The present technique relates to the field of integrated circuits. More particularly, the present technique relates to a level conversion circuit.
An integrated circuit may have different portions operating in different voltage domains. A level conversion circuit may be used to convert a signal from the voltage levels used in one domain to the voltage levels used in another domain. The present technique seeks to provide an improved level conversion circuit.
Viewed from one aspect, the present technique provides a level conversion circuit for generating an output signal in response to an input signal, the input signal having one of a high input level and a low input level, and the output signal having one of a high output level and a low output level; the level conversion circuit comprising:
(a) a shifting circuit configured to generate a shifted signal in response to the input signal, the shifted signal having one of the high output level and an intermediate level between the high output level and the high input level;
the shifting circuit comprising a first pullup transistor configured to enter a conductive state to pull the shifted signal to the high output level when the input signal has a first input level comprising one of the low input level and the high input level, and to enter a less conductive state to allow the shifted signal to drop to the intermediate level when the input signal has a second input level comprising the other of the low input level and the high input level;
(b) a second pullup transistor configured to enter a conductive state to pull an intermediate output node to the high output level when the shifted signal has the intermediate level, and to enter a less conductive state when the shifted signal has the high output level;
(c) a pulldown transistor configured to enter a conductive state to pull the intermediate output node to the low output level when the input signal has said first input level, and to enter a less conductive state when the input signal has said second input level;
(d) an output buffer configured to generate the output signal based on the signal at the intermediate output node; and
(e) a keeper circuit configured to retain the intermediate output node at the high output level when the input signal has said second input level.
The level conversion circuit converts an input signal having one of a high input level and a low input level into an output signal having one of a high output level and a low output level. The level conversion circuit comprises a shifting circuit which generates a shifted signal having one of the high output level and an intermediate level, with the intermediate level being between the high output level and high input level. The shifting circuit includes a first pullup transistor which pulls the shifted signal to the high output level when the input signal has a first input level (one of the low and high input level) and enters a less conductive state to allow the shifted signal to drop to the intermediate level when the input signal has a second input level (the other of the low and high input level). A second pullup transistor and pulldown transistor are provided to pull an intermediate output node to the high output level and low output level respectively. The second pullup transistor is controlled using the shifted signal generated by the shifting circuit, while the pulldown transistor is controlled using the input signal or a signal derived from the input signal. This circuit arrangement allows the intermediate output node to swing across the full range between the high output level and low output level in response to an input signal swinging between the low input level and high input level, and provides relatively high performance across process corners and local variation compared to other level converting designs. An output buffer then generates the final output signal based on the signal at the intermediate output node.
However, a problem arises when the input signal has the second input level. At this time, the first pullup transistor is in the less conductive state and so the shifted signal drops to the intermediate level to turn on the second pullup transistor. However, if the input signal remains at the same level for some time, then eventually leakage through the first pullup transistor will cause the shifted signal to rise back up to the high output level, making the second pullup transistor less conductive. When the input signal is at the second input level, the pulldown transistor is also in the less conductive state, and so this may cause the intermediate output node to float because neither of second pullup transistor and pulldown transistor would be on. This could cause reduced performance and incorrect generation of the output signal, which could lead to errors in downstream circuits which use the output signal.
To address this problem, a keeper circuit is provided to retain the intermediate output node at the higher output level when the input signal has the second input level, so that the intermediate output node cannot float. If a keeper circuit was provided to hold the shifted signal node at the required level, rather than the intermediate output node, then the efficiency of the shifted circuit would be reduced and the second pullup transistor and pulldown transistor would switch more slowly. The keeper circuit provided at the intermediate output node provides an improved level conversion circuit because it enables faster switching of the level conversion circuit while avoiding the problem of the floating intermediate output node discussed above.
The shifting circuit may have a capacitor coupled between a shifted signal node for carrying the shifted signal and an input node for carrying the input signal or a complementary input signal (an inverted version of the input signal). The capacitive coupling provided by the capacitor enables the shifted signal to be generated with the intermediate level between the high input level and high output level. When the input signal or complementary input signal is high, then a first plate of the capacitor is charged to the high input level, and a second plate of the capacitor is charged to the high output level using the first pullup transistor. When the input signal or complementary input signal switches low, then the first plate of the capacitor is discharged to the low input level and this causes the second plate to drop to the intermediate level between the high input level and high output level. Typically, the intermediate level may correspond to the difference between the high output level and high input level. Generating the shifted signal using a capacitor in this way enables a simpler circuit arrangement for the level conversion circuit with reduced static power consumption compared to other level conversion designs.
The capacitor may be implemented in various forms. However, it can be useful to provide a capacitor in the form of a metal on silicon field effect transistor (MOSFET), because this tends to have reduced circuit area compared to other forms of capacitor.
In general, the shifting circuit may comprise first and second complementary circuits which generate complementary shifted signals. Each of the complementary circuits may have a respective capacitor and first pullup transistor, and the first and second complementary circuits may be responsive to the input signal and a complementary input signal respectively. The second pullup transistor may receive the shifted signal generated by either of the first and second complementary circuits. The first and second complementary circuits may be cross coupled so that the first pullup transistor of one complementary circuit has its gate terminal coupled to the shifted signal node of the other complementary circuit, and vice versa. This cross coupled design allows each capacitor to be charged and discharged in the way discussed above.
The keeper circuit may have various forms. For example, the keeper circuit may have a keeper control transistor which enters a conductive state when the input signal has the second input level and enters a less conductive state when the input signal has the first input level. However, as the input signal switches between the low input level and high input level, while the keeper circuit acts in the output domain, it is preferable to control the keeper circuit using a signal switching between the low output level and high output level. For example, a complementary version of the signal at the intermediate output node can be used to control the keeper control transistor to switch between states. While this complementary signal could be generated in many ways, it can be convenient to use the output buffer to provide the complementary version of the signal at the intermediate output node. The output buffer is typically already provided for generating a stronger output signal, and so this signal can be reused to control the keeper control transistor, so that it is not necessary to provide further circuitry for generating the control signal for the keeper control transistor.
The keeper circuit helps maintain correct operation when the input signal has the second input level. However, when the input signal switches to the first input level then the pulldown transistor will turn on and may have to fight against the keeper circuit for some time before the keeper circuit is turned off. To ensure that the pulldown transistor can overcome the pull of the keeper circuit, it can be useful to provide a pulldown transistor which passes a larger current when in the conductive state than the current passed by the keeper circuit when it retains the intermediate output node at the high output level. For example, the relative widths of the transistors in the keeper control circuit and the pulldown transistor can be selected to ensure a greater drive strength for the pulldown transistor than the keeper circuit.
It would be possible to provide the keeper circuit with a fixed drive strength, for example by making the keeper control transistor narrower than the pulldown transistor. However, if the relative drive strength is fixed, then process variation or other conditions may affect the relative drive strengths, and could prevent the circuit working as desired. Therefore, a more flexible approach allows for varying drive strength of the keeper circuit. The keeper circuit may be provided with a current regulating transistor in series with the keeper control transistor. The current regulating transistor may limit the current passing through the keeper control transistor when in the conductive state. The gate voltage applied to the current regulating transistor controls the current passing through the current regulating transistor and hence changes the drive strength of the keeper circuit as a whole. A tuning circuit may be provided to tune the gate voltage of the current regulating transistor so as to adjust the drive strength of the keeper circuit. If the pullup of the keeper circuit is too weak, then the functionality of the level conversion circuit is impaired since the problem discussed above with the floating intermediate output node may still arise, and so in this case the tuning circuit may adjust the gate voltage of the keeper control transistor to allow more current to pass through the keeper circuit. On the other hand, if the pullup of the keeper circuit is too strong, then the performance of the level conversion circuit will be reduced since the pulldown transistor will take longer to overcome the keeper circuit, preventing the level conversion circuit operating at high frequencies. In this case the tuning circuit can adjust the gate voltage of the current regulating transistor to allow less current to pass through the keeper circuit. Hence, the tuning circuit and current regulating transistor provide a flexible way of controlling the level conversion circuit to ensure that it operates correctly and at a sufficiently high frequency.
The arrangement discussed above can be implemented using thin oxide transistors, which is suitable for use with relatively low voltages. To allow the level conversion circuit to be used for a wide range of voltage scaling, from a near threshold supply to full I/O voltage levels, the level conversion circuit may be implemented using thicker oxide transistors. However, in this case the basic design of level conversion circuit may encounter reduced performance because in sub-threshold operation, the switching speed of transistors is exponentially dependent on the overdrive voltage (difference between the high voltage level and the threshold voltage of the transistor), and as the threshold voltage of the thick oxide devices is much larger then the overdrive voltage is reduced.
To address this problem, the second pullup transistor and pulldown transistor may be retained as thin oxide transistors, but then a pair of cascode transistors coupled in series between the second pullup and the pulldown transistor with a thicker gate oxide can be provided to protect the second pullup transistor and pulldown transistor against drain-source voltage breakdown from the higher I/O voltage levels. This enables improved performance because the second pullup and pulldown transistors will switch more quickly because they have a lower threshold voltage than the thick oxide version, but the thicker oxide cascode transistors will shield the drains of the second pullup and pulldown transistors to prevent breakdown.
The cascode transistors may be held at a gate voltage level between the high output level and low output level. This allows the cascode transistors to remain on at all times to preserve speed, but in a state which reduces the voltage applied to the drains of the second pullup and pulldown transistors.
The other elements of the level conversion circuit, such as the shifting circuit, the output buffer and the keeper circuit can also be constructed using the thicker oxide transistors. Typically, the output buffer may comprise one or more buffer stages having a pullup portion which pulls an output of the buffer stage to the high output level and a pulldown portion which pulls an output of the buffer stage to the low output level. The pullup and pulldown portions respond to opposite transitions of the intermediate output node respectively. The output signal may then be derived from the output of a final buffer stage of the one or more buffer stages. The output buffer is useful for providing an output signal which has a more reliable voltage level and stronger drive current than the signal at the intermediate output node.
In the thick oxide embodiment, it can be useful to provide the output buffer such that the pullup portion of the first buffer stage of the one or more buffer stages is coupled to a node between the second pullup transistor and the at least one pair of cascode transistors, and the pulldown portion of the first buffer stage is coupled to a node between the pulldown transistor and the at least one pair of cascode transistors. While the circuit would still work if the buffer stage was coupled directly to the intermediate output node which is internal to the cascode transistors, the intermediate output node tends to transition more slowly than the external node between the pullup transistor and cascode transistors and the external node between the pulldown transistor and cascode transistors. Therefore, performance can be improved by coupling the pullup/pulldown portions of the output buffer to the respective external nodes instead of the intermediate output node. Nevertheless, the keeper circuit may still be coupled to the intermediate output node and the keeper circuit's retention of the intermediate output node at the output high voltage level will also propagate through to the external nodes to ensure the correct output at the output buffer.
Also, in the thick oxide embodiment, the pullup portion of each buffer stage may comprise several stacked transistors in series between the output of the buffer stage and a supply node for supplying the high output level. Similarly, the pulldown portion of each buffer stage may have multiple stacked transistors in series between the output of the buffer stage and a supply node for supplying the low output level. Using series stacked devices in this way reduces the short circuit current through the buffer stage. Since the transition speed of the output signal is relatively slow in the thicker oxide embodiment, then there is a longer period when the pullup portion of the buffer stage is turning on while the pulldown portion is turning off, or vice versa. Therefore, there is an increased time when both the pullup portion and the pulldown portion may be partially on, allowing a short circuit current to flow between the high output supply node and the low output supply node. By using several series stacked devices, this short circuit current can be reduced, to reduce power consumption.
Viewed from another aspect, the present technique provides a level conversion circuit for generating an output signal in response to an input signal, the input signal having one of a high input level and a low input level, and the output signal having one of a high output level and a low output level; the level conversion circuit comprising:
(a) shifting circuit means for generating a shifted signal in response to the input signal, the shifted signal having one of the high output level and an intermediate level between the high output level and the high input level;
the shifting circuit means comprising first pullup transistor means for entering a conductive state to pull the shifted signal to the high output level when the input signal has a first input level comprising one of the low input level and the high input level, and entering a less conductive state to allow the shifted signal to drop to the intermediate level when the input signal has a second input level comprising the other of the low input level and the high input level;
(b) second pullup transistor means for entering a conductive state to pull an intermediate output node to the high output level when the shifted signal has the intermediate level, and entering a less conductive state when the shifted signal has the high output level;
(c) pulldown transistor means for entering a conductive state to pull the intermediate output node to the low output level when the input signal has said first input level, and to enter a less conductive state when the input signal has said second input level;
(d) output buffer means for generating the output signal based on the signal at the intermediate output node; and
(e) keeper circuit means for retaining the intermediate output node at the high output level when the input signal has said second input level.
Viewed from a further aspect, the present technique provides a method of generating an output signal in response to an input signal, the input signal having one of a high input level and a low input level, and the output signal having one of a high output level and a low output level; the method comprising:
(a) in response to the input signal, generating a shifted signal having one of the high output level and an intermediate level between the high output level and the high input level,
the shifted signal being generated using a shifting circuit comprising a first pullup transistor configured to enter a conductive state to pull the shifted signal to the high output level when the input signal has a first input level comprising one of the low input level and the high input level, and to enter a less conductive state to allow the shifted signal to drop to the intermediate level when the input signal has a second input level comprising the other of the low input level and the high input level;
(b) switching a second pullup transistor to a conductive state to pull an intermediate output node to the high output level when the shifted signal has the intermediate level, and to a less conductive state when the shifted signal has the high output level;
(c) switching a pulldown transistor to a conductive state to pull the intermediate output node to the low output level when the input signal has said first input level, and to a less conductive state when the input signal has said second input level;
(d) retaining the intermediate output node at the high output level using a keeper circuit when the input signal has said second input level; and
(e) generating an output signal based on the signal at the intermediate output node.
Further aspects, features and advantages of the present technique will be apparent from the following detailed description of examples, which is to be read in conjunction with the accompanying drawings.
The two complementary half circuits 4-L, 4-R of the shifting circuit 4 are cross coupled so that the first pullup transistor 8-L, 8-R of one half circuit 4-L, 4-R is coupled to the shifted signal node x, xb of the other half-circuit 4-R, 4-L, and vice versa. The shifted signal xb generated by the left-hand complementary circuit 4-L is used to control a subsequent inverter stage 9 of the level conversion circuit 2. In other embodiments, the inverter stage 9 could receive the shifted signal x from the right hand complementary circuit 4-R instead of the shifted signal xb from the left hand complementary circuit 4-L.
The inverter stage 9 comprises a second pullup transistor 10 and a pulldown transistor 14. The second pullup transistor 10 pulls an intermediate output node 12 (out) to the VDDH level in response to the shifted signal xb being at the intermediate level VDDH-VDDL. When the shifted signal xb is at the VDDH level then the second pullup transistor 10 is turned off (less conductive). The pulldown transistor 14 is controlled by the complementary input signal inb to pull the intermediate output node 12 to ground when inb is at VDDL, and when inb is at 0 then the pulldown transistor 14 is turned off (less conductive).
A keeper circuit 20 is provided on the intermediate output node 12 to retain the intermediate output node 12 at the VDDH level. The keeper circuit 20 has a keeper control transistor 22 which turns on and off in response to a complementary output signal outb, and a current regulating transistor 24 which regulates the current passing through the keeper circuit. A tuning circuit 30 is provided to control the gate voltage Vkeep of the current regulating transistor 24. In this example, the tuning circuit 30 has a number of diodes 32 arranged in series, and different levels of the keeper voltage Vkeep can be tapped from different points of the chain of diodes 32. A selection circuit not illustrated in
The level conversion circuit 2 also includes an output buffer 40 which comprises at least one buffer stage (only one buffer stage in the example of
As shown in
If the input signal in now remains at the high input level for some time, then ideally the state of the level conversion circuit should remain as shown in
If the input remains at the low input zero for a time then the circuit will gradually return to the state shown in
Hence, in summary the keeper circuit 20 enables correct operation even if the level conversion circuit 2 remains in the same state for long enough to be affected by the leakage current 50 through transistor 8-L.
To shield the thin oxide second pullup transistor 10 (M4) and the pulldown transistor 14 (M1) from drain-source breakdown caused by the higher voltages at their drains, a pair of cascode transistors 210 are provided in series between the second pullup transistor 10 and pulldown transistor 14. To leave the cascode transistors 210 always on, the cascode transistors 210 are held at a gate voltage of (VDDH/2) which is generated using a chain of diodes Dl to D4. This can be the same chain of diodes that is used in the tuning circuit 30 to generate the keeper voltage for the current regulating transistor 24. In other examples, the gate voltage of the cascode transistors need not be exactly VDDH/2. While
In the example of
The operation of the level conversion circuit 200 is otherwise the same as shown in
A test chip has been taped out and simulation results show performance improvements over prior art. The below results are for converting from VDDL=0.4V to VDDH=1.0V in 45 nm SOI. When the embodiment of