PRELIMINARY  
					CYM26KNP72AV25  
					mode, a HIGH selects an interleaved burst sequence. Both  
					burst counters use A0 and A1 in the burst sequence, and will  
					wrap-around when incremented sufficiently. A HIGH input on  
					ADV/LD will increment the internal burst counter regardless of  
					the state of chip enables inputs or WE. WE is latched at the  
					beginning of a burst cycle. Therefore, the type of access (Read  
					or Write) is maintained throughout the burst sequence.  
					Introduction  
					Functional Overview  
					The CYM26KNP72AV25 is a synchronous-pipelined Burst  
					NoBL SRAM module designed specifically to eliminate wait  
					states during Write/Read transitions. All synchronous inputs  
					pass through input registers controlled by the rising edge of  
					the clock. The clock signal is qualified with the Clock Enable  
					input signal (CEN). If CEN is HIGH, the clock signal is not  
					recognized and all internal states are maintained. All synchro-  
					nous operations are qualified with CEN. All data outputs pass  
					through output registers controlled by the rising edge of the  
					clock. Maximum access delay from the clock rise (tCO) is  
					4.2 ns (133-MHz device).  
					Single Write Accesses  
					Write access are initiated when the following conditions are  
					satisfied at clock rise: (1) CEN is asserted LOW, (2) CE1, CE2,  
					and CE3 are ALL asserted active, and (3) the write signal WE  
					is asserted LOW. The address presented to Ax is loaded into  
					the Address Register. The write signals are latched into the  
					Control Logic block.  
					Accesses can be initiated by asserting all three Chip Enables  
					(CE1, CE2, CE3) active at the rising edge of the clock. If Clock  
					Enable (CEN) is active LOW and ADV/LD is asserted LOW,  
					the address presented to the device will be latched. The ac-  
					cess can either be a read or write operation, depending on the  
					status of the Write Enable (WE). BWS[H:a] can be used to con-  
					duct byte write operations.  
					On the subsequent clock rise the data lines are automatically  
					three-stated regardless of the state of the OE input signal. This  
					allows the external logic to present the data on DQ and DQP  
					(DQa-h/DPa-h for CYM26KNP72AV25). In addition, the ad-  
					dress for the subsequent access (Read/Write/Deselect) is  
					latched into the Address Register (provided the appropriate  
					control signals are asserted).  
					Write operations are qualified by the Write Enable (WE). All  
					writes are simplified with on-chip synchronous self-timed write  
					circuitry.  
					On the next clock rise the data presented to DQ and DP (DQa-  
					h/DPa-h for CYM26KNP72AV25) (or a subset for byte write  
					operations, see Write Cycle Description table for details) in-  
					puts is latched into the device and the write is complete.  
					Three synchronous Chip Enables (CE1, CE2, CE3) and an  
					asynchronous Output Enable (OE) simplify depth expansion.  
					All operations (Reads, Writes, and Deselects) are pipelined.  
					ADV/LD should be driven LOW once the device has been de-  
					selected in order to load a new address for the next operation.  
					The data written during the Write operation is controlled by  
					BWS (BWSa-h for CYM26KNP72AV25)  
					signals. The  
					CYM26KNP72AV25 provides byte write capability that is de-  
					scribed in the Write Cycle Description table. Asserting the  
					Write Enable input (WE) with the selected Byte Write Select  
					(BWS) input will selectively write to only the desired bytes.  
					Bytes not selected during a byte write operation will remain  
					unaltered. A Synchronous self-timed write mechanism has  
					been provided to simplify the write operations. Byte write ca-  
					pability has been included in order to greatly simplify  
					Read/Modify/Write sequences, which can be reduced to sim-  
					ple byte write operations.  
					Single Read Accesses  
					A read access is initiated when the following conditions are  
					satisfied at clock rise: (1) CEN is asserted LOW, (2) CE1, CE2,  
					and CE3 are ALL asserted active, (3) the Write Enable input  
					signal WE is deasserted HIGH, and (4) ADV/LD is asserted  
					LOW. The address presented to the address inputs is latched  
					into the Address Register and presented to the memory core  
					and control logic. The control logic determines that a read ac-  
					cess is in progress and allows the requested data to propagate  
					to the input of the output register. At the rising edge of the next  
					clock the requested data is allowed to propagate through the  
					output register and onto the data bus within 3.8 ns (150-MHz  
					device) provided OE is active LOW. After the first clock of the  
					read access the output buffers are controlled by OE and the  
					internal control logic. OE must be driven LOW in order for the  
					device to drive out the requested data. During the second  
					clock, a subsequent operation (Read/Write/Deselect) can be  
					initiated. Deselecting the device is also pipelined. Therefore,  
					when the SRAM is deselected at clock rise by one of the chip  
					enable signals, its output will three-state following the next  
					clock rise.  
					Because the CYM26KNP72AV25 is a common I/O device,  
					data should not be driven into the device while the outputs are  
					active. The Output Enable (OE) can be deasserted HIGH be-  
					fore presenting data to the DQ and DP (DQa-h/DPa-h for  
					CYM26KNP72AV25 ) inputs. Doing so will three-state the out-  
					put drivers. As a safety precaution, DQ and DP (DQa-h/DPa-h  
					for CYM26KNP72AV25 are automatically three-stated during  
					the data portion of a write cycle, regardless of the state of OE.  
					Burst Write Accesses  
					The CYM26KNP72AV25 has an on-chip burst counter that  
					allows the user the ability to supply a single address and con-  
					duct up to four WRITE operations without reasserting the ad-  
					dress inputs. ADV/LD must be driven LOW in order to load the  
					initial address, as described in the Single Write Access section  
					above. When ADV/LD is driven HIGH on the subsequent clock  
					rise, the chip enables (CE1, CE2, and CE3) and WE inputs are  
					ignored and the burst counter is incremented. The correct  
					BWS (BWSa-h for CYM26KNP72AV25 ) inputs must be driven  
					in each cycle of the burst write in order to write the correct  
					bytes of data.  
					Burst Read Accesses  
					The CYM26KNP72AV25 has an on-chip burst counter that  
					allows the user to supply a single address and conduct up to  
					four Reads without reasserting the address inputs. ADV/LD  
					must be driven LOW in order to load a new address into the  
					SRAM, as described in the Single Read Access section above.  
					The sequence of the burst counter is determined by the MODE  
					input signal. A LOW input on MODE selects a linear burst  
					Document #: 38-05111 Rev. **  
					Page 5 of 25