Programmable Logic Array Structure for Semiconductor Nonvolatile Memories, Particularly Flash-Eeproms

The present application claims priority from European Patent Application No 94830072.8, filed Feb. 18, 1994, and incorporated herein by reference. The present application is related to pending U.S. patent application Ser. No. 08/391,160 entitled "METHOD AND CIRCUIT FOR TIMING THE LOADING OF NONVOLATILE-MEMORY OUTPUT DATA", by Luigi Pascucci, Marco Maccarrone and Marco Olivo, and filed on Feb. 21, 1995, which claims priority from European Patent Application No. 94830069.4, filed on Feb. 18, 1994; U.S. patent application Ser. No. 08/391,159 entitled "INTERNAL TIMING METHOD AND CIRCUIT FOR PROGRAMMABLE MEMORIES", by Luigi Pascucci, Marco Olivo and Carla Maria Golla, and filed on Feb. 21, 1995, which claims priority from European Patent Application No. 94830070.2, filed on Feb. 18, 1994; U.S. patent application Ser. No. 08/391,146 entitled "LOAD SIGNAL GENERATING METHOD AND CIRCUIT FOR NONVOLATILE MEMORIES", by Luigi Pascucci and Carla Maria Golla, and filed on Feb. 21, 1995, which claims priority from European Patent Application No. 94830071.0, filed on Feb. 18, 1994; U.S. patent application Ser. No. 08/391,147 entitled "METHOD AND CIRCUIT FOR SUPPRESSING DATA LOADING NOISE IN NONVOLATILE MEMORIES", by Luigi Pascucci, Carla Maria Golla and Marco Maccarrone, and filed on Feb. 21, 1995, which claims priority from European Patent Application No. 94830073.6, filed on Feb. 18, 1994; U.S. patent application Ser. No. 08/391,920 entitled "METHOD AND CIRCUIT FOR TIMING THE READING OF NONVOLATILE MEMORIES", by Luigi Pascucci, Silvia Padoan, Carla Maria Golla, Marco Maccarrone and Marco Olivo, and filed on Feb. 21, 1995, which claims priority from European Patent Application No. 94830074.4, filed on Feb. 18, 1994, all of which are incorporated herein by reference.

1. Field of the Invention

The present invention relates to a programmable logic array structure for semiconductor nonvolatile memories, particularly flash-EEPROMs.

2. Background of the Invention

As is known, in present nonvolatile memories, the control of certain procedures formerly controlled externally tends to be integrated in the memory itself. For example, flash-EEPROM memories currently involve complex erase procedures wherein the entire memory is preprogrammed to "0", then erased, and the outcome of the operation checked and, if necessary, corrected. These operations, among other things, require sequential addressing of portions (sectors) of the array and hence state machines implementing at least a sequencer connected to a signal interpreter.

In known nonvolatile memories, the state machines required are formed by means of programmable logic arrays (PLA) in a so-called NOR-NOR configuration, i.e. in addition to input and output latch elements and an input buffer, the state machine also comprises a portion (AND plane) formed by OR-connected MOS transistors forming a number of minterms, followed by a portion (OR plane) formed by a further number of OR-connected MOS transistors connecting (if provided) each minterm to the PLA outputs.

The NOR-NOR configuration presents the disadvantage of high consumption, due to reading of the PLA requiring a first phase for precharging all the minterms, and a second evaluation phase. At the precharge phase, enabled by switching (e.g. the leading edge) of a clock signal, all the minterms are charged; and, at the evaluation phase, enabled by subsequent switching of the clock signal (in this case, the trailing edge), all but some of the minterms are discharged, and then recharged at the next read phase. This therefore results in high consumption, which increases in proportion to the size of the PLA.

In the case of digital signal processors, one proposal already made to solve the above problem is to employ, for the PLA, a so-called NAND-NOR configuration wherein the AND plane is implemented by a number of NAND transistors defining the various minterms and followed by an inverter.

Despite considerably reducing consumption by discharging only one minterm (or at any rate only the enabled minterms) after the precharge step, the above configuration is not readily applicable in cases (such as nonvolatile memories) involving high electric fields and voltages, which tend to discharge the charged capacitive nodes, thus impairing the reliability of the reading. This effect increases in direct proportion to an increase in the duration of the clock signal, in which case the time lapse between one precharge step and the next becomes increasingly comparable with the discharge time of the associated capacitances, and an even significant number of nodes may change state, thus resulting in serious read errors.

It is an object of the present invention to provide a programmable logic array for nonvolatile memories, that reduces the read time cycle and reduces read errors.

According to the present invention, there is provided a programmable logic array structure for semiconductor nonvolatile memories, particularly flash-EEPROMs, as claimed in claim 1.

A preferred, non-limiting embodiment of the present invention will be described by way of example with reference to the accompanying drawings, in which:

FIG. 1 shows a block diagram of the programmable logic array according to the present invention.

FIG. 2 shows a circuit diagram of the clock source of the FIG. 1 array.

FIG. 3 shows a plot of signals at various points in the FIG. 2 diagram.

For the purpose of this description, the term "programmable logic array structure" is intended to mean the whole formed by the actual PLA array and the clock generator generating the signals for operating the PLA.

Number 1 in FIG. 1 indicates a programmable logic array (PLA) substantially comprising, in a known manner, the following parts: an input buffer stage 2 (possibly preceded by an input latch circuit, not shown); an AND stage or plane 3; an OR stage or plane 4; and an output latch stage 5.

More specifically, input buffer stage 2 comprises a number of buffer circuits 7 (only one shown), each formed by a pair of inverters 8 and 9 supplying at the output the inverted natural value of the respective input.

AND plane 3 comprises a number of MOS transistors forming a number of minterms 10 of which only one is shown, partially. Each minterm 10 comprises a given number of (in this case, eighteen) N-channel transistors 11 (some short-circuited or absent, depending on the logic function implemented by the desired program) connected in series with one another, with the gate terminals connected to the inputs of AND plane 3 (natural and/or inverted outputs of input buffer stage 2). Between the bottom transistor 11 of each minterm 10 and ground, there is provided a first N-channel precharging transistor 12 with the gate terminal connected to an input 13 of AND plane 3, supplied with an AND plane precharging signal CPPA generated by a circuit described later on with reference to FIG. 2. A second P-channel precharging transistor 15 is interposed between the top transistor 11 of each minterm 10 and supply line 14, and also presents the gate terminal connected to input 13 and supplied with precharging signal CPPA. Node 16 between top transistor 11 and second precharging transistor 15 defines the output of minterm 10, and is connected to an inverter 17 for resetting the AND logic from the dynamic NAND configuration implemented by minterms 10. The outputs 17a of inverters 17 (only one shown) represent the output nodes of the AND plane. Between output 17a and ground, there exits a capacitance associated with the output node shown as capacitor 18.

OR plane 4 comprises a number of N-channel MOS transistors 19 OR-connected in groups, i.e. with the gate terminals connected in known manner to the inputs of OR plane 4, the source terminals grounded, and the drain terminals connected in groups to nodes 20. In known manner, some of transistors 19 may be absent, in which case, there is no connection between a predetermined output of the AND plane and a predetermined output of the OR plane. Between each node 20 and supply line 14, there is provided a P-channel precharging transistor 21, the gate terminal of which is supplied with an OR plane precharging signal CPPO. A capacitance 22 is interposed between node 20 and ground, represented by capacitor 22 being the equivalent capacitance of node 20. Node 20 is connected to an inverter 23 formed by two opposite transistors 23a and 23b. More specifically, transistor 23a is a P-channel type with the source terminal connected to supply line 14; the gate terminal connected to node 20 (together with the gate terminal of transistor 23b); and the drain terminal connected to the drain terminal of transistor 23b, which is an N-channel type with a grounded source terminal. The common node 23c between transistors 23a and 23b defines an output of OR plane 4, and is connected to an input of output latch stage 5.

Output latch stage 5 comprises a number of latch circuits 24, each comprising a first switch 25a supplied with two opposite storage enabling signals CPM and CPMN for closing switch 25a when CPM is high, and opening it when CPMN is high. A first inverter 26 is provided downstream from switch 25a, with its output 26a connected to a feedback inverter 27, the output of which is connected via a second switch 25b to the input 26b of first inverter 26. Second switch 25b is controlled in an opposite way with respect to first switch 25a by signals CPM and CPMN, so that it is open when signal CPM is high, and closed when CPMN is high. A third inverter 28 is input-connected to node 26a of first inverter 26, and defines, with its output 28a, one of the outputs of latch stage 5. A capacitance is interposed between output node 28a and ground, represented as the capacitor 29 being the equivalent capacitance of node 28a.

FIG. 2 shows the generator 30 for generating the timing signals for operating the programmable logic array in FIG. 1. As shown, generator 30 is composed of an array duplicating section 31 reproducing logic array structure 1 and including stages 3-5; and a monostable section 32 in which the PLA 1 evaluating and memorizing signals are generated.

As shown in FIG. 2, array duplicating section 31 comprises elements exactly duplicating the dimensions of corresponding elements in array 1, so that the propagation delays of generator 30 are identical to those governing precharging, evaluating and latching of array 1. Monostable section 32, on the other hand, generates the timing signals as of low-to-high switching of an external clock signal CP, taking into account the delays in the array duplicating section, and in such a manner that reading of the array lasts only as long as necessary, regardless of the period of clock signal CP. In this way, the PLA 1 signal sequence depends on a single switching edge of the clock signal.

More specifically, array duplicating section 31 comprises an AND plane duplicating portion 33 including a number of transistors 34 connected in series, with the gate terminals connected to one another and to supply line V.sub.DD corresponding to line 14 in FIG. 1. Transistors 34 correspond to transistors 11 of each minterm 10 of AND plane 3 of array 1. The source terminal of the bottom transistor 34 is grounded via an N-channel precharging transistor 35 supplied at the gate terminal with signal CPPA enabling precharging of the AND plane. The drain terminal of top transistor 34 (node 33a), which supplies signal A, is connected to supply line V.sub.DD via a P-channel precharging transistor 36 supplied at the gate terminal with signal CPPA. Node 33a is connected to an inverter 37, the output of which defines an output node 37a of AND plane duplicating portion 33 (supplying signal B) and is connected to the input of an OR plane duplicating portion 38. Portion 38 comprises a transistor 39 with the gate terminal connected to node 37a, the source terminal grounded, and the drain terminal defining a node 40 (supplying signal C). Node 40 is connected to supply line V.sub.DD via a P-channel precharging transistor 41 supplied at the gate terminal with signal CPPO enabling precharging of the OR plane. Node 40 is connected to an inverter 42 formed by a pair of complementary transistors 43 and 44 connected in series, with the gate terminals connected to node 40. More specifically, transistor 43 is a P-channel type, with the source terminal connected to supply line V.sub.DD and the drain terminal connected to the drain terminal of transistor 44 (N-channel type) and defining the output node 45 of OR plane duplicating portion 38 (at which signal O is supplied). The source terminal of transistor 44 is grounded.

In FIG. 2, capacitor 46, connected between node 37a and ground, simulates the parasitic capacitance associated with all the transistors 19 of OR plane 4 of PLA 1 to be driven by minterm 10; and capacitor 47 shown connected between node 40 and ground simulates the parasitic capacitance seen by nodes 20 of OR plane 4 (capacitance 22).

Node 45 is connected to an output latch duplicating circuit 48 having the same structure as latches 24 of PLA 1, for simulating the signal propagation delay through stage 5 of array 1, and ensuring the outputs of OR plane 4 of array 1 are memorized correctly in latches 24 before array 1 returns to the precharge state.

More specifically, circuit 48 comprises a switch 49 controlled by a storage enabling signal CPM so as to close when signal CPM is high, and open when inverted signal CPMN is high. Switch 49 is interposed between output 45 and a node 50 connected to one input of an AND gate 51, the second input of which is supplied with external clock signal CP. The output of AND gate 51 is connected to one input of a NOR gate 52, the other input of which is supplied with a signal SR generated by monostable section 32. The output of NOR gate 52 is connected to the input of an inverter 53, the output of which defines a node 54 and is connected to an inverter 55, the output of which defines a node 56 supplying a signal 01. A capacitor 57 is provided between node 56 and ground. Node 54 is also connected to node 50 via a switch 58 supplied with signals CPM and CPMN and operating in opposition to switch 49, i.e. is closed when signal CPM is low, and open when signal CPMN is high, so that node 50 is supplied alternatively with signal O at output 45 of OR plane duplicating portion 38, or the signal at node 54.

Monostable section 32 comprises a branch 62 for generating a delayed reset signal, and including four cascade connected inverters 63-66, the most upstream one of which 63 is supplied with a reset signal R which is doubly inverted by inverters 63 and 64. A capacitor 67 is provided between the output of inverter 64 and ground, for delaying the switching edge of the reset signal. The output of inverter 64 is also connected to inverter 65, the output node 68 of which thus presents a delayed inverted reset signal SRN which, inverted by inverter 66, produces a delayed reset signal SR which is supplied to NOR circuit 52.

Monostable section 32 also comprises a branch 72 for generating storage enabling signal CPM; and a branch 73 for generating precharging signals CPPA and CPPO of the AND and OR planes.

More specifically, branch 72 comprises an AND circuit 75 having two inputs connected respectively to nodes 45 and 56, and the output of which is connected to one input of a NOR circuit 76 supplied at a second input with reset signal R. The output of NOR circuit 76 is connected to an inverter 77, the output of which presents storage enabling signal CPM as well as a parasitic capacitance shown schematically by capacitor 78. The output of inverter 77 is connected to a further inverter 79, the output of which thus presents an inverted enabling signal CPMN as well as a further parasitic capacitance (shown schematically by capacitor 80).

Branch 73 comprises an OR gate 84 with two inputs supplied respectively with signal O1 and signal CPM. The output of OR gate 84 is connected to one input of a NAND gate 85 having a further two inputs supplied respectively with external clock signal CP and delayed inverted reset signal SRN. The output of NAND gate 85 is connected to an inverter 86 generating AND plane precharge enabling signal CPPA, and the output of which presents a capacitor 87 representing a parasitic capacitance. The output of inverter 86 is connected to a further inverter 88, the output of which presents inverted enabling signal CPPN and a parasitic capacitance shown schematically as capacitor 89. The output of inverter 88 is also connected to an inverter 90, the output of which presents OR plane precharge enabling signal CPPO as well as a parasitic capacitance (shown schematically as capacitor 91).

Operation of generator 30 generating the timing signals of array 1 will now be described with reference also to FIG. 3 which shows a plot of some of the main signals of generator 30.

The signal sequence commences upon high-to-low switching of reset signal R, after which it is governed by periodic low-to-high switching of clock signal CP.

Initially, between t.sub.0 and t.sub.1, reset signal R is high, so that SRN is low, SR is high, the output of NOR circuit 76 is low, signal CPM is high, and signal CPMN low. This condition is desired so that outputs 28a of output latch stage 5 of array 1 are all charged to "0" as explained below. Since SRN is low, the output of NAND circuit 85 is high, and the output of inverter 86 and signals CPPA and CPPO are low. Consequently, array 1 is in precharge mode (transistors 15, 21 on, transistor 12 off); transistor 36 is on; signal A at node 33a is high; signal B is low; transistors 39 and 41 are respectively off and on; signal C is high; signal O is low; switch 49 is closed; switch 58 is open; the output of gate 51 is low; but the output of NOR gate 52 is low due to signal SR being high, so that signal O1 is low. Moreover, in PLA 1, precharging transistors 21 are on, so that nodes 20 are high, nodes 23c are low, switches 25a are closed, and switches 25b open, so that outputs 28a are charged to "0".

At instant t.sub.1, reset signal R switches to low, at which point clock signal CP may be low or high. A low clock signal poses no problem; whereas, in the event of a high clock signal (as shown in FIG. 3 to indicate the characteristics of the generator) switching of the reset signal must be prevented from being interpreted as the leading edge of signal CP. For this purpose, generator 30 in FIG. 2 presents a delayed reset signal SR which switches with a delay with respect to reset signal R for preventing circuit error as explained below.

More specifically, between t.sub.1 and t.sub.2 (instant at which delayed signal SR switches), switching of R causes NOR gate 76 (which sees two "0" at the input) to switch to high and inverter 77 to low, so that signal CPM switches to low and signal CPMN to high. Switching of R, however, has no effect on branch 73, due to signal SRN, in the above time interval, remaining low and so preventing erroneous switching of NAND gate 85. Similarly, no change occurs in the state of circuit 48, due to signal SR maintaining a low output of NOR gate 52 and hence a high output 54. Switching of CPM, however, opens switch 49, and closes switch 58 to feed back its output (node 54) and isolate it from node 45, so that the inputs of AND gate 51 present two high signals (CP and the output of inverter 53).

At instant t.sub.2, SR also switches to low, but with no change in the state of circuit 30, due to no change occurring in the state of NAND gate 85 (low output of OR gate 84, due to O1 and, by now, also CPM being low) or in the state of NOR gate 52 (SR has switched to low, but the output of gate 51 is high). Conversely, if CP is already low when SR switches, resetting of output latch duplicating circuit 48 is commenced immediately, as described below with reference to interval t.sub.3 -t.sub.4.

Upon clock signal CP switching to low (instant t.sub.3), AND gate 51 switches to low and, after a delay depending on the capacitance of capacitor 57, signal O1 switches to high (resetting latch duplicating circuit 48). The rest of generator 30, however, remains in the same state as previously. More specifically, no switching occurs of gate 85, due to CP first switching to zero and O1 then switching to "1".

Upon CP switching back to high (instant t.sub.4), a sequence is enabled for evaluating the AND and OR planes of array 1 and memorizing the results (array reading).

More specifically, upon CP switching to "1", CPPA immediately switches to high (O1 and SRN high) thus terminating the precharge signal, so that, in array 1 in FIG. 1, second precharging transistor 15 is turned off (as is transistor 36 in portion 33 in FIG. 2), first precharging transistor 12 is turned on (as is transistor 35 in portion 33), and the PLA 1 input configuration at each minterm is evaluated in AND plane 3. Shortly after CPPA switches, CPPO also switches (instant t.sub.5), with a delay with respect to CPPA due to the double inversion by inverters 88, 90 and the presence of capacitor 89, so as to commence evaluation of OR plane 4. In practice, switching of signal CP results in the switching of predetermined outputs of the AND and OR planes and respective duplicating sections according to the sequence described below. More specifically, in PLA 1, the output nodes 16 of minterms 10 whose transistors 11 are all turned on (high gate terminal) switch to low (like node 33a in FIG. 2--see signal A in FIG. 3), so that respective nodes 17a switch to high (like node 37a in FIG. 2--see signal B in FIG. 3) thus permitting evaluation of OR plane 4 as soon as signal CPPO also switches. Subsequently, precharging transistors 21 (FIG. 1) and 41 (FIG. 2) are turned off, and OR plane transistors 19 connected to the enabled nodes 17a are turned on, thus switching the respective node 20 to low (just as transistor 39 is turned on to switch signal C to low in FIG. 3); and the enabled output nodes 23c of OR plane 4 and node 45 (supplying signal O) switch to high (see also FIG. 3).

Low-to-high switching of signal O enables the storage signal. More specifically, in FIG. 2, the output of AND gate 75 switches to high (switch 49 is still open, and O1 still high) followed by switching of NOR gate 76, the output of which switches to low, and of signals CPM and CPMN (instant t.sub.6, FIG. 3) which close switches 25a of latches 24 and switch 49 of circuit 48, and open switches 25b of latches 24 and switch 58 of circuit 48, so that the output signals of the OR plane and signal O may be sampled and loaded into the respective latches. In particular, propagation of signal O in circuit 48 (which reproduces the propagation delay of the OR plane outputs of PLA 1) switches AND gate 51 to high, NOR gate 52 to low, node 54 to high, and output 56 (signal O1) to low (FIG. 3).

With a delay depending on the capacitance of capacitor 57, high-to-low switching of O1 switches the output of AND gate 75 to low, gate 76 to high, and CPM to low (instant t.sub.7) thus terminating the latching signal, so that switches 25a of latches 24 in FIG. 1 open (as does switch 49 in circuit 48), and switches 25b of latches 24 close (as does switch 58) to freeze the respective output at nodes 28a and 56.

Switching of CPM restores the circuit to its initial state by switching OR gate 84 to low, NAND gate 85 to high, and signals CPPA and CPPO successively to low; and PLA 1 (and duplicating section 31) is restored to precharge mode by switching the formerly enabled outputs 16, 17a, 20, 23c (and 33a, 37a, 40, 45). As such, reading (including evaluation and latching) lasts only as long as strictly necessary, thus minimizing discharging of the non-enabled minterm outputs. Moreover, in the present precharge phase, only the minterms 10 discharged in the newly terminated read phase are charged from zero, while the charge of any nodes not involved in the reading but which may have been discharged capacitively to a small extent is completed, also taking into account the high fields and voltages present in the chip integrating the memory.

Subsequently (instant t.sub.8), high-to-low switching of signal CP again resets latch duplicating circuit 48 (O1 switches to high) as already described for interval t.sub.3 -t.sub.4.

As will be clear from the foregoing description, generator 30 generates the various signals required for reading PLA 1 by so separating them as to permit safe evaluation of both AND plane 3 and OR plane 4 of PLA 1. The output of AND plane duplicating portion 33 (signal A) in fact switches to zero at the speed permitted by discharging of the parasitic capacitance due to series transistors 34 and corresponding to the parasitic capacitance due to series transistors 11 of AND plane 3; node 37a supplying signal B is charged capacitively like the corresponding node 17a of AND plane 3; and node 40 supplying signal C is also charged like output nodes 20 of OR plane 4. All the circuit component elements of array duplicating section 31 may advantageously form part (in the device layout) of PLA 1 itself, in which case, reproduction of the evaluation delays is guaranteed.

As already stated, signal CPPO is generated as of signal CPPA to ensure node 40 supplying signal C remains in precharge mode throughout evaluation of AND plane 3. In any case, signal B may only switch to high if signal A switches to low, thus eliminating any possibility that signal B may damage a "1" value of signal C due solely to the stored charge of capacitance 47 connected to node 40.

Moreover, the output latch signal from PLA 1 may only commence upon signal C switching to zero as a result of signal O switching and signal O1 being reset and so switching CPM.

As already stated, at this point, circuit 48 provides for defining the "1" duration of the latch signal (CPM) by duplicating the propagation times in latches 24 and by means of capacitor 57 which takes into account the capacitance of the line connected to output 28a of each latch 24, for ensuring the output of PLA 1 is memorized correctly in latches 24 before PLA 1 returns to precharge mode.

The FIG. 2 circuit advantageously presents capacitors 57, 89 and 67, the value of which may be so determined at design level as to allow the circuit a certain margin of safety in the definition of the delays between the switching edges of the signals, to take into account the size of the specific array and the nonvolatile memory comprising the array and generator 30. More specifically, capacitor 57 at node 56 increases the "1" duration of signal CPM at the latching phase to allow for the rise time of outputs 28a of PLA 1 (depending on the capacitive charge seen by outputs 28a); capacitor 89 at the output of inverter 88 provides for regulating the switch delay of CPPO with respect to CPPA (interval t.sub.4 -t.sub.5); and capacitor 67 at the output of inverter 64 provides for regulating the delay of SRN and SR with respect to reset signal R.

As already stated, the "monostable" solution described for reading dynamic PLA 1 therefore presents the advantage that reading, i.e. the evaluation of nodes whose logic value depends on the parasitic capacitance of CMOS gates (subject to unpredictable losses), lasts only as long as strictly necessary, regardless of the period of the PLA clock signal (CP), so that a dynamic AND-OR structure may safely be implemented for the PLA despite the field and voltage conditions of the nonvolatile arrays involved.

Circuit 2 also provides for simplifying the structure of stage 5, while at the same time ensuring resetting. In fact, as described above in connection with interval t.sub.0 -t.sub.1, signal R provides for resetting the PLA and switching all the outputs 28a of output latch stage 5 to "0" with no need for special reset elements or inputs, thus simplifying the structure and enabling considerable saving in terms of layout.

Clearly, changes may be made to the array and clock generator as described and illustrated herein without, however, departing from the scope of the present invention.