Patent Application
20080278277
The present invention relates generally to resistors and in particular to methods and devices associated with fabricating digitally controllable on-chip resistors.
Resistors play a large part in almost all electronic circuits. In many cases the performance of a circuit is limited by the accuracy of the resistors which are available to implement the circuit. Complementary metal oxide semiconductor (CMOS) chip manufacturing processes are not currently capable of realizing precise resistance values. For example, values of resistors implemented in CMOS chips can vary by as much as 20-30% of their designed values.
Digitally controllable resistors, when implemented in CMOS to counter this probabilistic spread in CMOS resistor yields, rely on transistor switches to change their value according to control signals. However, even in their “on” state, these switches introduce some “on-resistance” in the signal path which may change the behavior of the circuit. Traditional methods try to reduce the effect of this on-resistance by increasing the channel width of the transistors in the switch, hence reducing their on-resistance. However, this also increases the parasitic capacitance of the switch. Thus, CMOS switches either have high parasitic capacitance or significant on-resistance, both of which may affect the performance of the digitally controllable resistors and/or circuit in which they are used.
These issues pose a problem in manufacturing precise resistor values on-chip, whereas the current growth of the telecommunications industry requires the manufacturers to include as much functionality on-chip as possible and avoid using off-chip components. Hence a method to implement precise, linearly variable on-chip resistance values is needed. In addition, temperature changes in electronic circuits during use cause a drift in the values of on-chip resistors. In order to combat this tendency, on-chip variable resistors that can be tuned reliably and accurately within a specified range are also needed.
Several existing approaches attempt to address these problems, some examples of which will now be described. For example, trimming is a post-processing (i.e., post manufacturing) step used to correct the values of on-chip passive components. However this processing adds greatly to the cost of the finished chip. Another approach involves using MOS transistors as variable resistors by biasing and sizing them appropriately. However, this approach is not suitable for applications where, for example, a linear/constant resistance step is needed for every increment in the digital control word because the parallel connection of binary weighted transistors results in non-linear resistance steps in the active resistance range.
A third approach used to address these problems with on-chip resistors involves using pulse width modulation (PWM) on a field effect transistor (FET) in series with a primary resistor. However, this approach has a drawback for communication systems given the possibility of additional noise due to clock feed through. Yet another approach is to use MOS transistors as active fuses to short out tuning resistors placed in series or parallel. However, this approach is not suitable for CMOS applications since implementing low-resistance switches consumes a large area on the chip and introduces considerable parasitic capacitance in the resistor, which may induce non-linear behavior.
Still another approach involves using grounded switched resistor strings. However, this technique causes constant current consumption in the variable resistor due to the ground terminals. This makes this approach unattractive for use in single ended and/or low power circuits. In addition, the number of passive (resistors) and active (switches) components in the circuit increases in an exponential manner as the number of bits in the digital control word increases linearly. Yet another approach uses a CMOS switch or transmission gate arrays as variable resistors. However, this approach uses a binary weighted structure resulting in non-linear resistance steps. Additionally, the transmission gate has non-linear voltage over current characteristics near the extremes of supply voltage range which may lead to a decrease in usable voltage swing.
Accordingly, it would be desirable to provide digitally controllable resistor methods and devices which achieve arbitrarily small, yet substantially linear, incremental resistance steps irrespective of the on-resistance associated with the switches.
According to an exemplary embodiment, a digitally controllable resistor includes a substrate, at least one digitally controllable resistance stage formed on the substrate, each of the at least one stages including a first resistor connected in series with a switch, a second resistor connected in parallel with the first resistor and the switch, and a control line connected to the switch for opening and closing the switch in response to a control bit associated therewith.
According to another exemplary embodiment, an integrated circuit chip includes a first circuit, disposed on the integrated circuit chip, for performing a function, the first circuit also capable of determining a compensating resistance value associated with performance of the function and generating a digital control word associated with the compensating resistance value; and a digitally controllable, variable resistor connected to the first circuit and including at least one digitally controllable resistance stage, each of the at least one stages including a first resistor connected in series with a switch, a second resistor connected in parallel with the first resistor and the switch, and a control line connected to the first circuit and to the switch for opening and closing the switch in response to a respective bit of the digital control word.
According to another exemplary embodiment, a method for compensating for an effect on an integrated circuit chip includes the steps of estimating a value associated with the effect, generating a digital control word associated with the value, and using at least one bit in the digital control word to operate a respective at least one switch in a digitally controllable, variable resistor, the variable resistor including at least one digitally controllable resistance stage, each of the at least one stages including a first resistor connected in series with one of the at least one switches, and a second resistor connected in parallel with the first resistor and the one of the at least one switches.
The exemplary embodiments described herein provide a number of potential benefits including, for example, the provision of a highly linear and digitally controllable resistor structure having a good frequency response which can be implemented in CMOS technology. The incremental resistance steps associated with the overall resistance of the digitally controllable resistor can be made arbitrarily small, irrespective of the on-resistance of the switch(es). Switches having a minimum channel width can be used in these exemplary architectures to reduce the parasitic capacitance in the resistor. This can provide a significant benefit for those exemplary applications where, for example, a precise RC constant is desirable. Additionally, the use of the digitally controllable resistors as described herein will increase the device yield and result in significant cost saving as compared to methods like trimming.
The accompanying drawings illustrate exemplary embodiments, wherein:
The following detailed description of the exemplary embodiments refers to the accompanying drawings. The same reference numbers in different drawings identify the same or similar elements. Also, the following detailed description does not limit the invention. Instead, the scope of the invention is defined by the appended claims.
According to exemplary embodiments, a highly linear and digitally controllable resistor structure having a good frequency response can be implemented in CMOS technology. The incremental resistance steps associated with the overall resistance of the digitally controllable resistor can be made arbitrarily small, irrespective of the on-resistance of the switch(es) which is effectively “absorbed”.
The term “CMOS” can be used to refer to a particular style of digital circuitry design, and/or to the family of processes used to implement that circuitry on integrated circuits (i.e., chips). Exemplary commercial CMOS products are integrated circuits having millions or hundreds of millions of n-type and p-type transistors on a substrate between, for example, 0.1 and 4 cm2 in size. In early CMOS fabrication processes, the gate electrode of these transistors was made of metal, e.g., aluminum. More recent CMOS processes switched from metal gate electrodes to polysilicon to better tolerate the high temperatures applied to the substrate after ion implantation. The CMOS substrate thus can include the metal (or polysilicon) layer disposed on top of an insulating oxide layer, which in turn is disposed on top of a semiconductor layer. There are several ways in which resistors can be implemented using CMOS technology. For example, polysilicon resistors can be constructed by depositing a layer of polysilicon on top of the CMOS substrate and adding contacts at both ends. Another way to fabricate resistors using CMOS technology is to implement them as N-well/P-well resistors. N-well/P-well resistors can be constructed by providing a layer of N- or P-doped semiconductor material over the substrate. The doping of the resistive material determines the resistivity (resistance per unit area) for a given process.
To fabricate a digitally controllable resistor according to these exemplary embodiments, a plurality of resistance stages or “building blocks” are provided on a CMOS substrate. An exemplary resistance stage 10 fabricated as integrated elements on a substrate 11, e.g., a CMOS elements on a CMOS substrate, is illustrated in
When the switch 14 is open, the switch resistance is high enough to be considered infinite for all practical purposes. In this case, the resistance between terminals A and B of the resistance stage 10 is Rup. However, when the switch 14 is closed, the effective resistance between terminals A and B of the resistance stage 10 is calculated by the following equation:
With an appropriate selection of the resistance values Rup and Rdown the difference between the two resistance values for stage 10 (i.e., the resistance value when the switch 14 is open and the resistance value when the switch 14 is closed) can be made equal to any desired step value (ΔR). This way, a resistance change of, for example, just a few ohms can be realized based on the position of the switch 14 irrespective of the value of its on-resistance Rswitch.
In order to fabricate a digitally controllable resistor with a larger resistance variation range than that which is provided by a single resistance stage device, the total resistance to be provided by the device can instead be divided between a plurality of the stages 10 fabricated on a substrate and connected together in series. An exemplary multi-stage digitally controllable resistor device 20 disposed on a substrate 21, e.g., a CMOS substrate, according to these exemplary embodiments is shown in
The effective resistance of all N stages in the exemplary digitally controllable resistor 20 is equal when all of the switches 14 are closed, i.e., the total effective resistance is uniformly distributed among all stages 10. This switch condition also provides the minimum resistance Rmin for the digitally controllable resistor 20. When some or all of the switches 14 are open, the effective resistance for each stage 10 is binary weighted, by selecting the resistance values as described in the equations below, to make the total resistance of the digitally controllable resistor 20 change linearly with the value of the digital control word. The maximum resistance (Rmax) is achieved when all the switches 14 are open. The intermediate resistance levels between Rmin and Rmax can be realized by varying the value of the digital control word between 0 and 2N−1.
To fabricate a multi-stage, digitally controllable resistor such as that shown in
Rmax is implicitly included in equations (2) by way of ΔR, Rmin and N. More specifically, the maximum resistance Rmax can be calculated as Rmax=Rmin+N*ΔR. Thus alternatives to equations (2) can be used to fabricate multi-stage, digitally controllable resistors according to exemplary embodiments. For example, the designer can either explicitly define Rmax and then determine ΔR or can define ΔR and determine Rmax.
These exemplary embodiments provide digitally controllable resistors having a number of beneficial qualities including, for example, linear voltage vs. current characteristics, good frequency response, low parasitic capacitance, linear resistance steps throughout the designed resistance range, and being completely monotonic over the whole range of the N-bit control word. To illustrate these characteristics an exemplary, digitally controllable resistor has been simulated using a 90 nm CMOS technology. This purely illustrative simulation was designed to have N=6 stages 10, an Rmin of 14 KΩ, a ΔR of 210Ω and an Rswitch of 1.9 KΩ.
a) and 4(b) illustrate the linearity of the resistance provided by the simulated, digitally controllable resistor described above. More specifically,
There are many different applications for digitally controllable resistors according to these exemplary embodiments. In addition to being used as a general purpose, digitally controllable, variable resistor, these devices can also be used in conjunction with other circuits. Thus, as shown generally in
Alternatively, the other circuit 60 can estimate the effect of temperature drift on the chip 66 and generate a unique control word (communicated via line 62) to control the resistance of unit 10 or 20 to minimize this effect. These, or other, types of tuning can be carried out in real time during operation of the chip 66. The other circuit 60 can be any type of other circuit which has a use for a controllable, variable resistor, e.g., a channel selection filter, examples of which can be found in, for example, the article entitled “Tunable, Multi-bandwidth channel select filter for an LTE radio receiver”, Section 6.2, F. Oredsson, I. Din, Lund University, 2006, the disclosure of which is incorporated here by reference.
Thus, it will be appreciated that, according to an exemplary embodiment, a general method for tuning a circuit can include the steps of
It will be appreciated that the foregoing embodiments are purely exemplary and that variations to the foregoing can be implemented. For example, minimum sized switches, i.e., switches having a minimum channel width, can be used in these exemplary architectures to reduce the parasitic capacitance in the resistor. This can provide a significant benefit for those exemplary applications where, for example, a precise RC constant is desirable. Additionally, the use of the digitally controllable resistors as described herein will increase the device yield and result in significant cost saving as compared to methods like trimming.
The above-described exemplary embodiments are intended to be illustrative in all respects, rather than restrictive, of the present invention. Thus the present invention is capable of many variations in detailed implementation that can be derived from the description contained herein by a person skilled in the art. All such variations and modifications are considered to be within the scope and spirit of the present invention as defined by the following claims. No element, act, or instruction used in the description of the present application should be construed as critical or essential to the invention unless explicitly described as such. Also, as used herein, the article “a” is intended to include one or more items.
