Circuit and method for providing a corrected duty cycle

[0001] 1. Field of the Invention

[0002] The present invention relates to a circuit and method for providing a corrected duty cycle.

[0003] 2. Description of the Related Art

[0004] Generally, a delay locked loop (DLL) is used in a synchronous semiconductor memory device, such as a double data rate synchronous dynamic random access memory (DDR SDRAM), to perform synchronization between an internal clock signal and an external clock signal of the synchronous semiconductor memory device. When the external clock signal is input to the synchronous semiconductor memory device, a time delay occurs due to a clock skew between the external clock signal and the internal clock signal. Therefore, the DLL is employed in the synchronous semiconductor memory device for generating the internal clock signal by compensating the clock skew.

[0005] The DDR SDRAM inputs and/or outputs data at rising and falling edges of the clock signal to increase the transmission speed of data. As the operational speed of the DDR SDRAM is increased, performance of the DDR SDRAM is greatly affected by performance of the DLL. In addition, variations in process, supply voltage and temperature can affect the driving ability of the circuit, thereby creating timing and/or noise problems and worsening the design margin of the duty cycle. Since reliable data transmission is achieved in DDR SDRAM when the duty cycle is equivalent to 50%, a duty cycle correction method applied to the DLL is required for ensuring sufficient design margin of the duty cycle.

[0006] FIG. 1 shows a digital duty cycle correction circuit 100 disclosed in U.S. Pat. No. 7,015,739. The digital duty cycle correction circuit 100 includes a duty cycle detector circuit 104 configured to receive a first internal clock signal ICLK and a second internal clock signal ICLK_B. A comparator circuit 106 is configured to receive output signals of the duty cycle detector circuit 104 and provide a comparison result. A counter circuit 108 is configured to perform an addition and/or a subtraction operation according to the comparison result so as to provide digital codes, and a digital to analog converter (DAC) 110 is configured to generate control signals according to the digital codes. Based on the above, the digital duty cycle correction circuit 100 requires the counter circuit 108 to perform a duty cycle correction and requires the DAC 110 to convert the signals. Therefore, the circuit 100 is complex and requires a large silicon area.

[0007] Alternatively, an analog structure to simplify the duty cycle correction circuit is desirable. Such circuit needs a fast response for correcting the internal clock signal larger or smaller than a desired duty cycle to avoid external noise disturbance.

[0008] An aspect of the present invention is to provide a duty cycle correction circuit.

[0009] According to one embodiment of the present invention, the duty cycle correction circuit comprises a duty cycle detector, a filter, an amplifier, a charge pump, a control circuit, and a duty cycle corrector. The duty cycle detector is configured to generate a first pair of control signals according to a pair of internal clock signals. The filter is configured to obtain average voltages of the first pair of control signals. The amplifier is configured to compare output voltages of the filter for generating an enable signal. The control circuit is configured to generate a selection signal according to the enable signal. The charge pump is configured to generate a second pair of control signals according to the enable signal and the selection signal, and the duty cycle corrector is configured to receive a pair of external clock signals, the first pair of control signals, and the second pair of control signals for generating the pair of the internal clock signals with a corrected duty cycle.

[0010] Another aspect of the present invention is to provide a method for providing a corrected duty cycle.

[0011] According to one embodiment of the present invention, the method for providing a corrected duty cycle comprises the steps of generating a first pair of control signals according to a first pair of internal clock signals; comparing averaged voltages of the first pair of the control signals for generating an enable signal; generating a second pair of control signals according to the enable signal; and generating a second pair of internal clock signals with the corrected duty cycle according to the first pair of control signals, the second pair of control signals, and a pair of external clock signals. The second pair of control signals is the average charge voltages controlled by a plurality of current sources.

[0023] FIG. 2 shows a block diagram of a duty cycle correction circuit 20 according to one embodiment of the present invention. The duty cycle correction circuit 20 comprises a duty cycle corrector 21, a control circuit 22, a duty cycle detector 23, a filter 25, an operational amplifier (OP) 27, and a charge pump 29. Referring to FIG. 2, the duty cycle corrector 21 is configured to generate an internal clock signal CK and a complementary internal clock signal CK_B according to an external clock signal ECK and a complementary external clock signal ECK_B, respectively. The duty cycle detector 23 is configured to generate a fine-tuning control signal DCF and a complementary fine-tuning control signal DCF_B according to the internal clock signals CK and CK_B. The filter 25 is configured to filter the control signals DCF and DCF_B for obtaining average voltages. After receiving the average voltages from the filter 25, the operational amplifier 27 generates an enable signal EN, which controls the charge pump 29 to charge or discharge a capacitor. The charge pump 29 outputs a coarse tuning control signal DCC and a complementary coarse tuning control signal DCC_B to the duty cycle corrector 21 for adjusting the initial duty cycle of the internal clock signals CK and CK_B to a corrected duty cycle. In addition, the control circuit 22 outputs control signals ENCPS and ENCPL to the charge pump 29 for adjusting bias current of the charge pump 29 according to the enable signal EN. The charge pump 29 may perform a coarse tuning operation with larger bias current. Alternatively, the charge pump 29 may perform a power saving operation with small bias current to consume less power.

[0024] FIG. 3 shows a detailed block diagram in accordance with one embodiment of the duty cycle corrector 21 in FIG. 2, wherein the duty cycle corrector 21 is represented as a differential amplifier with multi-input pairs in this embodiment. Referring to FIG. 3, the duty cycle corrector 21 comprises a first input pair including NMOS transistors N₁ and N₂ for receiving the internal clock signals CK and CK_B, a second input pair including NMOS transistors N₃ and N₄ for receiving the fine tuning control signals DCF and DCF_B, and a third input pair including NMOS transistors N₅ and N₆ for receiving the coarse tuning control signals DCC and DCC_B. The first, second, and third input pairs are all connected to diode-connected loads composed of PMOS transistors P₁ and P₂. The output signals OUT and OUT_B of the multi-input differential amplifier are output to an amplifier 32 for generating the internal clock signals CK and CK_B with the corrected duty cycle.

[0025] FIG. 4A shows a detailed block diagram in accordance with one embodiment of the duty cycle detector 23 in FIG. 2, wherein the duty cycle detector 23 is represented as a differential amplifier in this embodiment. Referring to FIG. 4A, the duty cycle detector 23 comprises an input pair including NMOS transistors N₁ and N₂ for receiving the internal clock signals CK and CK_B, and the input pair is connected to diode-connected loads composed of PMOS transistors P₁ and P₂, and the DCC function enable signal DCCEN.

[0026] FIG. 4B is an operational waveform of the duty cycle detector 23, which shows the relationship between the internal clock signals CK and CK_B and the fine-tuning control signals DCF and DCF_B. As shown in FIG. 4B, the control signals DCF and DCF_B are represented by a sawtooth waveform according to the clock signals CK and CK_B applied to the input pair in FIG. 4A. When the duty cycle of the clock signal CK is smaller or larger than the desired duty cycle, for example, 45%, the control signal DCF increases as the complementary control signal DCFB decrease, thereby increasing the difference between the two signals. In this case, the duty cycle detector 23 operates in a linear region, and thus it responds to the clock signals CK and CK_B instantly and accurately. However, when the duty cycle of the clock signal CK is much smaller or larger than the desired duty cycle, e.g., 20% duty cycle as shown in FIG. 4B, the control signal DCF increases rapidly as the complementary control signal DCF_B decreases such that the duty cycle detector 23 enters its saturation region quickly. In this case, it will have difficulty to respond a real change of the clock signals CK and CK_B.

[0027] FIG. 5 shows a detailed block diagram in accordance with one embodiment of the filter 25 and the amplifier 27 in FIG. 2, wherein the filter 25 is represented as an RC low-pass filter in this embodiment.

[0028] FIG. 6 shows a detailed block diagram in accordance with one embodiment of the control circuit 22 in FIG. 2. The control circuit 22 is configured to generate the control signals ENCPL and ENCPS according to the enable signal EN. A selection signal SCP is generated to provide the control signals ENCPS and ENCPL after the transmission of two D flip-flops enabled by the enable signal EN. The control signals ENCPL and ENCPS are sent to the charge pump 29 to control the operation of the charge pump 29. When the selection signal SCP is at a logic low level, the control signal ENCPL is activated, which enables the charge pump 29 to perform a coarse tuning operation with larger bias current. Alternatively, when the selection signal SCP is at a logic high level, the control signal ENCPS is activated, which enables the charge pump 29 to perform a power saving operation with less bias current.

[0029] FIG. 7 shows a detailed block diagram in accordance with one embodiment of the charge pump 29 in FIG. 2. The charge pump 29 is configured to generate average charge voltages DCC and DCC_B stored in a capacitor with different bias current, e.g., 10I or I in this embodiment. These bias current sources are controlled by the enable signal EN, control signals ENCPL and ENCPS, and a charge pump function enable signal DCP. The charge pump 29 performs a coarse tuning operation when the duty cycle detector 23 enters a saturation region, and performs a power saving operation when the duty cycle detector 23 enters a linear region.

[0030] In order to explain the operation method of the duty cycle correction circuit 20 of the present invention more clearly, FIG. 8 shows the flow chart of the method for providing a corrected duty cycle in accordance with an exemplary embodiment. In step S80, fine tuning control signals DCF and DCF_B are generated according to internal clock signals CK and CK_B. In step S82, an average voltage of the fine tuning control signal DCF is compared to an average voltage of the fine tuning control signal DCF_B to generate an enable signal EN. In step S84, coarse tuning control signals DCC and DCC_B are generated according to the enable signal EN, wherein signals DCC and DCC_B are average charge voltages controlled by a plurality of current sources. In step S86, the internal clock signals CK and CK_B with the corrected duty cycle are generated according to the fine tuning control signals DCF and DCF_B, the coarse tuning control signals DCC and DCC_B, and the external clock signals ECK and ECK_B. The details of the method for providing the corrected duty cycle of the present invention are described in accordance with FIGS. 2 to 9.

[0031] Referring to FIG. 9, an operation waveform showing a relationship between the fine tuning control signals DCF and DCF_B, the coarse tuning control signals DCC and DCC_B, and the enable signal EN, the DCC function enable signal DCCEN, the charge pump function enable signal DCP, and the control signals ENCPS, ENCPL are discussed. When the initial duty cycle of the internal clock signal CK is much smaller than the desired duty cycle, the control signal DCF increases rapidly as the complementary control signal DCF_B decreases rapidly after the DCC function enable signal DCCEN is activated. In this case, the duty cycle detector 23 enters a saturation region in a very short time. Under the condition of entering the saturation region, the difference between the averaged control signals DCF and DCF_B obtained by the filter 25 is high enough to trigger the operational amplifier 27 for generating the enable signal EN at a logic high level. After the delay time T₁ determined by a delay unit 72 as shown in FIG. 7, the charge pump 29 generates a gradually increased voltage DCC and a gradually decreased voltage DCC_B.

[0032] Subsequently, the increased voltage DCC and decreased voltage DCC_B output to the duty cycle corrector 21 for increasing and decreasing the voltages of the nodes OUT_B and OUT, respectively. Meanwhile, due to the increased voltage DCC and the decreased voltage DCC_B, the difference between the control signals DCF and DCF_B is reduced, the duty cycle detector 23 soon enters a linear region, which indicates that the duty cycle detector 23 has the ability to rapidly respond to the variation of the duty cycle of clock signals CK and CK_B. As a result, the duty cycle of the clock signals CK and CK_B can be immediately corrected to a desired duty cycle of about 50%. In one exemplary embodiment, the duty cycle detector 23 operates in the linear region when a duty cycle of the pair of internal clock signals is within a range of a predetermined value of the corrected duty cycle. The predetermined value is about ±10% and the corrected duty cycle is about 50%.

[0033] To reduce power consumption, when the duty cycle detector 23 enters the linear region, or the initial duty cycle of the internal clock signals CK and CK_B are close to the desired duty cycle, the selection signal SCP is activated, which enables the charge pump 29 to perform a power saving operation with less bias current. In this case, the overall circuit can consume less power and still have the ability to respond to the small variation of the duty cycle of clock signals CK and CK_B.

[0034] The above-described embodiments of the present invention are intended to be illustrative only. Numerous alternative embodiments may be devised by those skilled in the art without departing from the scope of the following claims.