are associated with device 100 a or 100 a′, while the lower signals nCE, DCLK, DATA, and nCEO are associated with device 100 b or 100 b′. FIG. 6 assumes that all of the devices 100 or 100′ in a network are constructed identically, for example, with the capability of producing a DCLK signal. As described above, however, only the main device 100 a or 100 a′ actually produces the DCLK signal.
 Considering FIG. 6 now in more detail, transition 120 is identical to transition 120 in FIG. 5. Immediately after transition 120, each device 100 or 100′ detects whether it is the main device of that type or an auxiliary device of that type. This determination can be made on the basis of the level of the applied nCE signal when transition 120 occurs. The device 100 or 100′ with the low nCE signal at transition 120 is the main device 100 a or 100 a′. Devices 100 or 100′ with a high nCE signal at transition 120 are auxiliary devices like 100 b or 100 b′. Thus in FIG. 6 the device 100 a or 100 a′ associated with the upper signals nCE, DCLK, DATA, and nCEO determines that it is the master device and begins producing synchronized DCLK and DATA pulses shortly after transition 120 as described above in connection with FIG. 5.
 When device 100 a or 100 a′ is about to produce its last bit (FIG. 2) or word (FIG. 4) of data m, device 100 a or 100 a′ causes its nCEO output signal to transition from high to low as shown at 130. This causes a similar transition 132 in the nCE input signal of first auxiliary device 100 b or 100 b′. Device 100 a or 100 a′ then produces its final data output m and thereafter stops producing data. However, device 100 a or 100 a′ continues to produce DCLK output pulses, and device 100 b or 100 b′ begins to respond to those pulses by producing DATA signals m+1, m+2, etc. in synchronism with the DCLK pulses from device 100 a or 100 a′.
 After device 100 b or 100 b′ has produced its last data n, the device(s) 10 or 10′ in the network signal a full condition by allowing the nCE signal applied to device 100 a or 100 a′ to rise to VCC as shown at 122. This causes device 100 a or 100 a′ to raise its nCEO output signal to VCC as shown at 134, which similarly raises the nCE input signal of device 100 b or 100 b′ to VCC as shown at 136. Device 100 b or 100 b′ is thereby placed in a low or no power mode, and after a predetermined number of further clock pulses from device 100 a or 100 a′, that device also enters a low or no power mode.
 If desired, the apparatus shown in FIGS. 1-4 may include various types of programming error detection signalling. For example, FIG. 7 shows any of devices 10 or 10′ using the level of the signal at node N1 to indicate that it has detected a programming error. Devices 100 or 100′ respond to such an indication by stopping and restarting the programming operation.
 With more detailed reference to FIG. 7, at 150 (similar to 120 in FIG. 5 or FIG. 6) the signal at node N1 (FIG. 2 or FIG. 4) goes high, indicating that all of devices 10 and 100 or 10′ and 100′ are ready for programming to begin. The nCE signal is also low, indicating that devices 10 or 10′ are as yet unprogrammed. Shortly after transition 150, device 100 or 100′ begins to output synchronized DCLK and DATA signals. The successive bits (FIG. 2) or words (FIG. 4) of DATA are numbered 1, 2, 3, . . . n, n+1, etc. in FIG. 7. At time 152 one of devices 10 or 10′ detects that it has not received correct programming data or that something else has gone wrong with the programming process. That device 10 or 10′ therefore uses its nSTATUS terminal to lower the level of the signal at node N1. This is detected by devices 100 or 100′ via their nERR terminals. Devices 100 or 100′ therefore shortly thereafter cease outputting DCLK and DATA signals and reset themselves to prepare to restart the programming process. All of devices 10 or 10′ also detect that the level of the signal at node N1 has been pulled down. Devices 10 or 10′ therefore also all reset themselves to prepare for the restarting of the programming process.
 After a suitable time-out interval, the device 10 or 10′ that detected the programming error and caused transition 152 allows the nSTATUS/nERR signal to again rise to VCC as shown at 154. Transition 154 is like transition 150, and so shortly thereafter device 100 or 100′ again begins outputting synchronized DCLK and DATA signals, beginning again with the programming data at the start of the programming data sequence.
 Another example of programming error detection signalling that may be used in systems of the type shown in FIGS. 2 or 4 is illustrated by FIG. 8. The first portion of FIG. 8 is identical to FIG. 7, except that FIG. 8 additionally shows a counter which is preferably located on device 100 a or 100 a′ for counting the number of data bits (FIG. 2) or words (FIG. 4) that have been output by devices 100 or 100′. Assuming that the entire program consists of n bits or words, when that amount of data has been output, the counter reaches a count of n and devices 100 or 100′ stop outputting data. Device 100 a or 100 a′ then waits a predetermined number of DCLK cycles for the signal at node N2 to rise to VCC. As described above, devices 10 or 10′ should allow this to happen when each of those devices recognizes that it is fully programmed. However, if for any reason one of devices 10 or 10′ has not been fully programmed, it does not allow the level of the signal at node N2 to rise to VCC. If the above-mentioned predetermined number of DCLK cycles passes without the signal at node N2 rising to VCC, this is detected by device 100 a or 100 a′ via that device's nCE terminal. Device 100 a or 100 a′ then knows that one of devices 10 or 10′ was not fully programmed and that the programming process should be repeated. Device 100 a or 100 a′ therefore pulls down the signal at node N1 as shown at 160. This resets all of devices 10 and 100 or 10′ and 100′′. After a predetermined time-out interval, device 100 a or 100 a′ allows the signal to transition back to VCC as shown at 162, which restarts the programming process as at transition 150.
 In order to facilitate programming of programmable logic array devices 10 or 10′ having different speed capabilities, device 100 a or 100 a′ may include a DCLK circuit having a programmably adjustable clock rate. An illustrative embodiment 200 of such a circuit is shown in FIG. 9. A signal pulse propagates repeatedly around the closed loop made up of inverters 210 a-210 w, although it will be understood that the number of inverters in this loop is arbitrary and that some of the inverters may sometimes be switched out of use as will be more fully explained below. The loop of inverters 210 is tapped at one location by inverters 220 to produce the DCLK output signal. The clock rate of the DCLK signal is determined by the time required for a signal to propagate all the way around the inverter loop.
 In order to adjust the DCLK rate, several groups of inverters 210 can be short-circuited to effectively remove them from the inverter loop. For example, inverters 210 a-210 d can be short-circuited by closing switch 230 a. Switch 230 b is opened whenever switch 230 a is closed to avoid having more than one path around the inverter loop at any one time. Similarly, inverters 210 e and 210 f can be short-circuited by closing switch 230 c and opening switch 230 d. Inverters 210 r and 210 s can be short-circuited by closing switch 230 e and opening switch 230 f. Inverters 210 t-210 w can be short-circuited by closing switch 230 g and opening switch 230 h. A programmable register 240 on device 100 a or 10 a′ controls the status of switches 230. Stage R1 of register 240 controls the status of switches 230 a and 230 b in complementary fashion. Stage R2 of register 240 similarly controls the status of switches 230 c and 230 d. Stage R3 of register 240 controls the status of switches 230 e and 230 f. And stage R4 of register 240 controls the status of switches 230 g and 230 h.
 From the foregoing, it will be apparent that the clock rate of the DCLK signal can be adjusted by appropriately programming register 240. For example, if the “normal” clock rate is the result of having inverters 210 a-210 f in the circuit, but having inverters 210 r-210 w short-circuited, the following table indicates how the clock rate can be increased (fewer inverter delays) or decreased (more inverter delays) from the normal rate:
 Device 100 a or 100 a′ can be programmed via register 240 to produce a slower DCLK rate when the programmable logic array devices 10 or 10′ being programmed are relatively slow. Device 100 a or 100 a′ can be programmed to produce a faster DCLK rate when the programmable logic array devices 10 or 10′ being programmed are relatively fast. This facilitates providing one type of device 100 a or 100 a′ that is suitable for programming a wide range of devices 10 or 10′.
FIGS. 10 and 11 show another type of programming signalling that can be used in accordance with this invention if desired. In FIG. 10 each of devices 310 a, 310 b, etc., can be similar to a device 10 in FIG. 2 or a device 10′ in FIG. 4. Device 400 can be similar to device 100 a in FIG. 2 or device 100 a′ in FIG. 4. Instead of producing a DCLK signal, however, device 400 produces a data available (“DAV”) signal transition 420 a short time after each possible transition 410 in the programming DATA signal. The DAV output signal of device 400 is applied to the DAV input terminal of each of devices 310. A short time after receiving each DAV signal transition 420, the device 310 currently being programmed shifts in the DATA signal currently being applied to its DATA input terminal. Then the device 310 currently being programmed produces a data acknowledge (“DACK”) signal transition 430 to acknowledge that it has received the DATA signal. The DACK signal is applied to device 400. After receiving each DACK signal transition 430, device 400 causes the DAV signal to transition (as at 422) back to its original condition. The device 310 currently being programmed detects each DAV signal transition 422 and responds shortly thereafter by causing the DACK signal to transition (as at 432) back to its original condition. Device 400 detects each DACK signal transition 432 and shortly thereafter (at 410) begins to output the next DATA signal pulse. This begins the next sequence of DAV and DACK signal transitions 420, 430, 422, and 432.
 An advantage of the signalling scheme illustrated by FIGS. 10 and 11 is that the programming data source device 400 automatically adjusts to whatever speed the device currently being programmed is capable of receiving programming data at. Without this type of signalling scheme, programming device 400 must be set to send out data no faster than the slowest device 310 that may need to be programmed. If, as is often the case, different devices 310 may be able to accept programming data at different speeds, this will mean that device 400 will have to be set to operate more slowly than many devices 310 are capable of having it operate. The result will be slower average programming time. By using the signalling technique illustrated by FIGS. 10 and 11, each device 310 is automatically programmed at whatever speed it can accept data. This will shorten programming time for many devices 310.
 Except as described above, the apparatus of FIG. 10 may be constructed and operate as previously described in connection with FIG. 2 or FIG. 4. Thus the DATA bus in FIG. 10 may be either a single lead (as in FIG. 2) or several parallel leads (as in FIG. 4).
 It will be understood that the foregoing is only illustrative of the principles of this invention, and that various modifications can be made by those skilled in the art without departing from the scope and spirit of the invention. For example, logic devices 10 and 10′ can have other, conventional, internal organizations of their programming and logic circuitry (e.g., elements 40 and 50 in FIG. 1). As another example of modifications within the scope of the invention, a microprocessor can be used in place of devices 100 or 100′ in networks of the type shown in FIGS. 2 and 4.