FR2783372A1 - Temperature stabilized RC oscillator, has capacitor,comparator, and mechanism for charging and discharging capacitor with constant current in dependence on output signal - Google Patents

Abstract

The RC oscillator comprises a capacitor (C), a comparator (A2) providing an output signal (CLK) which is a function of voltage on the capacitor. There are threshold voltage sources (ST1, ST2) (MP1, MP2, MN2, MN3, PG2) for discharging the capacitor as a function of output signal (CLK), with constant current (10). The value of the constant current is independent of temperature. The comparator receives the voltage on capacitor at one input, and a voltage determined by the first reference voltage (Vref1), and the potential divider (R2, R3), at the other. The comparator is a Schmitt trigger circuit. The constant current source comprises a resistor (R) and a branch (A1, MN1) applying the compensation voltage (Vref2) which varies as a function of temperature in the same manner as the resistance (R). The voltage sources (ST1, ST2) comprise three branches. The first branch contains a series circuit of a PMOS transistor, an NMOS transistor connected as a diode, and a bipolar transistor connected as a diode. The second branch contains a series circuit of a PMOS transistor connected as a diode, an NMOS transistor, a resistor, and a bipolar transistor connected as a diode. The third branch contains a series circuit of a PMOS transistor, a resistor, and a bipolar transistor connected as a diode. The ratio of surfaces of emitters of the second and the first bipolar transistors is equal to 8, and the ratio of the second and the first resistances is equal to 10 in the case of the first source (ST1), and different from 10 in the case of the second source (ST2).

Description

AT-

  TEMPERATURE STABILIZED RC OSCILLATOR

  The present invention relates to an RC oscillator, used in particular for generating a signal

clock for a microcontroller.

  Such an oscillator known in the state of the art is for example described in FIG. 1. It comprises a capacitor C1, a first armature of which is connected to the input of an inverter comparator COMP1, such as a hysteresis comparator ( Trigger de Schmitt), and whose second armature is connected to ground. It also includes a resistor R1 connected between a positive supply terminal receiving a supply voltage VDD and the first armature of C1, for charging the capacitor. Finally, it includes a switch such as an N-type MOS transistor, referenced Mi, which is connected in parallel on C1, that is to say that its drain and its source are respectively connected to the first and to the second armature. of Cl. It receives on its control gate a control signal delivered by the output of an inverter INV1 disposed at the output of the comparator COMP1, of

  so as to discharge the capacitor.

  Thus, the charge and discharge of Cl are controlled as a function of the signal CLK output from the oscillator, this being the signal delivered by the inverter INV1. More particularly, the capacitor C1 charges through R1 when the signal CLK is in the low logic state, and it is discharged by M1 when CLK is in the high logic state. In general, the value of Ri is high so that the charge of C1 is relatively slow, and the channel width of M1 is large, so that the discharge of Cl is relatively rapid. Both the charge and the discharge of Cl are exponential. The capacitor C1 and the resistor R1 are generally discrete components connected to the outside of the housing of the microcontroller. Indeed, given their respective typical values (for an 8 MHz oscillator by

  example), these are quite large components.

  This is why we often talk about an external oscillator

  to designate such an RC oscillator.

  For a capacitor C1 of determined value, the frequency of the oscillations is a function of the charging time and the discharging time. The first is determined in particular by the values of VDD and Ri, and the second in particular by the value of the conduction resistance of M1, the two being also determined by the threshold values of COMP1. These threshold values are determined by the reference voltage values generally obtained using a resistance bridge arranged between the supply terminal.

  positive and the mass (called "divider bridge").

  However, the resistance R1 and, to a lesser extent, the conduction resistance of Mi and the threshold values of COMP1 are dependent on the temperature. We are talking here of course about the operating temperature of each of these components. However, according to an acceptable approximation, it can be considered that it is the operating temperature of the oscillator as a whole. We know that this temperature increases during operation, due in particular to energy dissipation by the Joule effect.

in resistive components.

  In practice, there is a frequency drift due to phenomena related to temperature which can reach 10% of the nominal value of the frequency of the oscillator. Such a drift constitutes a defect which

  can be unacceptable in certain applications.

  The present invention aims to improve, from this point of view, the oscillators known in the prior art. To this end, the invention provides an RC oscillator comprising a charge capacitor, a comparator delivering an oscillating signal as a function of the voltage across the terminals of this capacitor and of at least one threshold voltage, and which comprises means for charging and / or means for discharging the capacitor as a function of the oscillating signal with a constant current, of value

independent of temperature.

  In one embodiment, a first input of the comparator receives at least one reference voltage and its second input receives the voltage across the charge capacitor, the oscillating signal being the signal at the output of this comparator. The means for charging the charge capacitor comprise a current source delivering a constant current, of value independent of the temperature, and means for charging the charge capacitor with this constant current. Thus, the capacitor is charged at a constant value current, whatever the operating temperature. It follows that the charging time of the capacitor becomes more independent of

this temperature.

  In one embodiment, the constant current source comprises a load resistance and means for applying across its terminals a compensation voltage varying as a function of the temperature substantially in the same way as the load resistance. Thus, the variation in the value of the load resistance as a function of the temperature is compensated by a correlative variation in the compensation voltage applied to these terminals. Ohm's law (U = RxI) then ensures independence from the temperature of the value of the current flowing through the load resistor. The oscillator includes a voltage source adapted to deliver the compensation voltage. According to another advantageous characteristic of the invention, the means for discharging the charge capacitor also comprise the source of constant current and means for discharging the charge capacitor by this constant current. In this way, the discharge of the charge capacitor is also carried out at a constant value current. Thus, the discharge time of the charge capacitor becomes

  also more independent of temperature.

  According to another advantageous characteristic of the invention, the reference voltage (s) applied to the first input of the comparator is (are) independent of the temperature. The comparator threshold (s) are therefore more independent of the temperature. The oscillator according to the invention comprises means for generating at least one voltage

  with the above property.

  The characteristics of the above invention, taken in isolation or in combination, make it possible to stabilize the oscillator in temperature, that is to say to substantially reduce the dependence link between on the one hand the variations of the temperature of operation due to the heating of the resistive components and on the other hand the variations of the

  frequency of the oscillator output signal.

  Other characteristics and advantages of the invention will become apparent on reading the

  description which follows. This is purely

  illustrative and should be read with reference to the accompanying drawings in which there is shown: - in Figure 1, already analyzed: the diagram of an RC oscillator according to the prior art; - in Figure 2: the diagram of an RC oscillator according to the invention; - Figures 3a and 3b: the equivalent diagram of two CMOS switches used in the diagram of Figure 2; - in Figure 4: the circuit diagram of a voltage source, used according to two different variants to produce the voltage sources delivering, respectively, the compensation voltage and the (or

the) reference voltage (s).

  In Figure 2, there is shown the diagram of an oscillator according to the invention. This comprises a charge capacitor C, a first armature of which is connected to a node A and the other armature of which is connected to ground. In the following, the voltages mentioned are referenced with respect to this mass. The non-inverting input of an operational amplifier A2 is also connected to node A. A resistor R2 is connected in series with a resistor R3 between a terminal receiving a reference voltage Vrefl and ground. These two resistors therefore form a dividing bridge. The inverting input of A2 is connected to the common node B between the two resistors R2 and R3. The output of amplifier A2 delivers an oscillating signal CLK which is the output signal of the oscillator. On the left of the figure, a resistance of

  load R is connected between a node D and the ground.

  The inverting input of an operational amplifier A1 and the source of an N-type MOS transistor, referenced MN1, are also connected to node D. The non-inverting input of A1 is connected to the output of a voltage source ST2 delivering a voltage Vref2, called compensation voltage. This voltage Vref2 varies as a function of the operating temperature of the oscillator in substantially the same way as the load resistance R, as will be seen below. The amplifier A1 and the transistor MNl have the function of applying the voltage Vref2 across the terminals of the resistance R. It is quite obvious that other means for achieving this function can be envisaged. We denote by Io the current flowing through the resistor R. A P-type MOS transistor, referenced MPl, is connected by its source to a terminal receiving a supply voltage VDD and is connected by its drain to the drain of the transistor MNl. The transistor MP1 is mounted "on a diode". It will be recalled that such an arrangement consists, for a MOS transistor, in connecting the control gate to the drain of the transistor. Two other P type MOS transistors, respectively referenced MP2 and MP3, have their sources connected to terminals receiving the supply voltage VDD. The control grids of MP2 and MP3 are linked to that of MPl. As will be understood, the transistors MP1, MP2 and MP3 are mounted as a current mirror. The drain of the transistor MP2 is connected to the drain of an N-type MOS transistor, referenced MN2, the source of which is connected to ground. The transistor MN2 is mounted as a diode, so that the current Io is found between the

MP2 and MN2 drains.

  The drain of the MP3 transistor is connected to the input of a CMOS switch going to the low logic state, referenced PG1. The diagram and the operation of such a switch are detailed below with reference to FIG. 3b. The output of the switch PGl is connected to point A. As will be understood, a current of the same value as Io flows through the switch PGl when

it is closed.

  The input of a CMOS switch passing to the high logic state, referenced PG2, is connected to node A. Its output is connected to the drain of an N-type MOS transistor, referenced MN3, the source of which is connected to the mass. The diagram and the operation of such a switch are described below with reference to FIG. 3a. The control gate of MN3 is linked to that of MN2, so that these transistors operate as a current mirror. In this way, a current of the same value as Io flows through the switch PG2 when the latter is on. Finally, a CMOS switch passing to the low logic state, referenced PG3, is connected by its input and its output in parallel on the resistor R2. In this way, the resistor R2 is short-circuited when PG3 is closed. The control inputs of switches PG1, PG2 and PG3 are connected to the output of the amplifier

  operational A2 to receive the CLK output signal.

  With reference to FIGS. 3a and 3b, we will now describe the diagram of an embodiment as well as the operation of the CMOS switches of the diagram of the

figure 2.

  In Figure 3a, there is shown the diagram of an embodiment of a CMOS switch passing to the high logic state. For simplicity, the switch PG2 of this type has been represented in FIG. 2 by a symbol which is reproduced on the left side of FIG. 3a. It has three connection terminals, namely an input terminal, an output terminal and a control terminal respectively denoted I, O and G. On the right-hand side of FIG. 3a, the actual diagram of the switch comprises two transistors MOS, one of type N referenced MN and the other of type P referenced MP. The MN drain and the MP source are connected together to terminal I. Likewise, the MN source and the MP drain are connected together to terminal O. Terminal G is connected directly to the MN control gate , and also the MP control grid via an inverter INV. In this way, the control grids of MN and MP receive

complementary signals.

  When a control signal carried on the terminal G is in the high logic state, the two transistors are conducting: one says then that the switch is closed or passing. When this signal is in the low logic state, the two transistors are blocked: we say that the

switch is open.

  In Figure 3b, there is shown the diagram of an embodiment of a CMOS switch passing to the low state. For simplicity, the switches PG1 and PG3 of this type have been represented in FIG. 2 by a symbol which is reproduced on the left part of FIG. 3b. On the right side, the actual diagram of this switch differs from that of Figure 3a only by the inversion of the MN transistors

and MP.

  Thus, when the control signal carried on terminal G is in the low logic state, the switch is closed. And it is open otherwise. This difference between the two types of switch is recalled by the presence of a small circle on the terminal G of the switch of FIG. 3b, by analogy with the classic distinction between the symbols of the N type MOS transistors on the one hand and of type P

on the other hand.

  Functionally, the terminals I and O are interchangeable, the terms "input" and "output" referring only to the direction of current flow in the

switch when closed.

  As can be seen, the CMOS switches of FIGS. 3a and 3b fulfill the function of a controlled switch. Any other embodiment fulfilling this function can be envisaged. However, these CMOS switches are preferred because they have low static consumption and good switching speed, both on rising edges and on

  Falling edges of the control signal.

  We now return to the diagram in Figure 2.

  The voltage source ST2 on the one hand, the resistance R on the other hand, and finally the means of applying the voltage Vref2 delivered by ST2 at the terminals of the resistor R (constituted by the amplifier Al and the transistor MNl) , fulfill the function of a constant current source of value Io. As the voltage Vref2 varies as a function of the temperature substantially like the resistance R, the value of the current Io is substantially independent of the temperature by application of Ohm's law (U = RxI). By the expression "substantially" used above, it should be understood that the law of variation of Vref2 as a function of the operating temperature is identical in the first order to the law of variation of R as a function of this same temperature. This is a linear law. To describe the operation of the oscillator, it

two separate cases should be considered.

  Assume first that the capacitor C is completely discharged (situation when the oscillator is powered up). The voltage at node A is then equal to the zero voltage. Since the voltage at node B is strictly positive as being equal to the voltage drop across the terminals of R3, the signal CLK at the output of amplifier A2 is in the low logic state. Consequently, the PG2 switch is open and the

  PG1 and PG3 switches are closed.

  This means that capacitor C charges through MP3 and PG3. The tension at node A rises. It will be noted that, due to the mounting of the transistors MP1 and MP3 in current mirror, the charge of C is carried out at constant current, of value Io. In other words, the correctly controlled transistors MPl and MP3 and the switch PGl make it possible to charge the capacitor C with a constant current, of value substantially independent of the operating temperature. In addition, the resistor R2 being short-circuited by PG3, the voltage at node B is equal to Vrefl. Consequently, the signal CLK only changes logic state when the voltage at node A reaches the value of Vref2 by lower values. The charge of C is then maximum. In fact, as soon as CLK is in the high logic state, the switch PG2 is closed and the switches PG1 and

PG3 are open.

  This means that the capacitor C discharges through PG2 and the transistor MN3. The tension at node A drops. It will be noted that, due to the current mirror mounting on the one hand of the transistors MP1 and MP2 and on the other hand of the transistors MN2 and MN3, the discharge of C is carried out at constant current, of value Io. In other words, the transistors MP1, MP2, MN2 and MN3 and the switch PG2 allow the capacitor C to be discharged with a constant current, of substantially value

  independent of operating temperature.

  In addition, the voltage at node B is determined by the resistance values R2 and R3 connected in divider bridge and by the value of Vrefl. For resistors R2 and R3 of the same value, the voltage at II node B is 1/2 x Vrefl 1. Consequently, the signal CLK does not change logic state again until the voltage at node B reaches 1/2 x Vrefl by values

  higher. The charge of C is then minimal.

  We then understand that the voltage at node B oscillates between 1/2 x Vrefl and Vrefl, these two values constituting the thresholds of the hysteresis comparator formed by A2, R2, R3 and STl. It is an advantage to have these two threshold values determined by the only value of the voltage Vrefl delivered by STl. Indeed, this voltage which, in order to obtain thresholds of values substantially independent of the temperature, must itself have this characteristic, is generated by a voltage source of a particular type. It is therefore advantageous to generate the two threshold voltages of the hysterisis comparator with a single such voltage source. In fact, this saves the silicon surface, and reduces the

  static current consumption of the oscillator.

  In FIG. 4, there is shown the diagram of an embodiment of the voltage sources ST1 and ST2,

  delivering the voltages Vrefl and Vref2 respectively.

  This diagram, known per se, has three branches connected in parallel between a supply terminal

VDD and mass.

  A first branch B1 includes a P-type MOS transistor, referenced MP11, the source of which is connected to the positive supply terminal and the drain of which is connected to the drain of an N-type MOS transistor, referenced MN11. The transistor MN11 is mounted as a diode. The source of MN11 is connected to the emitter of a bipolar transistor PNP, referenced BP1, the collector of which is connected to ground. The transistor BP1 is also mounted "as a diode". It will be recalled that, for a bipolar transistor, such an arrangement consists in connecting together the base and the collector of the transistor. To summarize, it can be said that the first branch comprises, in series between the supply terminal and the ground, a P-type MOS transistor, an N-type MOS transistor and a diode (BP1). The second branch comprises a P-type MOS transistor, referenced MP12, the source of which is connected to the supply terminal and the drain of which is connected to the drain of an N-type MOS transistor, referenced MN12. The transistor MP12 is mounted as a diode. The source of MN12 is connected to a first terminal of a resistor R12, the second terminal of which is connected to the emitter of a bipolar transistor PNP, referenced BP2. The transistor BP2 is mounted as a diode and its collector is connected to ground. In other words, the last branch comprises, in series between the supply terminal and the ground, an N-type MOS transistor, a resistor and a diode (BP2). It will be noted that the surface of the emitter of transistor BP2 is significantly larger than that of transistor BP1. In one example, it is eight times larger. It will also be noted that the control gates of the transistors MN11 and MN12 are connected together. The third branch includes a P-type MOS transistor, referenced MP13, the source of which is connected to the supply terminal and the drain of which is connected to a first terminal of a resistor R13. The other terminal of R13 is connected to the emitter of a bipolar PNP transistor, referenced BP3. This is mounted on a diode and its collector is connected to ground. The output OUT of the circuit is taken on the drain of MP13. It releases a tension which presents certain characteristics

  related to the operating temperature.

  Indeed, we can show that, if we note V (OUT) the voltage at the output OUT of the circuit, V (OUT) = Vbe + (Rl3 / R12) x ln (8) x (k / q) x T o : Vbe designates the base / emitter voltage of a bipolar transistor; ln denotes the natural logarithm function; k denotes the Boltzmann constant; q denotes the charge of the electron;

  T designates the operating temperature.

  Now, Vbe varies as a function of the operating temperature at a rate of -2mV / C. In addition, ln (8) is roughly equal to 2. Finally, the term (k / q) x T which is sometimes called thermodynamic potential, varies depending on

  the operating temperature at 0.1 mV / C.

It is about 25 mV at 25 C.

  Therefore, if R12 and R13 have values such that R13 / R12 is substantially 10, the value of the output voltage is independent of the operating temperature. As will be understood, the voltage source ST1 in FIG. 2 can be produced according to the diagram in FIG. 4 with R13 substantially equal to 10 x R12. In addition, with an appropriate value of the ratio R13 / R12 substantially different from 10, it is possible to obtain a linear (first order) dependence of V (OUT) on the operating temperature of constant amplitude. Knowing that the value of a resistor is also a linear (first order) function of the operating temperature, we see that an appropriate voltage V (OUT) applied across the resistor R in Figure 2 can compensate for variations of R due to variations in the operating temperature. Indeed, as will be understood, the voltage source ST2 of FIG. 2 can be produced according to the diagram of FIG. 4 with R13 substantially different from 10 x R12 (for a ratio between the emitter surfaces of the bipolar transistors BP2 and BP1 substantially equal to 8), so as to deliver a voltage Vref2 varying as a function of the temperature as the resistance R. The skilled person knows how to calculate the values of the components R12 and R13 as a function of the value of R. If necessary, it can use appropriate simulation software. You can also modify the ratio of the emitter areas of BP1 and

of BP2.

  It will be noted that the value of R and C determine, as a function also of the values of Vrefl, R2 and R3, the frequency of the oscillations. The resistor R and the capacitor C can be discrete components connected to the outside of an integrated circuit, in particular a microcontroller, incorporating the oscillator according to the invention.

Claims (10)

R E V E N D I C A T ION 8   1. RC oscillator comprising a charge capacitor (C), a comparator (A2) delivering an oscillating signal (CLK) as a function of the voltage across this capacitor (C) and of at least one threshold voltage (Vrefl) , characterized in that it comprises means (MP1, MP3, PG1) for charging and / or means (MP1, MP2, MN2, MN3, PG2) for discharging the capacitor (C) according to the oscillating signal (CLK) with a constant current (Io), of value independent of the temperature. 2. Oscillator according to claim 1, characterized in that a first input of the comparator (A2) receives at least one reference voltage and a second input receives the voltage across the charge capacitor, the oscillating signal   (CLK) being the signal at the output of this comparator.   3. Oscillator according to claim 1 or according to claim 2, characterized in that the means for charging the charge capacitor (C) comprise a current source delivering a constant current (Io), of value independent of the temperature, and means (MPl, MP3, PGl) to charge the capacitor by this constant current.   4. Oscillator according to claim 3, characterized in that the constant current source comprises a load resistance (R) and means (Al, MNl) for applying across its terminals a compensation voltage (Vref2) varying according to the temperature in much the same way as the load resistance (R).   5. Oscillator according to claim 3 or according to claim 4, characterized in that the means for discharging the charge capacitor also comprise the constant current source and means (MP1, MP2, MN2, MN3, PG2) for discharging the capacitor of load (C) by this constant current.   6. Oscillator according to claim 2, characterized in that it comprises means for applying to the first input of the comparator (A2), either a first reference voltage (Vrefl), or a second reference voltage (1/2 x Vrefl) lower than the first reference voltage, as a function of the comparator output signal (CLK), so that the comparator operates as a hysteresis comparator. 7. Oscillator according to claim 6, characterized in that it comprises a divider bridge (R2, R3) for generating the second reference voltage in   function of the first reference voltage.   8. Oscillator according to claim 6, characterized in that it comprises means for generating the first and second reference voltages   with values independent of temperature.   9. Oscillator according to any one of   previous claims, characterized in that   comprises a voltage source (ST1) for delivering a voltage independent of the temperature, which comprises in series between a positive supply terminal (VDD) and the ground: - a first branch comprising in series a P-type MOS transistor (MP11 ), an N-type MOS transistor (MN11) mounted as a diode, and a first bipolar PNP transistor (BP1) mounted as a diode; a second branch comprising in series a P-type MOS transistor (MP12), an N-type MOS transistor (MN12), a first resistor (R12) and a second bipolar PNP transistor (BP2) mounted as a diode; - a third branch comprising in series a P-type MOS transistor (MP13), a second resistor (R13), and a third bipolar PNP transistor (BP3) mounted as a diode; and in that, for a ratio between the emitter surfaces of the second (BP2) and of the first (BP1) bipolar transistors substantially equal to 8, the ratio (R13 / R12) between the values of the second (R13) and of the first (R12) resistance is substantially equal to 10.   10. Oscillator according to any one of   previous claims, characterized in that   comprises a voltage source (ST2) for delivering a compensation voltage, which comprises in series between a positive supply terminal (VDD) and the ground: - a first branch comprising in series a P-type MOS transistor (MP11), an N-type MOS transistor (MN11) mounted as a diode, and a first bipolar PNP transistor (BP1) mounted as a diode; a second branch comprising in series a P-type MOS transistor (MP12), an N-type MOS transistor (MN12), a first resistor (R12) and a second bipolar PNP transistor (BP2) mounted as a diode; - a third branch comprising in series a P-type MOS transistor (MP13), a second resistor (R13), and a third bipolar PNP transistor (BP3) mounted as a diode; and in that, for a ratio between the emitter surfaces of the second (BP2) and of the first (BP1) bipolar transistors substantially equal to 8, the ratio (R13 / R12) between the values of the second (R13) and of the first (R12) resistance is substantially different from 10.