FIELD OF THE INVENTION
This invention relates generally to the field of electrical switching. More particularly, this invention relates to a method and apparatus for reducing charge injection in a Field-Effect Transistor (FET) switch.
BACKGROUND
Field-Effect Transistor (FET) switches often introduce undesired voltage transients. For example, an N-channel, enhancement mode, Metal-oxide Semiconductor Field-Effect Transistor (MOSFET) switch is turned on by applying a gate drive signal to the gate of the FET. The gate drive signal makes a voltage transition from a supply level to a level above the analog signal to be switched. When the FET switch turns off, the opposite transition is made. In either case, part of the drive signal transition occurs while the switch FET is on, and part while the switch FET is off. For the ‘off’ part of the transition, the gate-to-drain capacitance of the FET couples into the input node and injects charge into the input, causing a voltage transient. On the other hand, for the ‘on’ part of the drive signal transition, the sum of the gate-to-drain, gate-to-source and gate-to-channel capacitances of the FET couples into the input node and injects charge into the input, causing a voltage transient.
One approach to reducing or eliminating the charge injection is to use a compensating FET and a capacitor. In this approach, the gate drive voltage of the compensation FET and/or capacitor is equal in magnitude to that of the switch FET but opposite in direction. The length of time the FETs are on varies with the input signal level and therefore changes the total amount of charge transfer. Consequently, this compensation technique will be less effective for some voltages than others. Another approach uses a programmable digital-to-analog converter (DAC) in an auto-calibration loop. A zero voltage level is applied to the high impedance input of the DAC. Measurements are then made using the analog-to-digital converter (ADC) of a digital multi-meter while the switch is toggled on and off. A programmable capacitor is adjusted until the reading is zero. This technique is expensive, due to external components, and requires a calibration algorithm. Additionally, the speed of compensation is limited, so high frequency injection is not well compensated.
OVERVIEW OF CERTAIN EMBODIMENTS
The present invention relates generally to the compensation of charge injection in FET switches. Objects and features of the invention will become apparent to those of ordinary skill in the art upon consideration of the following detailed description of the invention.
In one embodiment of the invention a switch includes a switch FET and two compensating FETs coupled to an input node. Gate drive signals for the two compensating FETs are generated by a gate drive circuit dependent upon the analog input signal and gate drive signal to the switch FET.
BRIEF DESCRIPTION OF THE DRAWINGS
The novel features believed characteristic of the invention are set forth in the appended claims. The invention itself, however, as well as the preferred mode of use, and further objects and advantages thereof, will best be understood by reference to the following detailed description of an illustrative embodiment when read in conjunction with the accompanying drawing(s), wherein:
FIG. 1 is a schematic diagram of a charge compensation circuit in accordance with certain embodiments of the invention.
FIG. 2 is a diagram depicting a switch FET gate drive signal in a charge compensation circuit in accordance with certain embodiments of the invention.
FIG. 3 is a diagram depicting further gate drive signals in a charge compensation circuit in accordance with certain embodiments of the invention.
FIG. 4 is a timing diagram depicting logic signals in a charge compensation circuit in accordance with certain embodiments of the invention.
FIG. 5 is a schematic diagram of a gate-drive circuit in accordance with certain embodiments of the invention.
FIG. 6 is a graph showing reduction of charge injection by use of a charge compensation circuit.
DETAILED DESCRIPTION
While this invention is susceptible of embodiment in many different forms, there is shown in the drawings and will herein be described in detail one or more specific embodiments, with the understanding that the present disclosure is to be considered as exemplary of the principles of the invention and not intended to limit the invention to the specific embodiments shown and described. In the description below, like reference numerals are used to describe the same, similar or corresponding parts in the several views of the drawings.
FIG. 1 is a simplified schematic diagram of FET switch 100 incorporating an exemplary charge compensation circuit. Referring to FIG. 1, the signal to be switched is applied to input node 102 and denoted as Vin. The switch 100 employs two separate compensation circuits that act as “dummy capacitors” and are used to minimize charge injection into the input node 102. In this embodiment, the first compensation circuit or dummy capacitor comprises a FET 104. The drive for the first dummy capacitor is provided on the line 106 by a gate drive circuit 150 described below with reference to FIG. 5. Referring to FIG. 1, the gate drive signal for FET 104 is denoted by gatexlo. The drain of the FET 104 is coupled to the input node 102. The FET 108 provides the second dummy capacitor. The drain, channel and source of the FET 108 are all coupled to the input node 102. The drive for the FET 108 is provided on the line 110 by the gate driver circuit 150. The gate drive signal for FET 108 is denoted by gatexhi. An optional capacitor 112, which may be a poly/metal1/metal2 capacitor of variable capacitance, for example, is included to allow for ‘fine tuning’ of the capacitance provided by the dummy capacitors. The switching function itself is provided by the FET 114. The gate drive for the FET 114 is provided by the gate drive signal 116, and is denoted by the signal m1gate. The ‘x’ designation in the signals gatexhi and gatexlo signifies that these signals move opposite to the gate drive signal m1gate of the switch FET 114. That is, when the signal m1gate is rising, the signals gatexhi and gatexlo are falling and vice versa. The ‘hi’ and ‘lo’ designations indicate the part of the m1gate signal transition for which the gatexhi and gatexlo signals are being used to balance the charge injection, as will be explained below. Capacitors 120 and 122 represent stray capacitance and circuit capacitance on the Vin node 102 and Vout node 118, respectively.
In one embodiment, the FETs 104, 108 and 114 are of the same design to allow for accurate capacitance matching and tracking. The FETs may be Metal-Oxide Semiconductor Field-Effect Transistors (MOSFETs) or Junction Field-Effect Transistors (JFETs), or similar gate-controlled devices.
It will be apparent to those of ordinary skill in the art, that any of the FETs 104, 108 and 114 may, in practice, be an assembly of devices constituting a switch. For example, the FET may be a pair of back-to-back DMOS devices, a pair of MOSFET devices, or a CMOS device made of an NMOS device in parallel with a PMOS device. In the sequel the term “FET” will be taken to include a combination of devices forming a switch.
The operation of the circuit in FIG. 1 is now described for an exemplary embodiment. When the switch FET 114 turns on, the signal m1gate makes a voltage transition from a negative supply level VSS2 to approximately 10V above the analog signal to be switched (Vin). The switch is turned on when the gate signal is at a switching voltage level Vs=Vin+Vth, where Vth is a threshold voltage level. When the switch FET 114 turns off, the opposite transition is made. In either case, part of the m1gate transition occurs while the switch FET is on, and part while the switch FET is off. For the ‘off’ part of the transition, only the gate-to-drain capacitance of the FET 114 couples into the input node Vin and causes a voltage transient. On the other hand, for the ‘on’ part of the m1gate transition, the sum of the gate-to-drain, gate-to-source and gate-to-channel capacitances of the FET 114 couples gate voltage changes into the input node Vin and causes a voltage transient. If a single charge balancing capacitor is used, and it is driven between the same voltages as m1gate (but in the opposite direction), optimal charge injection compensation will occur only for one value of the input signal level. However, good performance is achieved over the entire range of input signal levels if charge injection compensation is performed separately for the two distinct parts of the m1gate voltage transition.
In the switch circuit, the dummy capacitance driven by the signal gatexlo compensates for the switch FET charge injection for the ‘off’ part of the m1gate signal transition, whereas the dummy capacitance driven by the signal gatexhi compensates for the switch FET charge injection for the ‘on’ part of the m1gate signal transition. As the signal m1gate rises (switch turning on), the gatexlo falls from the switching level Vs to the level VSS2, and gatexhi falls from the turn-on level (Vbias volts above the switching level, or approximately 10V above the input signal level) to the switching level. This equalizes the compensation charge, and the charge injected by the switch FET, for all levels of the input signal.
The gate control signals gatexlo, gatexhi and m1gate are generated by gate drive circuit 150. The drive circuit 150 receives the analog input signal Vin at input 152, a digital (logic) switching signal m1 at input 154, and voltage supply signals VSS1 and VSS2 at input 156 and 158, respectively. An exemplary embodiment of the gate drive circuit 150 is described below with reference to FIG. 5.
The gatexlo and gatexhi signals, in addition to transitioning between the correct voltage levels, are also timed correctly to properly balance the injected charge. The relative timing of the signals is shown in FIG. 2 and FIG. 3. The figures show the voltage V of the gate drive signals as a function of time, t when the switch is turned on and off. Referring to FIG. 2, the switch turn-on is initiated at time t1 when the digital signal m1 (154 in FIG. 1) goes high. The m1gate signal 204 that controls the switch FET begins to ramp upwards. At time t2 the m1gate signal 204 rises to the level 202 of the switching voltage, Vs=Vin+Vth, such that the voltage Vgs from the FET gate to its source is equal to Vth and the device just turns on. The switch turn-off is initiated at time t3 when the digital signal m1 (154 in FIG. 1) goes low. The m1gate signal 204 begins to ramp downwards. At time t4 the m1gate signal 204 falls below the level 202 of the switching voltage Vs, such that Vgs<Vth and the device turns off.
Corresponding gatexlo and gatexhi signals are shown in FIG. 3. The gate driver circuit is designed such that, when the m1gate signal starts to rise at time t1, the gatexlo signal 304 falls immediately, as shown in FIG. 3, but the gatexhi signal 302 does not fall until m1gate is above the switching voltage level at time t2. When the m1gate signal starts to fall at time t3, the gatexhi signal 302 rises immediately to a maximum level Vbias volts above the switching level Vs, but the gatexlo 304 does not rise until m1gate is below the switching voltage level at time t4.
In one embodiment, the FETs are of the same type. In this embodiment the gate drive signals are given by:
gatexhi=Vs+Vbias+c1.(Vs−m1gate)
gatexlo=(1−c1).(Vs−m1gate)+VSS2
where
The net voltage change of gatexhi, gatexlo and m1gate is zero and so the net charge injection will be zero.
A signal indicating if the m1gate signal is above or below the switching level Vs may be obtained by passing the m1gate signal and a signal at the switching voltage level to a comparator. The output of the comparator is denoted by the logic signal c1. FIG. 4 is a timing diagram showing the comparator output c1 and a digital (logic) switch signal m1 used to activate the FET switch. The assertion of the digital switch signal m1 at t1 (a signal is asserted when it takes the logic value ‘true’, which is the value 1 for positive logic, and is de-asserted when it takes the value ‘false’) causes the m1gate signal to start rising. The de-assertion of the digital switch signal m1 at t3 causes the m1gate signal to start falling. The comparator output c1 is asserted when the m1gate signal is greater than the switching voltage level. The gatexhi signal should be falling, or at its minimum, during the time period t2<t<t3 i.e. the period when both m1 and c1 are asserted. The signal denoted as h is asserted during this period. The gatexlo signal should be falling, or at its minimum, during the time period t1<t<t4 i.e. the period when either m1 or c1 is asserted. The signal denoted as l is asserted during this period. In the gate drive circuit, the signals h and l may be obtained using simple logic circuits and using control voltage levels.
A truth table showing operation of the corresponding logic circuit is given in Table 1. The inputs to the logic circuit are the digital switching signal, m1, and the output of the comparator, c1. Starting from the ‘off’ position, the signal l is asserted first and de-asserted last.
TABLE 1
|
|
lh
Conditionm1c1m1 OR c1m1 AND c1
|
switch off0000
initial turn-on1010
switch on1111
initial turn-off0110
switch off0000
|
FIG. 5 is a schematic diagram of an exemplary gate driver circuit 150 used to generate the gate drive signals m1gate, gatexhi and gatexlo. The analog input signal Vin enters at input 152. A level shifter 504 provides a voltage level at node 506 that is approximately 9V above the analog input signal level Vin. The analog input signal 152 is also coupled to unity gain buffer 508 to provide a buffered version of the analog input signal at node 510. An always-on current source 512 establishes a voltage at node 514, which is one diode drop above the analog input signal level. Another always-on current source 516, of equal value, establishes a voltage at node 518, which is one diode drop below the analog input signal level. The digital m1 signal 154 controls the operation of current sources 522 and 524. When the digital m1 signal is high, the m1 current supply 522 is enabled and the m1bar current supply 524 is disabled. When the digital m1 signal is low, the m1 current supply 522 is disabled and the m1bar current supply 524 is enabled. Prior to the switch FET coming on, the digital m1 signal 154 is low, hence the m1gate signal 116 is pulled to the bottom rail VSS2 by the m1bar current source 524 and the m1 current source 522 is off. In one embodiment of the gate drive circuit 150, the current sources shown in FIG. 5 are provided by the output of FET current mirrors, and so behave like simple resistor pullups/pulldowns when the voltage across them falls below the level required to keep the output FET in saturation. The output signal l of the OR gate 528 controls the current sources 530, 532 and 534. When l is high the source 530 is on; when l is low the current sources 532 and 534 are on. With m1 low and c1 low, the signal l is low. Hence the l current source 530 is off and the lbar current sources 532 and 534 are on. This establishes the analog signal level Vin+Vth at the gatexlo output 106. The output signal h of the AND gate 536 controls the current sources 540, 542 and 544. When h is high the sources 540 and 542 are on, when h is low the current source 544 is on. Hence, with m1 low, the output signal h of the AND gate 536 is also low. The h current sources 540 and 542 are off and the hbar current source 544 is on. This establishes a voltage level of Vs plus 9V plus one diode drop on (i.e. the static turn-on level) on the gatexhi output 110.
A switch turn-on is initiated by the m1 digital signal 154 going high. This turns off the m1bar current source 524 and turns on the m1 current source 522, causing the m1gate signal 116 to ramp upward from the voltage level VSS2. The output l of the OR gate 528 goes high, turning on the l current source 530 and turning off the lbar current sources 532 and 534. This causes the gatexlo signal 106 to ramp downwards from the switching level Vs to the negative supply level VSS2. The output signal h of the AND gate 536 stays low until the level of the m1gate signal 116 exceeds the switching level Vs, at which time the output from comparator 548 switches. The signal h goes high, turning on the h current sources 540 and 542 and turning off the hbar current source 544. This causes the gatexhi signal 110 to ramp downwards from the turn-on level to the switching level Vs as the m1gate signal continues to ramp positively to the static turn-on level.
A switch turn-off is initiated by the m1 digital signal 154 going low. This turns on the m1bar current source 524 and turns off the m1 current source 522, causing the m1gate signal 116 to ramp downwards from the static turn-on level. The output signal h of the AND gate 536 goes low turning off the h current sources 540 and 542 and turning on the hbar current source 544. This causes the gatexhi signal 110 to ramp upwards from the switching voltage level Vs=Vin+Vth to the static turn-on level. The output l of the OR gate 528 stays high until the level of the m1gate signal 116 falls below the switching voltage level, at which time the output from comparator 548 switches. In one embodiment of the invention, this is achieved by setting a threshold of Vth in the comparator, so that the comparator switches when m1gate=Vin+Vth. The signal l then goes low, turning-off the l current source 530 and turning on the lbar current sources 532 and 534. This causes the gatexlo signal 106 to ramp upward from the negative supply level VSS2 to the switching voltage level as the signal m1gate continues to ramp negatively to VSS2.
The diodes 560 in FIG. 5 are used to block current flow in particular states of the circuit and to set voltage levels.
In an alternative embodiment of the gate drive circuit 150, the comparator 548 switches when m1gate=Vin. In this embodiment, the capacitor 112 in FIG. 1 can be used to correct for the error introduced by neglecting the threshold voltage Vth of the switching FET. This approach works well when the threshold voltage Vth is constant over the entire input signal range.
FIG. 6 shows two graphs of “voltage transients” caused by charge injection in an N-channel enhancement mode MOSFET switch. The voltage is plotted as function of time. The upper graph shows the voltage due to charge injection without compensation. The peak voltage is approximately 25 mV. The lower graph shows the voltage when a charge compensation circuit is employed. The peak voltage is less than 1 mV.
Those of ordinary skill in the art will recognize that the present invention has been described in terms of exemplary embodiments based upon use of MOSFET devices, current sources and logic circuits. However, the invention should not be so limited, since the present invention could be implemented using hardware component equivalents.
While the invention has been described in conjunction with specific embodiments, it is evident that many alternatives, modifications, permutations and variations will become apparent to those of ordinary skill in the art in light of the foregoing description. Accordingly, it is intended that the present invention embrace all such alternatives, modifications and variations as fall within the scope of the appended claims.