The present invention relates to a driving circuit architecture for an output buffer stage with high speed and with reduced noise induced on the power supply. More particularly, the invention relates to a driving circuit as above and comprising first and second circuit portions, each one provided for the generation of a respective driving signal for a corresponding transistor of the buffer stage.
As it is well known in the specific field of the manufacturing of semiconductor integrated memory devices, at present the market for nonvolatile memories typically requires higher and higher access speeds against lower and lower supply voltages. This need results from the desire of reducing the power consumption as much as possible and increasing, in the meantime, the so-called throughput.
Within this field the output buffer stages play a fundamental role. Their switching speed is strictly connected to the memory access time and the several switchings they are subjected to strongly affect the power consumption. For example, to optimize the power consumption in an output stage realized according to CMOS technology, the periods of time when the PMOS and NMOS transistors are simultaneously in the saturation zone may have to be reduced to the minimum. In fact, during the transitions between one logic level and another, there is typically a period when both the MOS transistors are in the saturation zone and generate a current between the power supply and ground, called the OVERLAP current.
By way of example, we now consider the case of a CMOS output stage which should drive a capacitive load Cload, as shown in
However, to ensure that these conditions occur, the two control signals applied to the respective gate terminals of the NMOS and PMOS transistors should be split, as shown in
From what has been said, it can be inferred that the generation of the two control signals is anything but simple, and it directly affects the buffer performance. It should be also highlighted that the rise speed of the two control signals and their pattern are directly connected to the disturbance induced on the power supply by the buffer switchings. To avoid too high noise peaks, the turn-on of the PMOS transistor and of the NMOS transistor is controlled so that the charge/discharge current of the capacitor Cload has a constant slope (dI/dt=const).
The noise induced on the power supply Vdd is substantially caused by the voltage variation across the parasitic inductor/resistor group, inevitably in the supply line. To better understand the phenomena previously described, reference can be made to
The limitation of sudden current variations on the capacitor results in the lack of spikes on the supply voltage. The variations of inductive nature on the power supply are proportional to the variation of the current absorbed by the supply source LdI/dt, whereas the variations of a resistive nature are proportional to the current required by the source.
What has been so far described allows generating a control signal with a constant slope. In the period of time elapsing between the start of the control signal and the overtaking of the threshold voltage of the MOS transistors of the output driver, it is possible to speed up the times, since the load driving step is still to come. A control is then attained, so called with two slopes, the first steeper between the instant zero and the overtaking of the threshold voltage of the MOS transistor (i.e. the PMOS transistor during the charging and the NMOS transistor during the discharging) and a second slope, less steep, until the output switchings are reached.
This approach however, has a drawback due to the fact that the resistors used for obtaining the two different slopes make sure that the voltage values on the gate terminals of the transistors of the buffer stage do not go to zero (in the case of the NMOS) or to Vdd (in the case of the PMOS) also when the variation has occurred. To address this problem the known technique proposes the scheme of
With respect to the previous approach the presence can be appreciated of a MOS transistor which completely short-circuits the resistor once the transition has occurred. It is in fact to be remembered that after an initial period when the current across the capacitor would tend to vary in a sudden way, causing noise on the power supply, the current tends to zero in an exponential way, therefore the control is not necessary any more. The scheme of
Although advantageous under several aspects, this approach also exhibits a limitation. In the applications to multi-level memories there is the need of obtaining low output impedances from the buffers and the use of the latter approach described has revealed itself to be a problem since low output impedances result in the need of using resistors of too low a value that cannot be practically realized or which are too subject to variations due to the manufacturing process. Therefore the approach of generating control signals with the linear profile at intervals obtained by resistors has revealed itself to be difficult to be applied.
The problem addressed by the present invention is that of providing a driving circuit for an output buffer stage incorporated in a memory electronic device, having such structural and functional characteristics as to allow the generation of driving signals with a linear pattern at intervals, the signal being however generated by components of the MOS or CMOS type, thus overcoming the limits of the approaches proposed by the prior art.
In view of the foregoing background, it is therefore an object of the present invention to provide a driving circuit and associated methods for the efficient driving of the output buffer stage with reduced noise on the power supply.
This and other objects, features and advantages in accordance with the present invention are provided by a driving circuit comprising a first circuit portion for generating a driving signal for the NMOS transistor of the output buffer stage, and a second circuit portion for generating a driving signal for PMOS transistor of the output buffer stage. Moreover, each of the first and second circuit portions may comprise a complementary pair of MOS transistors coupled between supply voltage references, with the complementary pair of MOS transistors defining an interconnection node therebetween. Each circuit portion may also include a logic network to generate an activation signal, and a third MOS transistor having conduction terminals coupled between one of the voltage references and the interconnection node, and a control terminal responsive to the activation signal.
A method aspect is for driving an output buffer stage comprising a PMOS transistor and an NMOS transistor coupled together. The method may comprise generating a driving signal for the NMOS transistor of the output buffer stage using a first circuit portion, and generating a driving signal for PMOS transistor of the output buffer stage using a second circuit portion. The first and second circuit portions may be as described above.
The characteristics and advantages of the driving circuit according to the invention will be apparent from the following description of an embodiment thereof given by way of indicative and nonlimiting example with reference to the annexed drawings.
With reference to such figures, and in particular to the examples of
The circuit portions 1A and 1B are respectively provided for the generation of a driving signal to be applied to the gate terminal of a corresponding transistor of the buffer stage. More particularly, the portion 1A of
Advantageously, thanks to the mode of generation of the signals GATEN and GATEP, the buffer stage 2 is active at high speed with respect to the approaches proposed by the known techniques. Moreover, reduced noise is induced on the power supply.
Advantageously, the driving circuit does not use resistors and it allows generating the control signals GATEN and GATEP for the output buffer stage 2 through CMOS structures only.
As it has been previously said, the need of obtaining low output impedances has, as a consequence, a high gate capacitance seen by the signals GATEN and GATEP. This peculiarity, together with the high switching speed required for multi-level memories, has led to the realization of a control of a buffer 2 by signals GATEN and GATEP with only MOS transistors and a low Rdson.
Such transistors M1 and M3 receive, in parallel on the respective gate terminals, a signal DRIVE_N coming from the output of a logic gate NR2P. The logic gate NR2P is of the NOR type and it receives, on a pair of inputs A, B, an input signal IN and an enable signal ENAB_N.
The output Z of the logic gate NR2P is branched off through a series of inverters IV forming a delay chain 3 having a signal output connected to the input of a second logic gate OR2. The other input of such logic gate OR2 directly receives the signal issued from the first logic gate NR2P.
The output of the second logic gate OR2 issues a signal DRIVE_PN_PULSE and it is connected to the gate terminal of a third MOS transistor, of the PMOS type and indicated with M2, whose conduction terminals are respectively connected to the first voltage reference VDDQ and to the interconnection node O between the transistors M1 and M3 of the output complementary pair.
The structure of the other circuit portion 1B intended for the generation of the signal GATEP is substantially similar with respect to the previous one and for it the same reference numbers will be used to refer to circuit components identical to those of the circuit portion 1A.
Also portion 1B comprises a complementary pair of output MOS transistors M1, M3 connected in series to each other between a first supply voltage reference VDDQ and a second voltage reference, for example a signal ground GNDQ.
Such transistors M1 and M3 receive, in parallel on the respective gate terminals, a signal DRIVE_P coming from the output of a logic gate ND2P. The logic gate ND2P is of the NAND type and it receives, on a pair of inputs A, B, an input signal IN and an enable signal ENAB_P. The output of the logic gate ND2P is branched off through a series of inverters IV forming a delay chain 3 having a signal output connected to the input of a second logic gate AN2. The other input of such logic gate AN2 directly receives the signal issued from the first logic gate ND2P. The output of the logic gate AN2 issues a signal DRIVE_NP_PULSE and it is connected to the gate terminal of a third MOS transistor, of the NMOS type and indicated with M5, whose conduction terminals are respectively connected to the interconnection node O between the transistors M1 and M3 of the output complementary pair and to the second voltage reference GNDQ.
The driving method is carried out by the generation of respective control signals GATEN and GATEP applied to the gate terminals of the transistors of the buffer stage 2 and having a linear pattern at intervals with only two slopes. In a first stage, the steepest slope is obtained by the transistors M1 and M2, this latter being enabled through the pulse DRIVE_PN_PULSE, for portion 1A and by the transistors M3 and M5, this latter being enabled through the pulse DRIVE_NP_PULSE, for portion 1B.
The width and the delay of the above pulses, with respect to the signals DRIVE_N and DRIVE_P, are obtained by the delay chain 3 formed by the inverters IV and by the logic gates, in one case of the NOR type and in the other of the NAND type with two inputs.
The circuit, intended as a whole, i.e. as if it were formed by the two portions 1A and 1B, each being connected at the output to the respective gate terminal of the buffer stage 2, ensures also that the pulse timely shows up after the signals DRIVE_N and DRIVE_P, avoiding in such way that excessive OVERLAP currents are generated.
To better understand the operation of the buffer 2, once it is driven with the circuit according to the invention, the results of some simulations are shown. In particular,
In the output switching from the high value to the low value, in particular, the pulse intervenes when the NMOS transistor of the buffer stage 2 has not entered the conduction yet, this allowing acceleration of the output switching.
From the previous description it can be easily deduced how the driving circuit according to the invention addresses the problems and attains several advantages among which the following ones are to be particularly cited.
Although preserving the advantages of some of the prior art approaches, the circuit according to the invention may ensure the possibility of:
1. obtaining low output impedances;
2. minimizing the disturbances induced in the power supply; and
3. reducing the occupation in terms of layout of the buffer.
Moreover, with this approach the problem of the voltage amount generated by the resistors used in the circuits of the prior art is not present, and, thus, using a control signal with three slopes is no longer necessary.
Number | Date | Country | Kind |
---|---|---|---|
MI2004A2075 | Oct 2004 | IT | national |
Number | Name | Date | Kind |
---|---|---|---|
5389834 | Kinugasa et al. | Feb 1995 | A |
5430387 | Bechade et al. | Jul 1995 | A |
5804990 | Popat et al. | Sep 1998 | A |
5926651 | Johnston et al. | Jul 1999 | A |
6653873 | Nguyen | Nov 2003 | B1 |
6720794 | Seike | Apr 2004 | B2 |
7053679 | Rho | May 2006 | B2 |
Number | Date | Country | |
---|---|---|---|
20060091910 A1 | May 2006 | US |