The invention is directed generally to off chip driver (OCD) circuits, and more particularly to a method and apparatus for compensating such circuits for changes in operating temperature.
When it is necessary to transmit signals off-chip, a circuit known as an off chip driver (OCD) may be used. To be effective, the OCD must deliver signals of sufficient strength and having timing which is within certain margins of error. Figures of merit which are used to define such margins of error include output skew matching ratio, skew between rise and fall delay, and tDQSQ, which will be discussed in further detail below.
An OCD may be implemented in CMOS design and have a group of PMOS field effect transistor (FET) fingers and a group of NMOS FET fingers connected together in push-pull circuit relationship. A problem encountered with such circuits is that the PMOS and NMOS transistors perform differently over different temperature ranges. Thus, at low temperatures (e.g., <5° C.) the PMOS device gets slower and the NMOS device gets faster, while at high temperatures (e.g., >50° C.) the NMOS device signal strength becomes weaker. Due to the different PMOS and NMOS characteristics over the operating temperature range of the circuit, it is difficult to get good performance for tDQSQ and there is a mismatch of the rise and fall times and of the output slew matching ratio.
In a prior design, fuse options are provided to adjust the timing parameters over different temperature, process, and voltage variations. However, such fuse options may not provide adequate control for temperature variations, with the result that the circuit may operate out of specification at certain temperatures.
In accordance with the present invention, a method of temperature compensating an off chip driver (OCD) circuit having a plurality of transistor fingers is provided which renders active a normally inactive transistor finger in the OCD circuit when a predetermined temperature condition occurs.
The invention will be better understood by referring to the accompanying drawings wherein:
The end driver in the particular embodiment shown is comprised of a group 8 of PMOS field effect transistor (FET) fingers connected in push-pull circuit relationship with a group 10 of NMOS FET fingers. Both groups of FET fingers include at least one relatively high power FET finger (for example a finger of 8× is depicted) and a plurality of relatively lower power FET fingers (e.g., fingers of 2× and 1× are depicted). The specific PMOS FET group 8 in
In the embodiment depicted, the source electrodes of the PMOS FET fingers are connected to voltage VDDQ while the source electrodes of the NMOS FET fingers are connected to VSSQ, which in the embodiment shown is at ground potential. When a low signal is applied on a conductor GP to the gate of a PMOS FET, the transistor is turned on. At the same time a low signal is applied on a conductor GN to the gate of a corresponding NMOS FET finger, causing the transistor to turn off. Thus, there is an output signal of voltage magnitude approximately VDDQ appearing between the common line 21 and ground.
On the other hand, when a high signal is applied on a line GN to the gate of an NMOS FET finger, the transistor is turned on, and when a high signal is applied to the gate of a corresponding PMOS FET finger, the transistor is turned off. Thus, there is an output signal of about ground potential on common line 21.
As mentioned above, when the end driver stage is used at operating speed, timing problems can result, and pre-driver stages 4 and 6 are utilized to remedy this. It should be understood that the invention is directed to improving the timing of the driver generally. While there are certain specific figures of merit relating to timing which will be discussed, these are exemplary only, as there are many ways to evaluate how good or bad driver timing may be. A first figure of merit which may be used is the output slew matching ratio which compares the slew of the leading edge of the output signal with the slew of the trailing edge, a matching ratio of “one” being perfect. A second figure of merit is the skew between the rise delay and the fall delay, which is a measurement of the delay between the rising edge of the system clock and the rising and falling edges of the data signal (TAC). Still a third figure of timing merit is tDQSQ which is a measure of the delay between the rising and falling edges of the data strobe signal DQS and the rising and falling edges of the data signal DQ.
Referring again to
Referring to
Additionally, NAND gates 42 and 44 have fuses <0> and <1> respectively inputted thereto. When these fuses are set high, during the occurrence of high DP data signals, low signals are outputted from NAND gates 42 and 44, thus turning FET fingers 14 and 16 on. The fuses are selectively set to adjust the timing parameters over different temperature, process and voltage variations. Thus a drive strength trimming range is provided. When the fuses are set high FET fingers 14 and 16 are rendered active in the circuit and when the fuses are set low fingers 14 and 16 are inactive.
However, it was found that when the fuse options were used alone, i.e., without the improvement of the present invention, the device may operate out of specification over certain portions of the temperature range. For example, this could be the case when the OCD is incorporated in a particular low power (LP) dynamic random access memory (DRAM), for which a temperature range of −30° C. to +85° C. is specified.
In accordance with the present invention, a normally inactive FET finger is rendered active in response to a predetermined temperature condition. This provides additional control of timing parameters responsive to operating temperature, as well as additional control of drive strength. As discussed above, when the temperature falls to below about −5° C. the PMOS devices get slower while the NMOS devices get faster. Thus, in the embodiment of
It is noted that pre-driver 6 is comprised of inverter 54, and NOR gates 48, 50, and 52. Data signal DN inputted to inverter 54 is the same as data signal DP discussed above, and the inverter also has an SR CTRL <0:1> input as discussed above. The DN signal is applied to one input of NOR gates 48, 50, and 52, while fuse options are applied to the other inputs of NOR gates 48 and 50. To provide the proper outputs on lines GN<1> and GN<2> the fuses would be set low if it is desired to render active FET fingers 38 and 40.
As discussed above, the NMOS device signal strength gets weaker above about 50° C., so the temperature control input to NOR gate 52 is arranged to go low when the operating temperature exceeds about 45° C., thus causing normally inactive FET finger 34 to be rendered active. This provides additional power to the NMOS devices to compensate for the power loss caused by rising temperature. In addition to improving the timing characteristics of the OCD, the present invention also improves the PU/PD current ratio, which relates to the current/voltage characteristics of the PMOS and NMOS devices.
It is noted that the term “temperature” as used herein refers to the operating temperature at the chip. Many chips have on-chip temperature sensors, thus making implementation of the invention easier.
It is also noted that while the illustrative embodiment depicts CMOS technology (PMOS and NMOS devices) the invention may be implemented in any type of circuitry which is comprised of transistor fingers. Further, the actual temperatures mentioned herein are illustrative only and other specific temperatures may be used.
The end driver and pre-driver stages operate as described in connection with
There thus has been described an improved method and apparatus for compensating OCD circuits for changes in operating temperature. The system and methods described herein may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The foregoing embodiments are therefore to be considered in all respects illustrative and not meant to be limiting.