JP2012248082A - Receiver circuit, system device and timing adjusting method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To adjust timing to take in data.SOLUTION: A receiver circuit includes: a mask generation section 37 of a memory interface circuit 12 for generating an inside mask signal CKM corresponding to a duration for receiving a data stream DQ on the basis of a strobe signal DQS (an inside strobe signal IRS) in synchronization with the data stream DQ; a DLL circuit 41 for generating an internal clock signal CKS obtained by delaying an internal clock signal CK2 according to a set value of a resister 41a; flip-flop circuits 33 and 34 for latching a reception data stream RDQ according to an inside strobing signal CST obtained by combining the internal mask signal CKM with the internal clock signal CKS by an AND circuit 38; and an adjustment circuit 50 for detecting the phase relationship between the internal mask signal CKM and the inside strobe signal CST on the basis of output signals from the DLL circuit 51 and 52, to update the set value of the register 41a to adjust a phase of the inside strobe signal CST.

Description

  The present invention relates to a receiving circuit, a system device, and a timing adjustment method.

  Conventionally, a DRAM (Dynamic Random Access Memory) is used as a semiconductor memory device. In recent years, a double data rate method in which data is input / output at both the rising and falling edges of the clock has been adopted in order to cope with an increase in the operating speed of the system. Such a semiconductor memory device is called DDR-SDRAM (Double Data Rate Synchronous Dynamic Random Access Memory), DDR2-SDRAM, or DDR3-SDRAM.

  For example, when the system apparatus reads data from a DDR-SDRAM (memory), the DDR-SDRAM outputs read data and a data strobe signal synchronized with the read data. The receiving circuit in the system apparatus takes in the read data based on the data strobe signal or the clock signal whose timing is adjusted (see, for example, Patent Documents 1 and 2).

JP 2007-334943 A JP 2010-122842 A

  However, a relative timing shift between the data string DQ and the strobe signal DQS is also caused by a change in the environmental temperature of the system circuit or a change in the operating power supply voltage of the system circuit. Such a timing shift causes an error in the read data.

  According to one aspect of the present invention, a first signal generation circuit that generates an internal strobe signal corresponding to a clock signal, a latch circuit that latches data in response to the internal strobe signal, and an external strobe signal A second signal generation circuit that generates a mask signal according to a period for receiving the data, and an adjustment circuit that adjusts a phase of the internal strobe signal according to a phase relationship between the mask signal and the internal strobe signal. Have.

  According to one aspect of the present invention, it is possible to adjust the timing for capturing data.

1 is a schematic block diagram of a system. It is a partial block circuit diagram of a memory interface. It is a timing diagram of a read operation. It is a timing diagram which shows the signal which concerns on timing adjustment. It is a timing diagram which shows the signal which concerns on timing adjustment. It is a timing diagram which shows the signal which concerns on timing adjustment. It is a flowchart which shows a training sequence. (A) (b) is explanatory drawing of a training sequence. It is a timing diagram in a training sequence. (A)-(c) is explanatory drawing of a training sequence.

Hereinafter, an embodiment will be described with reference to the accompanying drawings.
As shown in FIG. 1, the system includes a system circuit 10 and a memory 20 accessed by the system circuit 10. The system circuit 10 is, for example, one chip (System on Chip: SoC), and is formed in a package (PKG) having a predetermined structure (for example, BGA: Ball Grid Array). The memory 20 is a synchronous semiconductor memory device, for example, a double data rate dynamic random access memory (DDR3-SDRAM). 20 is formed in a package having a predetermined structure (for example, BGA) including a chip, and this package is mounted on the substrate.The system circuit 10 is an example of a system device, and the memory 20 is an example of a target circuit. It is.

  The core circuit 11 of the system circuit 10 is a circuit including, for example, a central processing unit (CPU) and a memory controller. The core circuit 11 supplies the clock signal CK to the memory 20. Further, the core circuit 11 supplies a command for the memory 20 to the memory 20 via the memory interface circuit (denoted as memory I / F) 12. Further, the system circuit 10 supplies an address, data, and the like to the memory 20 according to the type of command. For example, the system circuit 10 outputs a read command for reading data in the memory 20 and an address at which the data is stored to the memory interface circuit 12. Further, the core circuit 11 outputs a write command for writing data to the memory 20, data, and an address for storing the data to the memory interface circuit 12.

  The memory interface circuit 12 outputs a strobe signal DQS generated based on the clock signal CK to the memory 20. The memory interface circuit 12 outputs the data string DQ to the memory 20 in synchronization with the strobe signal DQS. In response to the write command, the memory 20 stores the data string DQ captured based on the strobe signal DQS. Further, the memory 20 outputs the strobe signal DQS and the data string DQ generated based on the clock signal CK in response to the read command. The memory interface circuit 12 fetches the data string DQ based on the strobe signal DQS and outputs the fetched data string DQ to the core circuit 11.

  In addition, the system circuit 10 includes a training circuit 13 that adjusts the timing at which the memory interface circuit 12 captures the data string DQ. The training circuit 13 adjusts the timing for taking in the data string DQ at a predetermined timing (for example, when the operation of the system circuit 10 is started). For example, the training circuit 13 adjusts the phase of the strobe signal for capturing data by changing the delay time of the delay circuit included in the memory interface circuit 12 and repeating the operation of reading data of a predetermined value from the memory 20. .

Next, a circuit unit that captures the data string DQ in the memory interface circuit 12 is described.
As shown in FIG. 2, the data string DQ is supplied to the buffer circuit 31 of the memory interface circuit 12 via the input terminal P <b> 1 of the memory interface circuit 12. The strobe signal DQS is supplied to the buffer circuit 32 of the memory interface circuit 12 via the input terminal P2.

  The output terminal of the buffer circuit 31 is connected to data input terminals of D-type flip-flop circuits (FF circuits) 33 and 34. The buffer circuit 31 outputs the received data string RDQ in response to the data string DQ. The D-type flip-flop circuits 33 and 34 are an example of a latch circuit.

The output terminal of the buffer circuit 32 is connected to the AND circuit 35. The buffer circuit 32 outputs a reception strobe signal RDS in response to the strobe signal DQS.
The AND circuit 35 is supplied with a mask shift signal DMS generated by a mask generation unit 39 and a delay locked loop (DLL) circuit 40. The AND circuit 35 performs an AND operation on the reception strobe signal RDS and the mask shift signal DMS, and outputs an internal strobe signal IRS corresponding to the calculation result. The internal strobe signal IRS is supplied to the counter 36 and the mask generator 37.

  The mask generation unit 39 generates a strobe mask signal DSM based on the internal strobe signal IRS. The transmission path for transmitting the strobe signal DQS shown in FIG. 1 is used as a transmission path for bidirectional communication, and is in a high impedance state when not communicating. The high impedance state of the transmission line is set by a communication circuit included in the system circuit 10 (memory interface circuit 12) and a communication circuit included in the memory 20. Depending on the settings of the communication circuit that outputs data and the input circuit that inputs data, the transmission path is roughly divided into three states: a preamble period, a data transmission period, and a postamble period. For example, the system circuit 10 shown in FIG. 1 enables a termination circuit (ODT: On Die Termination) in preparation for data input. Thereby, the transmission line becomes the first level (for example, a level that is ½ of the amplitude of the transmission line), and transitions to the second level (for example, the L level). Thereafter, the memory 20 changes the strobe signal DQS according to the clock signal CK and outputs the data string DQ. When a predetermined number of data strings DQ are output, the transmission line transitions to the second level by the termination circuit (ODT) of the system circuit 10. Thereafter, when the termination circuit (ODT) is set to an invalid state, the transmission line is in a high impedance state. The mask generation unit 39 shown in FIG. 2 detects the preamble period (second level) based on the level of the internal strobe signal IRS, and outputs an H level strobe mask signal DSM. The mask generation unit detects a postamble (transition to the second level) and outputs an L-level strobe mask signal DSM. The mask generation unit 39 is an example of a detection circuit. A detection circuit for detecting the preamble and postamble may be provided separately from the mask generation unit 39, and the mask generation unit 39 may generate the strobe mask signal DSM according to the output of the detection circuit.

  The DLL circuit 40 outputs a mask shift signal DMS obtained by delaying (phase shifting) the input signal (strobe mask signal DSM) for a time corresponding to the set value (delay code DC1) stored in the register 40a. The mask shift signal DMS is supplied to the AND circuit 35 described above.

  For example, a burst length (BL) BLN is set in the counter 36 from the core circuit 11 shown in FIG. The burst length BLN is the number of data that the memory 20 continuously inputs and outputs, and the setting value is “4”, for example. The counter 36 is supplied with the internal clock signal CKS generated from the internal clock signal CK2 by the DLL circuit 41.

  The frequency of the internal clock signal CK2 is the same as the frequency of the clock signal CK supplied to the memory 20. The DLL circuit 41 delays the internal clock signal CK2 by a time corresponding to the set value (delay code DC2) stored in the register 41a, and generates a phase-shifted internal clock signal CKS.

  The counter 36 operates based on the internal clock signal CKS and counts the internal strobe signal IRS. The counter 36 outputs an end detection signal BLE when the count value becomes equal to the set burst length BLN. The end detection signal BLE is supplied to the mask generation unit 37.

  When the end detection signal BLE is at the L level, the mask generation unit 37 outputs the H level internal mask signal CKM in response to the H level internal strobe signal IRS. As a result, the mask generator 37 raises the internal mask signal CKM to H level at the first rising timing of the internal strobe signal IRS. Further, the mask generation unit 37 outputs an internal mask signal CKM at the L level in response to the end detection signal BLE at the H level. The AND circuit 35, the counter 36, and the mask generation unit 37 are included in the second signal generation circuit.

  The AND circuit 38 is supplied with the shifted internal clock signal CKS and the internal mask signal CKM output from the mask generation unit 37. The AND circuit 38 performs an operation (logical product operation) on the internal clock signal CKS and the internal mask signal CKM, and outputs an internal strobe signal CST corresponding to the operation result. For example, the AND circuit 38 outputs the L-level internal strobe signal CST in response to the L-level internal mask signal CKM, and responds to the H-level internal mask signal CKM in substantially the same timing as the internal clock signal CKS. The internal strobe signal CST which rises / falls at is output. The DLL circuit 41 and the AND circuit 38 are included in the first signal generation circuit. The AND circuit 38 is an example of a synthesis circuit. The internal strobe signal CST is supplied to the clock input terminals of the flip-flop circuits 33 and 34.

  The flip-flop circuit 33 latches the received data string RDQ in response to the internal strobe signal CST at the H level, and outputs read data RD1 having a level equal to the latched level. The flip-flop circuit 34 latches the received data string RDQ in response to the L-level internal strobe signal CST, and outputs read data RD2 having a level equal to the latched level.

  That is, the flip-flop circuit 33 latches the received data string RDQ at the rising timing of the internal strobe signal CST. The flip-flop circuit 34 latches the received data string RDQ at the falling timing of the internal strobe signal CST. Therefore, the timing at which the flip-flop circuit 33 latches the received data string RDQ and the timing at which the flip-flop circuit 34 latches the received data string RDQ are shifted from each other by 180 degrees.

  Therefore, the reception circuit included in the memory interface circuit 12 takes in the data string DQ (reception data string RDQ) in response to the internal strobe signal CST generated based on the internal clock signal CK2. Therefore, the data string DQ can be captured without being affected by skew generated on the transmission path between the system circuit 10 and the memory 20 shown in FIG.

  The internal strobe signal CST changes at the same timing as the shifted internal clock signal CKS. The internal clock signal CKS is generated by the DLL circuit 41 according to the set value stored in the register 41a based on the internal clock signal CK2.

Next, adjustment of the timing of the internal clock signal CKS and the mask shift signal DMS will be described.
The internal mask signal CKM generated by the mask generator 37 is supplied to the DLL circuit 51 of the adjustment circuit 50. The DLL circuit 51 gives a first delay amount (for example, 2/4 period (180 degrees) of the phase of the internal clock signal CK2) to the internal mask signal CKM to generate the delay mask signal DCM. The DLL circuit 51 is an example of a first delay circuit.

  The internal strobe signal CST output from the AND circuit 38 is supplied to the DLL circuit 52. The DLL circuit 52 gives the second delay amount (for example, a quarter period (90 degrees) of the phase of the internal clock signal CK2) to the internal strobe signal CST to generate the delayed strobe signal DST. The DLL circuit 52 is an example of a second delay circuit.

  The phase detector 53 detects the phase relationship between the delayed strobe signal DST and the delayed mask signal DCM, and generates a phase detection signal PDR corresponding to the detection result. Specifically, the phase detection unit 53 determines the phase relationship between the first rising edge (1st Rise Edge) of the delay strobe signal DST and the delay mask signal DCM (advance / delay of the delay mask signal DCM relative to the 1st Rise Edge of the strobe signal DST). To detect. And the phase detection part 53 outputs the phase detection signal PDR of the level according to a detection result.

  As described above, the internal mask signal CKM rises to the H level at the first rising timing of the internal strobe signal IRS. The internal strobe signal IRS corresponds to the strobe signal DQS output from the memory 20 shown in FIG. 1 and has the same phase as the data string DQ output from the memory 20. The internal strobe signal CST is 90 degrees behind the center of the data window of the data string DQ, that is, the data string DQ, as will be described later, in order to capture the data string DQ. Therefore, when there is no power supply voltage fluctuation or ambient temperature fluctuation, the phase of the delay mask signal DCM is equal to the phase of the delay strobe signal DST.

  For example, in the system shown in FIG. 1, the amount of delay in the transmission path between chips changes due to fluctuations in power supply voltage and ambient temperature. Due to the change in the delay amount, the timing of the reception strobe signal RDS is advanced or delayed with respect to the timing of the internal clock signal CKS. When the phase of reception strobe signal RDS advances from internal clock signal CKS, delay mask signal DCM is at the H level at the first rising timing of delay strobe signal DST. At this time, the phase detector 53 outputs an H level phase detection signal PDR.

  When the phase of the reception strobe signal RDS is delayed from the internal clock signal CKS, the delay mask signal DCM is at the L level at the first rising timing of the delay strobe signal DST. At this time, the phase detector 53 outputs an L level phase detection signal PDR.

The calculation unit 54 generates an update code UC based on the phase detection signal PDR output from the phase detection unit 53.
Then, the calculation unit 54 updates the set values of the registers 40a and 41a of the DLL circuits 40 and 41 with the generated update code UC at a timing that does not affect the reception of the data string DQ. For example, the end detection signal BLE is used to update the set values of the registers 40a and 41a. The end detection signal BLE indicates that the reception of data corresponding to the read command (read operation) has ended. Accordingly, the arithmetic unit 54 waits for the end of the read operation and updates the set values of the registers 40a and 41a.

  For example, the calculation unit 54 outputs the update code UC in response to the end detection signal BLE output from the counter 36. The end detection signal BLE indicates that the reception of data corresponding to the read command (read operation) has ended. Therefore, the arithmetic unit 54 waits for the end of the read operation and outputs the generated update code UC.

  Each of the first timing control unit 55 and the second timing control unit 56 receives the end detection signal BLE. The first timing control unit 55 stores the delay code DC1 corresponding to the update code UC in the register 40a at a timing based on a signal obtained by delaying the end detection signal BLE for a predetermined time. Similarly, the second timing control unit 56 stores the delay code DC2 corresponding to the update code UC in the register 41a at a timing based on a signal obtained by delaying the end detection signal BLE for a predetermined time. By adjusting the delay time for the end detection signal BLE, the setting timing of the delay code DC1 for the first register 40a and the setting timing of the delay code DC2 for the second register 41a are synchronized.

Next, operations of the memory interface circuit 12 and the training circuit 13 will be described.
[Outline of training process]
The memory interface circuit 12 and the training circuit 13 shown in FIG. 1 execute the following processes to adjust the timing for taking in the data string DQ.
(A) System startup.
(B) Gate training.
(C) Data Eye Traning.
(D) Delay code calculation (Read Delay Code Cal.).
(E) Delay code update (Read Delay Code Update).

  The training circuit 13 executes the process (b) and adjusts the set value stored in the register 40a of the DLL circuit 40 shown in FIG. Further, the training circuit 13 executes the process (c) and adjusts the set value stored in the register 41a of the DLL circuit 41 shown in FIG. The memory interface circuit 12 repeatedly executes the processes (d) and (e) to adjust the set values of the registers 40a and 41a.

[Timing adjustment (1)]
Next, the operation of (c) Data Eye Training will be described.
The training circuit 13 shown in FIG. 1 accesses the memory 20 via the memory interface circuit 12 according to the flowchart shown in FIG. 7, and adjusts the timing for taking in the data string DQ.

First, in the initialization process, the delay value N stored in the register 41a of the DLL circuit 41 that generates the internal clock signal CKS is set to an initial value (= 0) (step 61).
Next, in the determination process, it is determined whether or not the process is completed up to a predetermined delay value N (step 62). For example, the delay value N is changed from an initial value (= 0) to a value corresponding to a predetermined delay time (Delay). An example of the delay time N and the delay time (Delay) is shown in FIG. In this example, the delay time is adjusted to 8 stages. Accordingly, the following processing is repeatedly executed until the delay value N reaches “8”.

  Next, in the delay value setting process (Delay Set), the delay value N is set in the register 41a shown in FIG. 2 (step 63). In the example shown in FIG. 8A, when the delay value N is the initial value (= 0), the DLL circuit 41 shown in FIG. 2 has a delay time of “0” with respect to the internal clock signal CK2, that is, the internal clock. An internal clock signal CKS having the same timing as that of the signal CK2 is generated. When the delay value N is maximum (= 7), the DLL circuit 41 shown in FIG. 2 is delayed from the internal clock signal CK2 by a delay time of “140 psec (picoseconds)” with respect to the internal clock signal CK2. An internal clock signal CKS with timing is generated.

  Next, in a write process (Write), a write operation is performed on the memory 20 by the write command via the memory interface circuit 12 shown in FIG. 1 (step 64). At this time, the data string to be written is set such that a plurality of continuous data can be identified, and for example, each bit of the 8-bit data string DQ is “01010101”. Thereby, it becomes possible to identify continuous data (for example, data D0 (= 0) and data D1 (= 1)).

Next, in the read process (Read), the previously written data is read by a read command (step 65).
Next, it is determined whether the read data matches (Pass) or different (Fail) with the written data, and the determination results are sequentially stored in a register (step 66).

Next, the delay value N is switched (N = N + 1) (step 67), and the process proceeds to step 62.
When the reading of a predetermined number (for example, “8”) of the data string DQ is completed in step 62 (determination: YES), the delay value of the DLL circuit 41 shown in FIG. 2 is calculated based on the determination result sequentially stored in step 66. Set (step 68). For example, the delay value at the center of the path area (a plurality of delay values N determined to be paths) is set as a set value in the register 41a shown in FIG. For example, as shown in FIG. 8B, a pass (Pass) is determined in an area where the delay value N is “1” to “5”. Accordingly, the delay value N (= 3) at the center of the region “1” to “5” is calculated, and the calculated delay value is set in the register 41a shown in FIG.

In the timing adjustment described above, the adjustment range of the internal clock signal CKS is set to one cycle, and the delay time adjustment step should be as small as possible.
For example, as shown in FIG. 9, a plurality of internal clocks having different delay timings N and increasing the delay value N with respect to the internal clock signal CKS (waveform shown by the solid line) when the delay value N is the initial value (= 0). A signal CKS (waveform indicated by a broken line) is generated. A data string DQ read from the memory 20 shown in FIG. 1 is fetched at the timing of rising edges of a plurality of internal clock signals CKS having different timings, and the data (read data) and the data written to the memory 20 (write data) Compare When both data match, it is determined as a pass, and when both data are different from each other, it is determined as a fail.

  A central value among a plurality of delay values N corresponding to the internal clock signal CKS determined to be a path is set as a set value. Therefore, the step (delay time) for adjusting the delay time of the internal clock signal CKS is preferably fine (short). Thereby, in the flip-flop circuits 33 and 34 for capturing the data string DQ, the timing margin in the setup (Setup) and the timing margin in the hold (Hold) can be made equal to each other.

  Further, in the timing adjustment described above, among the data string DQ in which a plurality of data (for example, a number equal to the burst length) is continuous, the data excluding the top data, that is, the second and subsequent data are used for pass-fail judgment. Use it. By adjusting the timing of the internal strobe signal CST (internal clock signal CKS) using the second and subsequent data, the edge of the internal strobe signal CST is aligned near the center of the data valid period (data window). Can do.

  That is, the AND circuit 38 shown in FIG. 2 outputs the L level internal strobe signal CST in response to the L level internal mask signal CKM, and is equal to the internal clock signal CKS in response to the H level internal mask signal CKM. A level internal strobe signal CST is output. The rising timing of the internal mask signal CKM corresponds to the rising timing of the strobe signal DQS, and the rising timing of the strobe signal DQS is equal to the timing of the data string DQ. Therefore, as shown in FIG. 10A, by determining the pass-fail of the data string DQ (data D0) from the rising edge of the internal mask signal CKM, the period for determining the pass, that is, the start and end of the data window is determined. Can be determined.

  However, as shown in FIG. 10B, if the rising timing of the internal mask signal CKM is delayed due to some factor, the determination is performed from the middle of the data D0. As a result, the data window of the data D0 cannot be correctly determined. The rising timing of the internal strobe signal CST adjusted by the data window thus determined is delayed from the center of the correct data window, and the hold timing margin is reduced.

  For this reason, as shown in FIG. 10C, by performing pass-fail judgment on the data D1, it is possible to correctly judge the period for judging that the data D1 is a pass, that is, the start and end of the data window. Can do. Thereby, the edge of the internal strobe signal CST can be aligned near the center of the period (data window) in which the data is valid. The timing margin in the setup (Setup) and the timing margin in the hold (Hold) can be made equal to each other.

[Import data]
Next, an operation for taking in the data string DQ will be described.
As shown in FIG. 3, the preamble (transition to L level) of the reception strobe signal RDS is detected, and an H level strobe mask signal DSM is generated.

Based on the H level strobe mask signal DSM, the internal strobe signal IRS is output according to the reception strobe signal RDS.
When the mask generation unit 37 detects the first rising edge of the internal strobe signal IRS, it outputs the internal mask signal CKM at the H level. The AND circuit 38 outputs an internal strobe signal CST corresponding to the internal clock signal CKS based on the internal mask signal CKM at the H level. The received data string RDQ is latched at the rising timing and falling timing of the internal strobe signal CST, and output as data D0 to D7.

  The counter 36 counts the number of cycles (number of pulses) of the internal clock signal CKS in response to the internal strobe signal IRS at the H level. When the count value becomes equal to the burst length BLN, an H level end detection signal BLE is output based on the falling edge of the internal strobe signal IRS.

  The mask generation unit 37 outputs an internal mask signal CKM at L level based on the end detection signal BLE at H level. The AND circuit 38 outputs an L level internal strobe signal CST in response to the L level internal mask signal CKM.

[Timing adjustment (2)]
Next, operations of (d) delay code calculation and (e) delay code update will be described.
As shown in FIG. 4, the internal mask signal CKM is delayed by a first delay time (for example, the phase of the internal clock signal CK2 by 180 °) by the DLL circuit 51 to generate a delay mask signal DCM. The internal strobe signal CST is delayed by a second delay time (for example, the phase of the internal clock signal CK2 by 90 °) by the DLL circuit 52 to generate a delayed strobe signal DST.

  2 detects the phase relationship between the delayed strobe signal DST and the delayed mask signal DCM. For example, the phase detection unit 53 detects the first rising edge of the delay strobe signal DST, and detects the level of the delay mask signal DCM at the timing of the rising edge. Then, the phase detector 53 outputs a phase detection signal PDR having a level corresponding to the level of the detected delay mask signal DCM. For example, the phase detection unit 53 outputs an H level phase detection signal PDR when detecting an H level delay mask signal DCM, and outputs an L level phase detection signal PDR when detecting an L level delay mask signal DCM.

  Based on the H level phase detection signal PDR, the arithmetic unit 54 determines whether to increase or decrease the delay time for generating the mask shift signal DMS and the internal clock signal CKS at the time of the next read. Generate. The update code UC is a 1-bit signal, for example, and an update code UC of “1” is generated when the delay time is increased, and an update code UC of “0” is generated when the delay time is decreased.

  The timing control unit 55 updates the set value of the register 40a based on the update code UC output from the calculation unit 54. Similarly, the timing control unit 56 updates the set value of the register 41a based on the update code UC output from the calculation unit 54. The setting values stored in the registers 40a and 41a are set in the data eye training executed by the training circuit 13, as described in [Timing Adjustment (1)] above.

  When the operating environment (ambient temperature) of the system apparatus and the voltage of the operating power supply fluctuate (VT fluctuation), the timing of the internal strobe signal CST and the timing of the data string DQ and the strobe signal DQS change relatively. The internal mask signal CKM is generated based on the strobe signal DQS and changes in the same manner as the reception data string RDQ.

  For example, as shown in FIG. 5, when the reception strobe signal RDS (internal mask signal CKM) becomes earlier, the delay mask signal DCM is at the H level at the timing of the first rising edge of the delay strobe signal DST. A phase detection signal PDR is output. Based on the H level phase detection signal PDR, the calculation unit 54 outputs an update code UC of “1” (H level) so as to reduce the delay time for generating the internal clock signal CKS at the next read time. To do.

  The timing control unit 56 shown in FIG. 2 updates (decreases) the set value of the register 41a based on the update code UC output from the calculation unit 54. For example, the timing control unit 56 stores a value obtained by subtracting a predetermined adjustment step from the set value of the register 41a in the register 41a, that is, decreases the set value of the register 41a. As a result, the DLL circuit 41 advances the timing of the internal clock signal CKS generated with the delay time corresponding to the set value based on the internal clock signal CK2.

  Similarly, the timing control unit 55 updates (decreases) the set value of the register 40a based on the update code UC output from the calculation unit 54. As a result, the DLL circuit 40 advances the timing of the mask shift signal DMS generated with the delay time corresponding to the set value based on the strobe mask signal DSM.

  As shown in FIG. 6, when the reception strobe signal RDS (internal mask signal CKM) is delayed, the delay mask signal DCM is at the L level at the timing of the first rising edge of the delay strobe signal DST. A phase detection signal PDR is output. Based on the L-level phase detection signal PDR, the arithmetic unit 54 outputs an update code UC of “0” (H level) so as to increase the delay time for generating the internal clock signal CKS at the next read time. To do.

  The timing control unit 56 illustrated in FIG. 2 updates the set value of the register 41a based on the update code UC output from the calculation unit 54. For example, the timing control unit 56 stores a value obtained by adding a predetermined adjustment step from the set value of the register 41a in the register 41a, that is, increases the set value of the register 41a. Thereby, the DLL circuit 41 delays the timing of the internal clock signal CKS generated with a delay time corresponding to the set value based on the internal clock signal CK2.

  Similarly, the timing control unit 55 updates (increases) the set value of the register 40a based on the update code UC output from the calculation unit 54. As a result, the DLL circuit 40 delays the timing of the mask shift signal DMS generated with a delay time corresponding to the set value based on the strobe mask signal DSM.

  As described above, the adjustment circuit 50 sets the delay amount of the DLL circuit 41 that generates the internal clock signal CKS (the setting stored in the register 41a) so that the phase of the delay mask signal DCM and the phase of the delay strobe signal DST are equal. Value). As a result, the phase difference between the internal mask signal CKM and the internal clock signal CKS, that is, the phase difference between the internal mask signal CKM and the internal strobe signal CST becomes 90 degrees. RDQ) is the center of the data window.

As described above, according to the present embodiment, the following effects can be obtained.
(1) The internal clock signal CK2 is delayed by the DLL circuit 41 to generate the internal clock signal CKS. The internal clock signal CKS is masked with the internal mask signal CKM generated based on the strobe signal DQS to generate the internal strobe signal CST. The flip-flop circuit 33 latches the received data string RDQ at the timing of the rising edge of the internal strobe signal CST and outputs read data RD1, and the flip-flop circuit 34 is at the timing of the falling edge of the internal strobe signal CST. The received data string RDQ is latched and read data RD2 is output.

  Therefore, this memory interface circuit 12 can suppress a decrease in timing margin due to factors such as signal reflection and crosstalk noise in the transmission path between the memory 20 and the memory interface circuit 12.

  (2) The receiving circuit of the memory interface circuit 12 generates the internal mask signal CKM corresponding to the period for receiving the data string DQ based on the strobe signal DQS synchronized with the data string DQ. The adjustment circuit 50 adjusts the phase of the internal strobe signal CST according to the phase relationship between the internal mask signal CKM and the internal strobe signal CST. Therefore, the timing margin for taking in the data string DQ by adjusting the phase of the internal strobe signal CST for latching the data string DQ according to the timing of the internal mask signal CKM, that is, the timing of the data string DQ due to the VT fluctuation. It is possible to suppress the decrease in data and to capture correct data.

  (3) The memory interface circuit 12 includes a counter 36 that counts the clock signal CKS in response to the internal strobe signal IRS. The counter 36 outputs an end detection signal BLE when the count value becomes equal to the burst length BLN. . The calculation unit 54 of the adjustment circuit 50 outputs the update code UC in accordance with the end detection signal BLE. In response to the end detection signal BLE, the timing control unit 56 updates the set value stored in the register 41a of the DLL circuit 41 that generates the internal clock signal CKS according to the update code UC. Therefore, the memory interface circuit adjusts the timing of the internal clock signal CKS, that is, the internal strobe signal CST every time the data string DQ is received from the memory 20. Therefore, the timing of the internal strobe signal CST can be adjusted following the VT fluctuation. Further, since the timing is adjusted every time data is read from the memory 20, it is not necessary to adjust the timing by operating the training circuit 13 during the operation. Therefore, there is no need to suppress (stop) the read operation of the core circuit 11 for timing adjustment, and the overhead for reading data from the memory 20 can be reduced.

In addition, you may implement each said embodiment in the following aspects.
In the system shown in FIG. 1, at least one of the system circuit 10 and the memory 20 may be mounted directly on the substrate.

  Although embodied in the memory interface circuit 12 that captures the data string DQ at the timing of the rising edge and falling edge of the internal strobe signal CST generated based on the internal clock signal CK2, the timing of the rising edge of the internal strobe signal CST, or A memory interface circuit that captures the data string DQ at the timing of the falling edge may be embodied.

The present invention is not limited to the memory interface circuit 12 and may be embodied in a device that performs communication using the data string DQ and the strobe signal DQS.
In the above embodiment, whether the phase difference between the internal mask signal CKM and the internal strobe signal CST is a phase difference suitable for taking in the data string DQ (for example, a quarter period (90 degrees) of the internal clock signal CK2). 2 can be detected, the delay amount of the DLL circuits 51 and 52 shown in FIG. 2 is not limited to the delay amount described in the above embodiment. For example, a signal obtained by delaying the internal mask signal CKM by 90 degrees and the internal strobe signal CST are supplied to the phase detection unit 53. For example, the delay amount of the DLL circuit 51 may be 45 degrees, and the delay amount of the DLL circuit 52 may be 135 degrees (= 90 degrees + 45 degrees). Alternatively, the delay amount of the DLL circuit 51 may be 90 degrees, and the delay amount of the DLL circuit 52 may be 0 degrees.

33, 34 Latch circuit 36 Counter 37, 39 Mask generator 40, 41 Delay locked loop circuit (DLL circuit)
50 adjustment circuit 51, 52 delay locked loop circuit (DLL circuit)
53 Phase Detection Unit 54 Operation Unit 55, 56 Timing Control Unit CK2 Clock Signal CST Internal Strobe Signal CKM Mask Signal DQ Data Sequence D0-D7 Data DQS Strobe Signal

Claims (8)

A first signal generation circuit for generating an internal strobe signal corresponding to the clock signal;
A latch circuit for latching data in response to the internal strobe signal;
A second signal generation circuit for generating a mask signal corresponding to a period for receiving the data in response to an external strobe signal;
An adjustment circuit that adjusts the phase of the internal strobe signal in accordance with the phase relationship between the mask signal and the internal strobe signal;
A receiving circuit. The first signal generation circuit includes:
A delay circuit that generates a delayed clock signal obtained by delaying the clock signal according to a set value;
A synthesis circuit for synthesizing the delayed clock signal and the mask signal to generate the internal strobe signal;
The receiving circuit according to claim 1, comprising: The adjustment circuit includes:
A first delay circuit for delaying the mask signal;
A second delay circuit for delaying the internal strobe signal;
A phase detector for detecting the phase of the output signal of the first delay circuit and the output signal of the second delay circuit;
An arithmetic unit that generates an update code so as to make the phase of the output signal of the second delay circuit equal to the phase of the output signal of the first delay circuit according to the detection result of the phase detection unit;
A control unit for updating a set value of the first delay circuit according to the update code;
The receiving circuit according to claim 1, comprising: A detection circuit for detecting a preamble of the external strobe signal and outputting a first signal;
A second delay circuit that delays the first signal according to a set value to generate a second signal;
Including
The receiving circuit according to claim 3, wherein the second signal generation circuit generates the mask signal based on the second signal.   The receiving circuit according to claim 4, wherein the adjustment circuit includes a second control unit that updates a set value of the second delay circuit according to the update code. The second signal generation circuit includes:
A synthesis circuit for synthesizing the second signal and the external strobe signal to generate a strobe signal;
A counter that counts the delayed clock signal in response to the strobe signal and outputs an end signal when the count value reaches a set value according to a data string;
A mask generator that generates the mask signal based on the strobe signal and the end signal;
The receiving circuit according to claim 4, comprising: A core circuit and an interface circuit;
The core circuit reads data from the target circuit via the interface circuit,
The interface circuit includes a reception circuit that receives data and a strobe signal output from the target circuit,
The receiving circuit is
A first signal generation circuit for generating an internal strobe signal corresponding to the clock signal;
A latch circuit for latching data in response to the internal strobe signal;
A second signal generation circuit for generating a mask signal corresponding to a period for receiving the data in response to an external strobe signal;
An adjustment circuit that adjusts the phase of the internal strobe signal in accordance with the phase relationship between the mask signal and the internal strobe signal;
Including a system unit. Generate an internal strobe signal corresponding to the clock signal,
A latch circuit for latching data in response to the internal strobe signal;
Generating a mask signal according to a period of receiving the data according to an external strobe signal;
Adjusting the delay time for delaying the clock signal according to the phase of the mask signal and the internal strobe signal;
Adjusting the phase of the internal strobe signal to generate the delayed clock signal by generating the internal strobe signal based on a signal obtained by delaying the clock signal;
A timing adjustment method characterized by the above.

Images