NL2002640C2

NL2002640C2 - Ultra low three points oscillator assembly, oscillator circuit, and electronic device.

Info

Publication number: NL2002640C2
Application number: NL2002640A
Authority: NL
Inventors: Johannes Wilhelmus Theodorus Eikenbroek; Siti Amalina Enche Ab Rahim
Original assignee: Bruco B V
Priority date: 2009-03-19
Filing date: 2009-03-19
Publication date: 2010-09-21
Also published as: WO2010105932A1

Description

Ultra low power three points oscillator assembly, oscillator circuit, and electronic device

The present invention relates to an ultra low power 5 three points oscillator assembly. It further relates to an oscillator circuit, connectable to a resonator, for forming a three points oscillator assembly, and to an electronic device comprising a three points oscillator assembly.

In many fields, power consumption of electronic 10 equipment is a very important parameter. One of those fields is for instance Radio Frequency Identification (RFID}. Most of those applications require that power consumption is as low as possible in order to increase battery lifetime or to reduce the requirement on the extracted energy from the 15 incident Radio Frequency (RF) field for powering the electronic circuits of the RFID tag.

Many RFID systems have their RF operating frequency below 135 kHz, see for example Klaus Finkenzeller,"RFID Handbook, Fundamentals and Applications in Contactless Smart 20 Cards and Identification", 2nd edition, Wiley 2004, ISBN 0-470-84402-7.

In most active RFID tags, a (reference-) oscillator is required. Because this oscillator is on most of the time, i power consumption of this circuit is of the utmost 25 importance.

A widely used oscillator configuration is the three points oscillator. It comprises an assembly of an oscillator circuit and a resonator in a coupled state. The oscillator circuit comprises a transconductance amplifier having a 30 first and second terminal, wherein the transconductance amplifier has a predetermined inverting transconductance transfer.’ It further comprises a first capacitor connected in between ground and the first terminal of the 2 transconductance amplifier, and a second capacitor connected in between ground and the second terminal of the transconductance amplifier. An example of a transconductance amplifier is a voltage controlled current source. The 5 transconductance amplifier is arranged to generate negative impedances. If the resistive part of this impedance is larger than the resistive part of the resonator impedance, the assembly will start to oscillate at an oscillation frequency within a predetermined oscillation bandwidth. It 10 should be noted that the resonator is normally inductive at the oscillation frequency and that the exact oscillation frequency depends not only on the resistive parts and the resonance frequency of the resonator. For instance, the reactive parts of the negative impedance can cause a shift 15 in oscillation frequency compared to the resonance frequency of the resonator.

Usually the transconductance amplifier is a single active device, e.g. a bipolar transistor or a Metal-Oxide Semiconductor (MOS) transistor. Depending on the choice of 20 the signal ground, such an oscillator is called a Colpitts oscillator or a Pierce oscillator.

An attractive property of these configurations is the i fact that these oscillators are inherently stable for ! frequencies approaching DC as long as the impedance between i 25 input and output is inductive and/or resistive for low frequencies up to DC, which makes them especially attractive to use in combination with (Quartz-) crystal-resonators.

This is contrary to oscillators whose active core consists of a cross coupled differential pair. For those oscillators, 30 special arrangements must be made to guarantee stability at low frequencies when resonators are used that cannot conduct DC-currents like crystal resonators and electrostatically : driven Micro-Electro-Mechanical System (MEMS)-resonators.

3

Usually, a single transistor is used as the active core of the oscillator. For low power applications, the common-source (or common-emitter) configuration is used very often. This is called the Pierce oscillator. In the paper by Eric 5 A. Vittoz et al.,"High-Performance Crystal Oscillator

Circuits: Theory and Application", IEEE Journal of Solid-State Circuits, vol. 23, No 3, June 1988, pp. 774 - 783, a design strategy is outlined to obtain low power operation.

An example of the implementation of an oscillator where 10 the inverting voltage to current transfer is implemented by means of a single transistor is shown in Figure 1, wherein only the equivalent signal circuit is shown.

In Figure 1, the series resonant branch comprising Ls, Cs, and Rs, constitutes an inductive behavior together with 15 C3 at the oscillation frequency. The resonant branch plus (a part of) C3 could for instance represent the equivalent electrical circuit of a MEMS-resonator or a Quartz crystal. Capacitances Cl and C2 form the capacitive load of the resonator.

20 Next, the working principle of the oscillator in figure 1, will be described under reference to figures 2A and 2B.

For a single transistor oscillator, the capacitances of the transistor can be thought of to be part of the capacitances of the external passive components and the 25 resonator (Cl, C2, and C3 in figure 1). This implies that (at least to first order), the transconductance of the transistor can be considered to be frequency independent, ignoring the phase shift due to the presence of the base resistance of a bipolar transistor or the Non-Quasi-Static 30 (NQS) effect of MOS devices.

In the above mentioned paper by Vittoz, it is explained that it is most convenient to describe the impedance what 4 the motional-series Ls, Rs, Cs -branch of the resonator observes. This impedance is called Zc, see figure 2B.

Impedance Zc should exhibit a negative real part and it exhibits a reactive part. The magnitude of the negative real 5 part must be larger than the series loss resistance of the motional-part of the resonator in order to fulfill the oscillation conditions. While a non-zero reactive part will slightly shift the oscillation frequency from the series resonance frequency. The equations of the negative real part 10 and the reactive part for the single transistor oscillator are given in the above mentioned paper by Vittoz.

If the frequency of all the frequency dependent components is set to the oscillation frequency ω0, then (in the complex plane) Zc is a bilinear function of the 15 transconductance gm of the transistor. For the single transistor oscillator, the transconductance of the transistor can be considered to be frequency independent.

The maximum loss resistance RioaE which can just be compensated is: 20 1

EQl ^lass < 7 _ 1 V

2®oQ

V. lic2 y

This expression can be rewritten to a requirement for 25 oscillation. This requirement is: EQ2 V / *

In practice, the value of C1C2/(C1+C2) is much larger 30 than C3. As a result, the requirement for oscillation can be approximated to: EQ3 mCR < — α,0υ3Λ/ο» 2 5

Unfortunately, capacitively excitated MEMS-resonators do not always meet this requirement, or they have a value only slightly smaller than 0.5. If the oscillation 5 requirement is not fulfilled, a single transistor oscillator is not able to oscillate.

For the single transistor oscillator, a one to one relation exists between the bias current Ibias of the transistor and the transconductance. For the MOS transistor, 10 the ratio gm/Ibias is maximum when the MOS device operates in weak inversion. Hence, for a given resonator, the relation between gm-Crit» which is the minimum required gm in order to maintain oscillation, and the power consumption is fully defined, see for instance the above referred paper by Vittoz 15 et al. Reduction of power consumption is therefore limited.

Another drawback of the known oscillator, and in particular when a MEMS resonator is used, is that the oscillation frequency is not constant with temperature. In general, environmental factors such as the ambient 20 temperature will change the mechanical characteristics and therefore the oscillation frequency.

An object of the present invention is to modify the existing three points oscillator to improve the ' applicability for MEMS resonators.

25 A further object of the present invention is to reduce the power consumption of the oscillator.

A still further object of the present invention is to stabilize the oscillation frequency.

At least one of these objects is achieved with a three 30 points oscillator assembly according to the present invention which is characterized in that the inverting i transconductance transfer of the transconductance amplifier is a low pass transfer function with a dominant pole, the 6 influence of said dominant pole being significant at said oscillation frequency as to allow for a higher achievable negative resistance of said transconductance amplifier.

If the inverting transconductance transfer of the 5 transconductance amplifier, which preferably is a multistage amplifier, is a low pass transfer function with a dominant pole, it can be described by G(co)=- ~G°—

EQ4 v ' 1+jmG

10 where ω is the frequency, iG the time constant corresponding to the dominant pole, and Go the low frequency transconductance limit. The equations of the negative real part and the reactive part of Zc, when the frequency 15 dependent transconductance is used, can be derived in a similar manner as in the paper by Vittoz. So, Zc can still be described by a real and an imaginary part, which now become dependent on Go and tg for a given ω.

If the oscillation frequency is denoted by ωο (rad/s), Zc 20 can be split in a resistance Rc and a reactance Xc:

Zc (0¾ ) = (ffl0 ) + j%c (0¾ ) EQ5 , ..

; =/(<?..«·<,)

It turns out that, when the frequency is set to the 25 oscillation frequency, Zc is a bilinear function f of Go with tG as a parameter. So for a certain tg, Zc represents a circle in the complex plain, when Go is varied from -« to +00.

The radius and the centre of this circle are described by the following mathematical representations: i 30 7 ΛΛα.-,

EQ 6 0 \ CXC2 3J

5 ( c +c 1 1+2.^1ΙΗς

Center =-γ- ω°Τβ->- - j--,-—-r

( r +c ^ ( C +C

2a>0C3 2a>0C3

V J V L1L2 J

10 Besides the. radius of the circle, the presence of a .dominant pole (1/tg) in the transconductance only affects the real part of the location of the center of the circle, see figure 3.

The maximum loss resistance Ri0Ss which can just be 15 compensated can be found from: n . φ + αΪ4+β>ότσ EQ7 < f r ,r \ 2(dqC3 1+^C3 0 3 3

Thus, by replacing the single transistor with the 20 transconductance amplifier with a dominant pole, the allowable loss resistance range which can be undamped can be substantially increased compared with a frequency independent transconductance. So, the introduction of a frequency dependent transfer is especially beneficial for 25 oscillators with a MEMS resonator. The requirement for oscillation, when tg is not 0, becomes: ^ / * 30 So, the presence of the dominant pole in the transconductance "widens" the requirement for oscillation: a larger value of Ri0Ss can be accepted in order to fulfil the 8 oscillation conditions. The required G0 in order to realise the maximum negative resistance, which is denoted as Gopt in accordance with the notation as used by Vittoz in his paper, is 5 _ , cc x EQ9 ^opt ~-\/l+öJo,<yo V C3

So, by introducing a low pass transfer in the transconductor, the maximum loss resistance, which can be 10 compensated, compared with a frequency independent transconductor, is increased by a factor equal to the square root of 1+(ü)oTg)2 plus ωοίε at the expense of an increase in the required low frequency transconductance by a factor equal to the square root of 1+(ωοτε)2 compared with a 15 frequency independent transconductance.

The minimum required transconductance Gcrit necessary to exactly compensate a loss resistance (which, of course, must be less than the allowable maximum loss resistance Rioss) is also affected by the presence of the frequency dependent 20 transfer of the transconductor. It turns out that the 'relation between Gcrit and Rioss, with tg as a parameter is: E01° j t-3 ^ ZKloss ^ y V ^‘oss'-'eq ^ 25 with Ceq denoted by ( c +C λ EQ11 ce9=c3. i + ^-^c3 . Due to the presence of tG the critical conductance is 30 increased.

An example of Zc as a function of G0 for a certain value of tG, showing the location of Gcrit and Gopt, is shown in Figure 4.

9

As already derived analytically, it can also be observed that the maximum negative resistance increases with increasing tg.

It is advantageous if the time constant is adjustable.

5 As shown in figure 3, by adjusting the time constant rG, the reactive part of Zc can be changed, thereby shifting the oscillation frequency. This property can be useful for controlling the oscillation frequency.

Additionally or alternatively, it is advantageous if the 10 low frequency limit of the transconductance Go is adjustable, for instance by varying the bias to the active components of the transconductance amplifier.

Several interesting useful combinations of Go and tg can be chosen.

15 As a first example, it is noted that a minimum sensitivity of the oscillation frequency occurs for variations in Go. If the real part of the centre of the Ζς-circle is set to -Rioss, the imaginary part of Zc hardly varies for (small) variations in Go because GCrit coincides 20 with the top of the circle and small deviations in Go hardly affect the imaginary part of Zc. According to equation 6, the real part is only affected by tg and not by Go as far as the parameters of the transconductance are concerned.

Another possibility of the new oscillator is to realise 25 an impedance Zc with a zero reactive part. The following two equations with the two unknowns Go and tg must be solved in order to realise this: EQ12 RefcW}= 3Q lm{zc(i»0)}=0

An advantage is that the oscillation frequency equals the resonance frequency of the (mechanical) series resonant branch of the resonator.

10 A temperature dependent resonance frequency of the resonator can be compensated for if the assembly comprises a temperature sensor for measuring a temperature of the resonator, and a first- controller for controlling the time 5 constant of the dominant pole in response to the measured temperature. This is especially useful for MEMS-resonators. For example silicon based MEMS structures exhibit a very large temperature dependence of 30 ppm/°C.

Additionally or alternatively, the assembly can comprise 10 a frequency.detector for measuring the oscillation _ frequency, and a second controller for controlling the time constant of the dominant pole in response to the measured frequency. In this way, a feedback system is realized with which the oscillation frequency can be maintained and/or 15 stabilized at a predetermined level, which could be adjustable.

The inverting transimpedance amplifier comprises active circuitry that is preferably at least partly based on Metal-Oxide Semiconductor Field-Effect Transistors (MOSFETs), 20 arranged to operate in weak inversion. In weak inversion, the ratio between transconductance and bias current is maximum, making weak inversion very suitable for low power operation.

The transconductance amplifier can be arranged for 25 differential operation. In that case, baluns are used to convert between the single ended output of the resonator and the differential input of the transconductance amplifier. To this end, the differential terminals of the baluns are connected to the respective differential terminals of the 30 transconductance amplifier.

If purely differential signals are applied to the transconductance amplifier, i.e. no common mode signal, virtual grounds exist within the transconductance amplifier.

11

These virtual grounds can be connected to the actual ground. This presents the advantage that two single-ended topologies can be used and no conunom-mode instability is likely to occur.

5 In a preferred embodiment, the transconductance amplifier comprises a cascode combination of two n-channel MOSFETs, wherein a first of said two n-channel MOSFETs has its gate connected to a bias voltage source, its drain connected to a current source. The drain is the current 10 output of the transconductance amplifier. The second of the two n-channel MOSFETs has its drain connected to the source of the first n-channel MOSFET. This drain is also connected to ground via a capacitor. The source of the second n-channel MOSFET is connected to ground and the gate of the 15 second n-channel MOSFET is the voltage input of the transimpedance amplifier. This configuration enables the time constant to be set by the capacitor in between ground and the drain of the second n-channel MOSFET.

By using a plurality of cascaded voltage-current and 20 current-voltage conversion stages, preferable three, instead öf a single transistor, the total power consumption can be reduced. It should be noted that current-voltage conversion does not exclude any conversion from voltage to current.

Schematic examples of multi-stage transconductance 25 amplifiers are given in figures 6, 7A and 7B. These figures illustrate multi-stage variants of the single stage general circuit depicted in figure 5.

In figure 5, an inverting voltage-to-current conversion block represents an amplifying core consisting of one or 30 more active devices. In figures 6, 7A and 7b, the overall inverting voltage-to-current transfer can be constructed from a combination of voltage-to-current transfers and current-to-voltage transfers. The only boundary condition of 12 the combination of the transfers is that the input quantity of the overall transfer must be a voltage, the output quantity a current and that the overall transfer must be an inverting transfer.

5 The goal is to use less bias current for the overall voltage-to-current transfer when compared with the single transistor solution for realising the same transconductance.

An example of one of the possible transfer combinations is: 10 I.u, L· EQ 13 v„ u„ ix u, A high level implementation example is shown in figure 6. This can be simplified to a configuration with only two 15 voltage-to-current conversion blocks, by replacing the current-to-voltage conversion block in figure 6 with a single resistive element, such as a resistor. This is shown in figures 7A and 7B.

The resistive element can also be realized by an active 20 resistor, wherein the active resistor comprises a second transconductance amplifier having a voltage input and a current output, a first resistor connected in between the voltage input and ground, and a second resistor connected in between the voltage input and current output.

25 In a preferred embodiment, the first stage of the three stage transconductance amplifier comprises an n-channel MOSFET having its source connected to ground and its gate 1 being the first terminal of the transconductance amplifier, and its drain being a current output of the first stage. The 30 second stage comprises a MOSFET current source arranged for biasing the n-channel MOSFET via the drain terminal thereof, wherein the output resistance of the MOSFET current source is used for converting the current at the output of the 13 first stage into a voltage at the drain of the n-channel MOSFET of the first stage. The third stage comprises a noninverting transconductor comprising a combination of a p-channel MOSFET and a n-channel MOSFET, wherein the drain of 5 the n-channel MOSFET is connected to a supply, its gate being coupled to the drain of the first stage, and its source being coupled to the source of the p-channel MOSFET, the drain of the p-channel MOSFET being the second terminal of the transconductance amplifier and being coupled to a 10 current source for biasing the p-channel MOSFET and the n-channel MOSFET of the third stage, and the gate of the p-channel MOSFET being connected to a bias voltage source.

In another preferred embodiment, the first stage of the three stage transconductance amplifier comprises a non-15 inverting transconductor comprising a combination of a p-channel MOSFET and a n-channel MOSFET, wherein the drain of the n-channel MOSFET is connected to a supply, its gate being the first terminal of the transconductance amplifier, and its source being coupled to the source of the p-channel 20 MOSFET, the drain of the p-channel MOSFET being a current Output of the first stage, and the gate of the p-channel MOSFET being connected to a voltage bias source. The second stage comprises a MOSFET current source arranged for biasing the combination of the p-channel MOSFET and the n-channel 25 MOSFET, wherein the output resistance of a MOSFET is used for converting the current at the output of the first stage into a voltage. The third stage comprises an n-channel MOSFET having its source connected to ground and its gate to ; the drain of the p-channel MOSFET of the first stage, and 30 its drain being connected to a current source for biasing. The drain of the n-channel MOSFET of the third stage is also i the second terminal of the transconductance amplifier.

14

The three points oscillator assembly according to the present invention is particularly suited for MEMS oscillators due to the high negative resistance that can be achieved. In such an oscillator, the resonator comprises a 5 MEMS resonating device.

The present invention also provides an electronic device comprising the three points oscillator assembly in coupled state as described above. It further provides for an oscillator circuit, connectable to a resonator, for forming 10 the three points oscillator assembly as defined above.

Next, the present invention will be described in more detail under reference to the accompanying drawings in which:

Figure 1 illustrates a prior art three points 15 oscillator;

Figures 2A and 2B show a general schematic of a three points oscillator;

Figure 3 illustrates Zc as a function of Gq and according to the present invention; 20 Figure 4 illustrates Zc as a function of Go according to the present invention;

Figure 5 shows a general schematic of a three points j oscillator; ! Figure 6 illustrates a general schematic of a muli- 25 stage three points oscillator according to the present invention;

Figures 7A and 7B illustrate reduced schematics of the multi-stage three points oscillator in figure 6 according to the present invention; 30 Figure B shows a first embodiment of the ; transconductance amplifier according to the present I invention; i.

15

Figure 9 shows a second embodiment of the transconductance amplifier according to the present invention;

Figure 10 shows a third embodiment of the 5 transconductance amplifier according to the present invention;

Figure 11 illustrates an active resistor used for voltage-current conversion according to the present invention; 10 Figure.12 shows a fully-differential transconductance amplifier according to the present invention;

Figure 13 shows a semi-differential transconductance amplifier according to the present invention;

Figure 14 illustrates a temperature stabilized three 15 points oscillator according to the present invention;

Figure 15 illustrates a frequency stabilized three points oscillator according to the present invention;

As an example some embodiments of the present invention will be shown for a low frequency reference oscillator (32 20 kHz for instance). Such a frequency is commonly used as the reference frequency for electronic watches. But the example circuits can be used for other oscillation frequencies as well.

In figure 8, it is assumed that the overall voltage-to-25 current transfer is realised by means of a combination of an inverting voltage-to-current, a current-to-voltage and a voltage-to-current transfer in a CMOS process. The goal is to use as little bias current as possible.

The (input-) voltage-to-current transfer can be realised 30 by a single common-source MOS transistor, while the current-to-voltage transfer can be realised by means of a resistor. The resistor can be realised by a MOS transistor. Both the 16 input transistor and the MOS-resistor use the same bias-current.

Because ultra low power consumption is required, a very high resistor-value is required (e.g. > 10 ΜΩ). It turns out 5 that a MOS transistor, operating in weak inversion and in saturation can be used to realise this.

The intrinsic voltage amplification factor of a MOS device, operating in weak inversion is approximately independent of its bias current, see for Jeroen Kuenen, "A 10 Design Strategy For Low-Power Low-Voltage Integrated

Transconductance Amplifiers", Ph.D. thesis, Delft University Press, 1997, ISBN 90-407-14320. The output resistance is denoted by:

, ~rM

15 EQ14 / in which Veq(L) is a process and transistor-length (L) dependent "Equivalent Voltage"-constant, which is independent of the bias current. On the other hand, the 20 transconductance of the common-source transistor, operating in weak inversion is described by: ; EQ15 25 The low-frequency voltage-to-voltage transfer function then becomes: 'rji)

¢4, nVr ID

30 EQ16 , v

-yJA

nVT

where Ux is the voltage at the drain of the input-transistor.

17

Sor the overall voltage-to-voltage transfer of a common-source transistor, loaded with a current-source load and both operating in weak inversion, is to first order independent of the bias current. By choosing the proper 5 length of (for instance) the current-source transistor, the voltage gain can approximately be set to the desired value (the smaller the length, the lower the output resistance).

The second stage of the amplifier consists of a noninverting transconductance transfer, made by a n-channel 10 MOSFET p-channel MOSFET combination. The transconductance of this combination is approximately equal to gm/2 assuming that the two transistors have equal transconductance gm. The low-frequency overall transfer of the total transconductor circuit (Go) becomes: 15 Q ~ ~Ip\ Ip2

0 nVT 2 nVT

EC17 -rjfi.

~2{„rTJ m 20 As mentioned before, the low frequency transfer is independent of the bias current of the first transistor.

The feedback resistor Rfb takes care of proper biasing of the circuit. Its value must be large enough to avoid loading the resonator. The value of Rfb must fulfil the same 25 requirements as for the single transistor oscillator, see the paper by Vittoz.

The total bias current of the amplifier is the sum of the bias currents of the first stage and the second stage: : 30 EQ18 Im -^di + Idi

For this transconductance example, the bandwidth is mainly set by the internal node at the output of the first 18 stage where the amplified voltage of the input signal is present. Due to the presence of the very high load resistance, the bandwidth can be rather small (less than a few hundred kHz). The total capacitance as present at the 5 drain of the first transistor is virtually constant.· This implies that the bandwidth of the voltage gain of the first stage can be controlled by means of the bias current of this first stage. Consequently, because the voltage-to-current transfer of the second stage can be considered to be 10 frequency independent, the bandwidth of the overall voltage-to-current transfer can be controlled by the bias current of the first stage.

The frequency dependent overall transfer of the two stage transconductor can be described by: 15 β(φ2(,pry' EQ19 —

V J

= -Oq

\+jmG

20 where Ctot represents the total capacitance across the equivalent load resistor (Veq/IDi) at the output of the first stage.

Summarizing, the voltage gain of the first stage is set 25 by the ratio of the finite output resistance of current source II and the transconductance of the input transistor Ml, while the bandwidth of the voltage-to-current transfer can be controlled by the bias current II.

For a given resonator, a certain G^-it is required. As 30 long as Itotai is less than Gcrit*nVT, the overall amplifier is ; more power efficient than the single transistor solution.

Figure 9 presents another embodiment of a transconductance amplifier according to the present 19 invention. In this circuit, the non-inverting voltage gain of the first stage is set by the ratio of the transconductance of the transistor-combination Ml and M2 and the finite output resistance of the current source II. The 5 voltage at the drain of M2 is converted to a current by means of transistor M3. The feedback resistor Rfb takes care of proper biasing of the circuit. For those skilled in the art, other two-stage configurations can simply be constructed as well.

10 Figure 10 presents still another embodiment of a transconductance amplifier according to the present invention. Apart from the realisation of a transconductance with a multi stage amplifier (which consumes less current compared with a single transistor solution), another 15 possible configuration serves to realise a transconductor which does not have the benefit of the reduced bias current, but does exhibit the benefit of introducing a pole (phase shift) in order to create a higher negative resistance.

An example of such an embodiment is a cascode transistor 20 combination with a capacitor connected from the interconnection of the two devices to ground, as shown in figure 10. The low frequency voltage-to-current transfer is set by transistor Ml, while the phase shift is realised by the input resistance of M2 in combination with C. The 25 transfer of this cascode with capacitor is: EQ 20 jj ~ Sui (->

Uin 1 + >— &M2 30 If both transistors operate in weak inversion, the transconductance of transistor Ml is approximately equal to 5mi = Ii/ (nVT) , while the input conductance giM2 of M2 is 20 approximately equal to giM2 = Ii/VT. So, for a given bias current, the time constant can be set by the capacitor C.

Figure 11 shows an embodiment of a voltage to current transfer using an active resistor. Such device can be used 5 as the second stage of a three stage transconductance amplifier for example. In figure 11, two feedback resistors are used around a voltage-to-current transfer, realized by means of a transconductance amplifier. For this example, the output resistance Rout can be found from: 10 EQ21 r 0Hf =^-^- m 1+ GR,

This method can be useful for those cases where the output resistance of a transistor, operating in weak-15 inversion and saturation, is too high.

Figures 12 and 13 present schematic examples of a differential transconductance amplifier. In figure 12, the overall transconductance is fully differential whereas in figure 13, two single-ended versions can be combined into a 20 semi-differential transconductance. It is noted that both examples have the same signal transfer.

The advantage of differential and semi-differential circuits is that common-mode disturbances are much more suppressed compared with its single-ended counterpart.

25 However, the (undesired) conversion of common-mode to differential-mode signals is much less for the fully differential version than the semi-differential version.

Figures 14 and 15 present schematic overviews of two different embodiments that make use of the fact that the 30 oscillation frequency depends on the time constant corresponding to the dominant pole. This dependency can be 1 used to compensate the temperature dependent resonance frequency of the whole oscillator or of the resonator. This 21 is especially useful for MEMS-resonators. To this end, a temperature sensor is used to measure a temperature of the resonator. This temperature is fed to a controller, which controls the time constant in response to the measured 5 temperature, see figure 14.

Of course, instead of compensating for a temperature dependence, a frequency detector can be used to determine the oscillation frequency. This frequency can then be fed to a controller, which in turn controls the time constant. In 10 this way, a feedback system is obtained by which the oscillation frequency can be maintained and or stabilized, as shown in figure 15.

As further examples, other possible extensions are noted to be evident to the skilled person after having been 15 confronted with the above disclosure. Apart from the possibility to realise different transconductor topologies, the oscillator core can be extended by an Automatic Gain Control (AGC) loop in order to minimise the supply current under steady state conditions. The purpose of the AGC-loop 20 is to realise a large negative resistance at start-up. For instance it is possible to set it to -Rmav by configuring the transconductance to be Gopt at start-up. Then, steadily, the bias current of (part of) the oscillator core is to be reduced, once the oscillator voltage is increasing to its 25 final value. Once the desired (voltage) signal swing is reached, the overall (average) transconductance is reduced to GCrit (which depends on the value of tg as well) . In the paper by Vittoz, an example is shown of an implementation of an AGC circuit.

30 Another extension is to add a buffer circuit that converts the oscillator signal to a rail to rail voltage swing, rëquired to drive (standard) logic circuits. In the 22 paper by Vittoz, an example is shown of a possible implementation as well.

Further also other topologies are possible within the framework of the present invention as defined in the 5 appended claims, without departing from the scope of the literal definitions of the claims or the spirit thereof.

For example, other transistor technologies, e.g. bipolar transistor technology, can be used than the MOSFETs discussed above. In addition, the circuits according to the 10 present invention can also be implemented using PMOS

technology. It should be obvious to the skilled person that a combination of these various technologies is not excluded.

Claims

A three-point oscillator assembly, comprising: a resonator, such as a Micro-Electric Mechanical System (MEMS) resonating device, which is provided with electrical resonance characteristics between a first and second clamp thereof; an oscillator circuit comprising: - a transconductance amplifier which is provided with a first and a second terminal, the transconductance amplifier having a predetermined inverting transconductance transfer; a first capacitive element, such as a capacitance, connected between ground and the first terminal of the transconductance amplifier; - a second capacitive element, such as a capacitance, connected between ground and the second terminal of the transconductance amplifier; wherein, in a coupled state, the first and second terminals of the transconductance amplifier are coupled to the first and second terminals of the resonator respectively, and wherein the transconductance amplifier is adapted to generate a negative impedance for causing an electronic oscillation of the composition at an oscillation frequency within a predetermined oscillation bandwidth; characterized in that the transconductance transfer from the transconductance amplifier is an inverting low pass transfer function with a dominant pole, wherein the influence of said dominant pole is considerable on said oscillation frequency to allow a higher attainable negative resistance of said transconductance amplifier.

The three-point oscillator assembly according to claim 1, wherein the transconductance amplifier is a multi-stage amplifier.

The three-point oscillator assembly according to claim 1 or 2, wherein the time constant corresponding to the dominant pole is adjustable.

4. Three-point oscillator assembly according to claim 3, further comprising a temperature sensor for measuring a temperature of the resonator, and a first controller for checking the time constant of the dominant pole in response to said measured temperature to compensate for a temperature dependent resonance frequency of said resonator.

5. Three-point oscillator assembly according to claim 3 or 4, further comprising a frequency detector for measuring the oscillation frequency, and a second controller for checking the time constant of the dominant pole in response to said measured frequency for maintaining a predetermined frequency determined oscillation frequency.

6. Three-point oscillator assembly according to any of the preceding claims, wherein the transconductance amplifier comprises an active circuit based at least in part on Metal-Oxide Semiconductor Field-Effect Transistors (MOSFETs) arranged to operate in weak inversion.

7. Three-point oscillator assembly according to any of the preceding claims, wherein the transconductance amplifier is adapted for differential operation, the assembly further comprising two baluns connected to the first and second terminals of the resonator respectively, with differential output of the baluns connected to the respective first and second differential terminal of the transconductance amplifier.

8. Three-point oscillator assembly according to any of the preceding claims, wherein the transconductance amplifier comprises a cascode combination of two n-channel MOSFETs, wherein a first of said two n-channel MOSFETs has its gate connected to a voltage source, its drain connected to a drain current source, wherein said drain is the current output of the transconductance amplifier, and wherein a second of said two n-channel MOSFETs has its drain connected to the source of the first n-channel MOSFET, the drain also being connected to ground via a capacitance, and wherein a source of said second n-channel MOSFET is connected to ground, and wherein a gate of said second n-channel MOSFET is the voltage input of said transconductance amplifier.

9. Three-point oscillator assembly according to any of the preceding claims, wherein the transconductance amplifier comprises three cascaded voltage-to-current conversion stages.

10. The three-point oscillator assembly according to claim 9, wherein at least one of the voltage-to-current conversion stages comprises a resistive element for obtaining the current-to-voltage conversion.

11. A three-point oscillator assembly according to claim 9 or 10, wherein said resistive element comprises an active resistor, said active resistor comprising: a second transconductance amplifier which is provided with a voltage input and a current output; a first resistor connected between the voltage input and ground; a second resistor connected between the voltage input and the current output.

12. Three-point oscillator assembly according to any of claims 9-11, wherein a first stage of the transconductance amplifier comprises an n-channel MOSFET which has its source connected to ground and wherein its gate is the first terminal of the transconductance amplifier, and its drain a current output is from the first stage; wherein a second stage of the transconductance amplifier comprises a MOSFET current source adapted to set the n-channel MOSFET via its drain terminal, the output resistance of the MOSFET current source being used to convert the current at the output of the first stage in a voltage at the drain of the n-channel MOSFET of the first stage; wherein a third stage of the transconductance amplifier comprises a non-inverting transconductor which comprises a combination of a p-channel MOSFET and an n-channel MOSFET, wherein the drain of the 20 n-channel MOSFET is connected to a power supply, and wherein its gate is coupled to the drain of the first stage, and wherein its source is coupled to the source of the p-channel MOSFET, wherein the drain of the p-channel MOSFET is the second terminal of the transconductance amplifier and which is coupled to a power source for setting said p-channel MOSFET and said n-channel MOSFET of the third stage, and wherein the gate of the p-channel MOSFET is set with a predetermined voltage.

A three-point oscillator assembly according to any of claims 9-12, wherein a first stage of the transconductance amplifier comprises a non-inverting transconductor which comprises a combination of a p-channel MOSFET and n-channel MOSFET, the drain of the n-channel MOSFET is connected to a power supply, where its gate is the first terminal of the transconductance amplifier, and where its source is coupled to the source of the p-channel MOSFET, the drain of the p-channel MOSFET being a current output of the first stage, and wherein the gate of the p-channel MOSFET is connected to a bias voltage source; wherein a second stage of the transconductance amplifier comprises a MOSFET current source adapted to set the combination of the p-channel MOSFET and the n-channel MOSFET, wherein the output resistance of the MOSFET is used to convert the current at the output of the first stage in a tension; 15 wherein a third stage of the transconductance amplifier comprises an n-channel MOSFET which has its source connected to ground and its gate connected to the drain of the p-channel MOSFET of the first stage, and wherein its drain is connected to a power source 20 for setting, wherein the third stage n-channel MOSFET drain is the second terminal of the transconductance amplifier.

14. Electronic device comprising the three-point oscillator assembly in coupled condition as defined in any of claims 1-13.

An oscillator circuit, connectable to a resonator, for forming a three-point oscillator assembly as defined in any one of claims 1-13.