Advanced Microwave Circuits and Systems Part 13 pptx

Frequency depending attenuation Figure 9 shows the voltage which can be induced in the transponder in comparison to a fer via air.. Example of Energy Transmission in a Sensor Transponder

absorption which is transformed into heat is calculated for different frequencies In order to

classify the impact of energy transfer to the transponder, the available energy is set in relation

to transfer in air in a FDTD simulation

3.2.1 Simplified Body Model

If a transponder is directly attached to the heart and the reading device on the chest of a

patient, there is more than one type of tissue between the antennas Table 1 shows a list of

tissues and their conductance values In order to calculate the losses, the volume of each

tissue, their conductance values and the distribution of current density induced have to be

taken into consideration However, here the aim is to have a simplified estimation, this is why

a homogeneous model of the body is used in the following It only contains one conductance

value and consists of a simple geometry For a “worst case scenario” conductance values

of blood can be used because it has the highest conductivity in comparison to other types

of tissues For a second calculation an average conductance value is used For the analytic

calculation of induced eddy currents in bodies, a cylinder is especially useful because of its

simple geometric form Thus, a cylinder is assumed which has a field-creating coil at its front

surface The length is chosen so that the transponder is included and the distance to the

field-creating coil is approximately 50cm The diameter can be chosen accordingly Figure 7 shows

the model

Fig 7 Modell for estimation of power absoption

A length of l = 80 cm and a radiant of a = 15 cm were choosen

3.2.2 Estimation of Power Loss

As described at the beginning, an alternating magnetic field, which occurs within a conductive

medium, induces eddy currents These ultimately lead to heating the medium The heat

capacity which is transformed in a cylinder - figure 7 - needs to be estimated The heat capacity

can be assessed as follows (3):

- H is the magnetic field strength on the axis of the cylinder, µ is the permittivity, σ the

con-ductance value, x the length of the cylinder and K a correction factor The correction factor Kdescribes the inhomogeneous distribution of current density an depends on the radius of thecylinder and penetration of the skin The magnetic field within the cylinder is not homoge-neous Furthermore, the current density of eddy currents over the radius is not homogeneousdue to energy-displacement effects

The magnetic field strength H decreases with increasing distance x to the field-creating coil.This can be described with the Biot-Savart law

In comparison, the electronik of the transponder consumes 90µW The volume of the cylinder

in which the heat capacity is transformed is about 56,5dm2 The resulting heat capacity perunit volume is about 80nW This value is quite safe for medical accounting purposes In order

to analyse the influence to the transponder-system, it is necessary to see the energy, whichthe transponder has at its disposal, in relation to a transfer via air This is how the impact ofabsorption of the human body is visible

Deeply Implantable Medical Sensor Transponders 415

absorption which is transformed into heat is calculated for different frequencies In order to

classify the impact of energy transfer to the transponder, the available energy is set in relation

to transfer in air in a FDTD simulation

3.2.1 Simplified Body Model

If a transponder is directly attached to the heart and the reading device on the chest of a

patient, there is more than one type of tissue between the antennas Table 1 shows a list of

tissues and their conductance values In order to calculate the losses, the volume of each

tissue, their conductance values and the distribution of current density induced have to be

taken into consideration However, here the aim is to have a simplified estimation, this is why

a homogeneous model of the body is used in the following It only contains one conductance

value and consists of a simple geometry For a “worst case scenario” conductance values

of blood can be used because it has the highest conductivity in comparison to other types

of tissues For a second calculation an average conductance value is used For the analytic

calculation of induced eddy currents in bodies, a cylinder is especially useful because of its

simple geometric form Thus, a cylinder is assumed which has a field-creating coil at its front

surface The length is chosen so that the transponder is included and the distance to the

field-creating coil is approximately 50cm The diameter can be chosen accordingly Figure 7 shows

the model

Fig 7 Modell for estimation of power absoption

A length of l = 80 cm and a radiant of a = 15 cm were choosen

3.2.2 Estimation of Power Loss

As described at the beginning, an alternating magnetic field, which occurs within a conductive

medium, induces eddy currents These ultimately lead to heating the medium The heat

capacity which is transformed in a cylinder - figure 7 - needs to be estimated The heat capacity

can be assessed as follows (3):

- H is the magnetic field strength on the axis of the cylinder, µ is the permittivity, σ the

con-ductance value, x the length of the cylinder and K a correction factor The correction factor Kdescribes the inhomogeneous distribution of current density an depends on the radius of thecylinder and penetration of the skin The magnetic field within the cylinder is not homoge-neous Furthermore, the current density of eddy currents over the radius is not homogeneousdue to energy-displacement effects

The magnetic field strength H decreases with increasing distance x to the field-creating coil.This can be described with the Biot-Savart law

In comparison, the electronik of the transponder consumes 90µW The volume of the cylinder

in which the heat capacity is transformed is about 56,5dm2 The resulting heat capacity perunit volume is about 80nW This value is quite safe for medical accounting purposes In order

to analyse the influence to the transponder-system, it is necessary to see the energy, whichthe transponder has at its disposal, in relation to a transfer via air This is how the impact ofabsorption of the human body is visible

Table 1 Conductivities in S/m of different tissues at different frequencies (2)

Fig 8 Approximation der Volumen-Information (MRT-picture: Deutsches R¨ontgen-Museum)

A cross section of the human body at the level of the heart can be seen In order to create a

field and measure the strength close to the transponder, two types of antenna were used With

this simulation absorption and frequency behaviour can be analysed quickly

In order to assess the absorption strength of the human body, it is necessary to eliminate

factors from the antenna which might have an impact For this matter, a type of reference

simulation was carried out In this simulation the human body was replaced with air The

measured voltage values at the transponder antenna will then be offset with the results of

following simulations

Simulations with a further model were used to assess absorption effects realistically In the

following this model is referred to as “homogeneous model” It reflects a worst-case scenario

For this purpose dielectric parameters of blood were used for all tissues since blood has a

higher conductivity than other tissues

In order to extract information about frequency-dependent absorption from the results of thethree simulations, quotients were made from conductance values of the homogeneous modeland the inhomogeneous model based on the reference model

Fig 9 Frequency depending attenuation

Figure 9 shows the voltage which can be induced in the transponder in comparison to a fer via air If there is only air in the transfer system, the quotient is one for all consideredfrequencies First, it is clearly visible that the absorption capacity generally increases withhigher frequencies and thus induced voltage decreases For a frequency of 40 MHz the voltagedecreases to 24 and 64 per cent respectively in the homogeneous model In a low-frequencyarea, on the other hand, absorption is hardly detectable However, which frequency is best fortransferring a maximum of energy does not only depend on the absorption but also on furthercharacteristics of the transmission According to the induction-law, for instance, the inducedvoltage stands in proportion to frequency Thus, it can be expected that there is a frequency atwhich the induced voltage is at its maximum Furthermore, the characteristics of the antennasused have to be analysed The following chapters deal with this topic In chapter 4.2 the idealfrequency will be established with regard to all findings

trans-4 Example of Energy Transmission in a Sensor Transponder System

4.1 Frequency Behaviour of Induced Voltage at the Transponder

The voltage induced in the transponder coil is used to provide the power supply to thetransponder electronic To improve the efficiently, an parallel resonant circuit is formed by

an additional capacitor connected in parallel with the transponder coil Figure 10 shows theequivalent circuit of the transponder

The resistor Ri represents the natural resistance of the transponder coil L1 and the currentconsumption of the transponder electronic is represented by the load resistor RL If a voltage

Ui is induced in the coil L1, the voltage Ul can be measured at the load resistor RL It is a result

of the voltage Ui minus the current i multiplied with the coil impedance and Ri The so calledquality factor represents the relationship between the induced voltage at L1 and the voltage

at the transponder electronic A higher quality factor causes a higher voltage ul and a higher

Table 1 Conductivities in S/m of different tissues at different frequencies (2)

Fig 8 Approximation der Volumen-Information (MRT-picture: Deutsches R¨ontgen-Museum)

A cross section of the human body at the level of the heart can be seen In order to create a

field and measure the strength close to the transponder, two types of antenna were used With

this simulation absorption and frequency behaviour can be analysed quickly

In order to assess the absorption strength of the human body, it is necessary to eliminate

factors from the antenna which might have an impact For this matter, a type of reference

simulation was carried out In this simulation the human body was replaced with air The

measured voltage values at the transponder antenna will then be offset with the results of

following simulations

Simulations with a further model were used to assess absorption effects realistically In the

following this model is referred to as “homogeneous model” It reflects a worst-case scenario

For this purpose dielectric parameters of blood were used for all tissues since blood has a

higher conductivity than other tissues

In order to extract information about frequency-dependent absorption from the results of thethree simulations, quotients were made from conductance values of the homogeneous modeland the inhomogeneous model based on the reference model

Fig 9 Frequency depending attenuation

Figure 9 shows the voltage which can be induced in the transponder in comparison to a fer via air If there is only air in the transfer system, the quotient is one for all consideredfrequencies First, it is clearly visible that the absorption capacity generally increases withhigher frequencies and thus induced voltage decreases For a frequency of 40 MHz the voltagedecreases to 24 and 64 per cent respectively in the homogeneous model In a low-frequencyarea, on the other hand, absorption is hardly detectable However, which frequency is best fortransferring a maximum of energy does not only depend on the absorption but also on furthercharacteristics of the transmission According to the induction-law, for instance, the inducedvoltage stands in proportion to frequency Thus, it can be expected that there is a frequency atwhich the induced voltage is at its maximum Furthermore, the characteristics of the antennasused have to be analysed The following chapters deal with this topic In chapter 4.2 the idealfrequency will be established with regard to all findings

trans-4 Example of Energy Transmission in a Sensor Transponder System

4.1 Frequency Behaviour of Induced Voltage at the Transponder

The voltage induced in the transponder coil is used to provide the power supply to thetransponder electronic To improve the efficiently, an parallel resonant circuit is formed by

an additional capacitor connected in parallel with the transponder coil Figure 10 shows theequivalent circuit of the transponder

The resistor Ri represents the natural resistance of the transponder coil L1 and the currentconsumption of the transponder electronic is represented by the load resistor RL If a voltage

Ui is induced in the coil L1, the voltage Ul can be measured at the load resistor RL It is a result

of the voltage Ui minus the current i multiplied with the coil impedance and Ri The so calledquality factor represents the relationship between the induced voltage at L1 and the voltage

at the transponder electronic A higher quality factor causes a higher voltage ul and a higher

Fig 10 Equivalent Circuit of a Transponder

maximum possible distance between reader and transponder It can be calculated with the

following formula relating to the equivalent circuit (4):

By analysis of this formula it can be seen, that for every pair of Ri and RL there is a L1 at

which the quality factor is at its maximum And this maximum value of the quality factor

is different for every frequency So if the optimal L1 is calculated for every frequency, the

maximum possible quality factor versus frequency could be calculated

4.2 Optimal Frequency

The induced voltage Ui is reduced by the loss effects described in chapter 4.1 Because Ui is

proportional to the quality factor, it is allowed to multiply the quality factor calculated with 17

together with the results of the graph’s in figure 9 Figure 11 shows the evaluation of equation

?? considering the effects described before.

Fig 11 Influence of the human tissue to the optimum frequency

First of all, a great difference in induceable voltages between LF and HF frequencies can be

seen For low frequencies, the quality factor is much smaller than for the HF case The

simu-lation shows a maximum quality factor for all simusimu-lations between 7 MHz and 9 MHz If thecoils are sourrounded by air, there will be an optimal frequency of about 9 MHz This opti-mal frequency becomes lower, when human tissue is between the coils For the homogeneousmodel, in worst case, an optimal frequency is about 7 MHz In the inhomogeneous model,that is more realistic, the highest quality factor could be optained with 8.4 MHz It can besaid, that the human tissue reduces the optimal frequency value, at which the most voltagecan be induced respectively the highest transmission range could be achieved The optimalfrequency can be observed near to the 6,78 MHz ISM band In comparison to LF ISM Bandthe amount of induced voltage is about 4 times higher In comparison to 13.56 MHz a power

of maximum 20 % less is necessary to get the same transmission range

4.2.1 Practical Measurement

An experimental measurement shall determine the maximum achievable distance For thisexperiment, a circular coil with a single winding and an aperture of 26 cm was used to pro-duce the magnetic field A frequency of 13,56 MHz was chosen A test transponder wasdeveloped to measure the energy that can be provided to an implanted transponder It con-sists of a ferrite rod coil, an HF front-end and a load resistor that simulates the impedance of

a transponder circuit To create a substitute that simulates the electric properties of the man body, a phantom substance was prepared following a recipe described in [2] The maingoal of the experiment is to measure the voltage induced at the transponder coil when it isplaced inside this substance at different distances from the reader coil 50 L of the phantomsubstance is obtained It was placed in a container large enough to allow the transponder to beplaced in a similar position as in a human body Following the specifications in the article, thecontainer should be made of an electrical insulator and non-magnetic material In our case,the container has a cylindrical form, which is sufficiently similar to a human body Anotherrequirement is a minimum volume of substance It is specified that a mass of at least 30 kg

hu-of phantom material is necessary Generally, a homogeneous phantom is accurate enough tosimulate a human body, in this way it is not necessary to incorporate materials of differentconductivity inside the container Figure 12 (a) shows the measurement setup

With this measurement it is possible to determine in how much surrounding tissue a der can work To measure the provided energy for different distances from the reader, thevoltage at the load resistor in the test-transponder was measured versus the distance Thechip used in sensor transponders usually works with voltages greater than 3 V Therefore, thetransponder would be provided with enough energy at a distance where the voltage is stillhigher than this voltage Figure 12 (b) shows the measurement results

transpon-The measurement was done with a voltage amplitude at the reader coil of 300 V and a loadresistor in the test transponder of 60 kOhm and 100 kOhm These values were chosen empiri-cally The diagram shows the voltage that would be available for a chip in different distances.The voltage is grater than 3 V for distances up to 43 cm

The experimental measurement shows, that a sensor transponder can work inside a humantissue up to a distance of 40 cm

5 Conclusion

The influence by the human tissue on the inductive energy transmission was considered forthe design of a sensor transponder system For the given constraints to the transponder an-tenna an optimal frequency could be found The loss effects decrease this optimum frequency

Deeply Implantable Medical Sensor Transponders 419

Fig 10 Equivalent Circuit of a Transponder

maximum possible distance between reader and transponder It can be calculated with the

following formula relating to the equivalent circuit (4):

By analysis of this formula it can be seen, that for every pair of Ri and RL there is a L1 at

which the quality factor is at its maximum And this maximum value of the quality factor

is different for every frequency So if the optimal L1 is calculated for every frequency, the

maximum possible quality factor versus frequency could be calculated

4.2 Optimal Frequency

The induced voltage Ui is reduced by the loss effects described in chapter 4.1 Because Ui is

proportional to the quality factor, it is allowed to multiply the quality factor calculated with 17

together with the results of the graph’s in figure 9 Figure 11 shows the evaluation of equation

?? considering the effects described before.

Fig 11 Influence of the human tissue to the optimum frequency

First of all, a great difference in induceable voltages between LF and HF frequencies can be

seen For low frequencies, the quality factor is much smaller than for the HF case The

simu-lation shows a maximum quality factor for all simusimu-lations between 7 MHz and 9 MHz If thecoils are sourrounded by air, there will be an optimal frequency of about 9 MHz This opti-mal frequency becomes lower, when human tissue is between the coils For the homogeneousmodel, in worst case, an optimal frequency is about 7 MHz In the inhomogeneous model,that is more realistic, the highest quality factor could be optained with 8.4 MHz It can besaid, that the human tissue reduces the optimal frequency value, at which the most voltagecan be induced respectively the highest transmission range could be achieved The optimalfrequency can be observed near to the 6,78 MHz ISM band In comparison to LF ISM Bandthe amount of induced voltage is about 4 times higher In comparison to 13.56 MHz a power

of maximum 20 % less is necessary to get the same transmission range

4.2.1 Practical Measurement

An experimental measurement shall determine the maximum achievable distance For thisexperiment, a circular coil with a single winding and an aperture of 26 cm was used to pro-duce the magnetic field A frequency of 13,56 MHz was chosen A test transponder wasdeveloped to measure the energy that can be provided to an implanted transponder It con-sists of a ferrite rod coil, an HF front-end and a load resistor that simulates the impedance of

a transponder circuit To create a substitute that simulates the electric properties of the man body, a phantom substance was prepared following a recipe described in [2] The maingoal of the experiment is to measure the voltage induced at the transponder coil when it isplaced inside this substance at different distances from the reader coil 50 L of the phantomsubstance is obtained It was placed in a container large enough to allow the transponder to beplaced in a similar position as in a human body Following the specifications in the article, thecontainer should be made of an electrical insulator and non-magnetic material In our case,the container has a cylindrical form, which is sufficiently similar to a human body Anotherrequirement is a minimum volume of substance It is specified that a mass of at least 30 kg

hu-of phantom material is necessary Generally, a homogeneous phantom is accurate enough tosimulate a human body, in this way it is not necessary to incorporate materials of differentconductivity inside the container Figure 12 (a) shows the measurement setup

With this measurement it is possible to determine in how much surrounding tissue a der can work To measure the provided energy for different distances from the reader, thevoltage at the load resistor in the test-transponder was measured versus the distance Thechip used in sensor transponders usually works with voltages greater than 3 V Therefore, thetransponder would be provided with enough energy at a distance where the voltage is stillhigher than this voltage Figure 12 (b) shows the measurement results

transpon-The measurement was done with a voltage amplitude at the reader coil of 300 V and a loadresistor in the test transponder of 60 kOhm and 100 kOhm These values were chosen empiri-cally The diagram shows the voltage that would be available for a chip in different distances.The voltage is grater than 3 V for distances up to 43 cm

The experimental measurement shows, that a sensor transponder can work inside a humantissue up to a distance of 40 cm

5 Conclusion

The influence by the human tissue on the inductive energy transmission was considered forthe design of a sensor transponder system For the given constraints to the transponder an-tenna an optimal frequency could be found The loss effects decrease this optimum frequency

(a) Measurement setup (b) Results

Fig 12 Practical measurement

A carrier frequency around 6,78 MHz is an optimal choice for our constraints Measurementshave determined the achievable transmission distance through human body

[4] Klaus Finkenzeller; RFID-Handbook; Hanser; M ¨unchen Wien; 2006

[5] A Hennig; RF Energy Transmission for Sensor Transponders Deeply Implanted in man Bodies; EmuW IEEE 2008

Tobias Feldengut, Stephan Kolnsberg and Rainer Kokozinski

Fraunhofer Institute for Microelectronic Circuits and Systems (IMS)

Germany

1 Introduction

The importance of wireless sensors in medical systems, automotive applications, and

environmental monitoring is growing continuously A sensor node converts physical values

such as pressure, temperature, or mechanical stress to digital values The wireless interface

connects it to a base station or a network for further data processing Most of these products

are required to be light, cheap, long lived, and maintenance free Remote powering of

transponder tags is a key technology to meet these demands, because it obviates the need

for a battery Near field systems usually operate in the low frequency range, typically

between the 133 kHz (LF) and the 13.56 MHz (HF) ISM bands While LF and HF systems

operate in the magnetic near field via inductive coupling between two coils, UHF systems

use electromagnetic waves in the far field of the base station The range of the available

inductive systems is typically limited to less than one meter, which motivates the use of far

field energy transmission at ultra high frequencies This chapter presents the design of a

passive long range transponder with temperature sensor The system is shown in figure 1

reader

energy data Application

data

tag 1

tag k tag 2

Fig 1.passive far-field transponder system

A base station transmits an 868 MHz carrier wave that is modulated with the forward link

data In the transponder chip, the antenna voltage is rectified and multiplied to serve as the

supply voltage for the integrated circuits including the sensor and a digital part When the

tag is transmitting data to the reader (backward link), it switches its input impedance

20

between two different states to modulate its own radar cross section The transponder is

shown in figure 2 It consists of an integrated circuit and an antenna The ASIC comprises an

analog front-end as an air interface, a digital part for protocol handling, as well as

non-volatile memory The temperature sensor and the readout circuit are integrated on the same

chip

sensor transponder ASIC

sensor-readout analog UHF front-end

POR

sensor

temperature-voltage regulator

Vdd

data

calibration-Fig 2 sensor transponder architecture

The power supply block generates a stable 1.5 V voltage for the other circuit blocks by

rectifying and regulating the incoming RF signal The modem contains a simple low-power

ASK demodulation circuit and a modulation switch The carrier frequency from the reader

is far too high to serve as a clock for the digital part, so that a local oscillator circuit is

required A bandgap circuit generates supply independent reference voltages and bias

currents It also generates a temperature-dependent voltage that is amplified to serve as the

temperature sensor This chapter is focused on the design of the analog front-end

Antenna equivalent circuit Chip Input Impedance

Fig 3 simple equivalent circuit of transponder input

According to the well known Friis relation

2

2)4

( d G P

[Curty et al, 2005] Antenna matching is used to achieve high input voltage amplitude as

well as power matching The amplitude of the incoming signal is often as low as the threshold voltage of the rectifying devices, and sufficient rectifier efficiency is therefore difficult to achieve Chapter 2.1 is focused on the rectifier optimisation

2.1 Rectifier

The rectifier is the most critical circuit for efficient energy transmission The input from the antenna is a high frequency (868 MHz) signal with amplitude of less than 500 mV at large distance from the base station Rectifying diodes are required to operate at (or slightly below) the threshold voltage Recent research efforts have focused on the modelling and the optimisation of the typically used multi-stage Dickson charge pump [Curty et al., 2005]; [Karthaus & Fischer, 2003] This circuit is shown in figure 4

Ideally, diode D1 and capacitor C1 lift the AC input voltage up by its peak value Diode D2 and capacitor C2 create a peak detector, so that the output voltage of the first stage is set to twice the input amplitude Several stages are cascaded to reach an output voltage that is high enough for the reliable operation of all circuits At high frequencies and at low input voltage levels, the behavior of actual implementations differs significantly from the predictions of this simplified explanation [Karthaus & Fischer, 2003] This fact results from the parasitics of real world devices, especially in cheap standard CMOS solutions The following effects are detrimental to the rectifier performance The diodes exhibit

 forward voltage drop,

 significant substrate and junction capacitances,

 reverse leakage current, and substrate leakage

These values depend not only on the diode area, but also on the output current of the rectifier, on the temperature, and on random process variations In addition to the diode parasitics, integrated capacitors exhibit parasitic substrate capacitances, series resistance, limited capacitance values, and leakage current Finally, package parasitics, pad capacitance,

between two different states to modulate its own radar cross section The transponder is

shown in figure 2 It consists of an integrated circuit and an antenna The ASIC comprises an

analog front-end as an air interface, a digital part for protocol handling, as well as

non-volatile memory The temperature sensor and the readout circuit are integrated on the same

chip

sensor transponder ASIC

sensor-readout analog UHF front-end

POR

sensor

temperature-voltage regulator

Vdd

data

calibration-Fig 2 sensor transponder architecture

The power supply block generates a stable 1.5 V voltage for the other circuit blocks by

rectifying and regulating the incoming RF signal The modem contains a simple low-power

ASK demodulation circuit and a modulation switch The carrier frequency from the reader

is far too high to serve as a clock for the digital part, so that a local oscillator circuit is

required A bandgap circuit generates supply independent reference voltages and bias

currents It also generates a temperature-dependent voltage that is amplified to serve as the

temperature sensor This chapter is focused on the design of the analog front-end

Antenna equivalent circuit Chip Input Impedance

Fig 3 simple equivalent circuit of transponder input

According to the well known Friis relation

2

2)4

( d G P

[Curty et al, 2005] Antenna matching is used to achieve high input voltage amplitude as

well as power matching The amplitude of the incoming signal is often as low as the threshold voltage of the rectifying devices, and sufficient rectifier efficiency is therefore difficult to achieve Chapter 2.1 is focused on the rectifier optimisation

2.1 Rectifier

The rectifier is the most critical circuit for efficient energy transmission The input from the antenna is a high frequency (868 MHz) signal with amplitude of less than 500 mV at large distance from the base station Rectifying diodes are required to operate at (or slightly below) the threshold voltage Recent research efforts have focused on the modelling and the optimisation of the typically used multi-stage Dickson charge pump [Curty et al., 2005]; [Karthaus & Fischer, 2003] This circuit is shown in figure 4

Ideally, diode D1 and capacitor C1 lift the AC input voltage up by its peak value Diode D2 and capacitor C2 create a peak detector, so that the output voltage of the first stage is set to twice the input amplitude Several stages are cascaded to reach an output voltage that is high enough for the reliable operation of all circuits At high frequencies and at low input voltage levels, the behavior of actual implementations differs significantly from the predictions of this simplified explanation [Karthaus & Fischer, 2003] This fact results from the parasitics of real world devices, especially in cheap standard CMOS solutions The following effects are detrimental to the rectifier performance The diodes exhibit

 forward voltage drop,

 significant substrate and junction capacitances,

 reverse leakage current, and substrate leakage

These values depend not only on the diode area, but also on the output current of the rectifier, on the temperature, and on random process variations In addition to the diode parasitics, integrated capacitors exhibit parasitic substrate capacitances, series resistance, limited capacitance values, and leakage current Finally, package parasitics, pad capacitance,

and metal line parasitics can not be neglected All of the above mentioned effects need to be

considered and make the rectifier design a challenging task

VA = 2(Vin - VDrop) D1

C1

D2

C2 V(t) = Vin sin( t)

V(t)

Additional

Casc

ed Stages

Fig 4 basic rectifier stage, linearised model, and input impedance [Curty et al., 1005]

Figure 4 also shows a linear model of the rectifier circuit [1] It consists of an input resistance

and an input capacitance, as well as an output voltage source and an output resistance The

antenna should be inductively matched to this input impedance of the rectifier The tag also

exhibits an input capacitance due to the above mentioned parasitic capacitances

Fig 5 multi stage voltage multiplier

At high quality factors, the bandwidth of the system is reduced, and inductive antenna matching is more difficult to achieve, so the input capacitance should be minimised In summary, the goal of the rectifier design is to minimise the required input voltage and to achieve a large input impedance and low input capacitance for a given output current consumption The major design concerns are therefore minimising the capacitance between

AC and DC nodes as well as reducing the voltage drop Figure 5 shows a multiplier stack with several stages, as well as the resulting waveforms at the AC terminal and the DC nodes Each diode reduces the achievable output voltage by one threshold voltage drop This voltage drop is a significant issue, as 12 diodes create a voltage drop of more than 2 V One way to reduce this voltage drop is to use transistor diodes with threshold voltage compensation This approach is depicted in figure 6 [Nakamoto et al, 2007] The gate of the transistor diode is forward biased by one threshold voltage, so that the device effectively acts as a diode with zero threshold voltage The threshold voltage of transistors varies with temperature and process fluctuations, and the compensation voltage needs to track this variation A compensation voltage that is too large would significantly increase reverse leakage currents, because the transistor could not close properly (see figure 7)

The DC voltage at the anode of M2 is VS When the voltage at terminal VB is larger than

VS+Vth, the compensation voltage at the gate of the rectifying diode equals one threshold voltage in a first order approximation, independent of VB When the diodes are matched, the compensation works independent of the temperature, the bulk-source voltage and process variations The compensation of PMOS transistors works in an analog fashion

and metal line parasitics can not be neglected All of the above mentioned effects need to be

considered and make the rectifier design a challenging task

VA = 2(Vin - VDrop) D1

C1

D2

C2 V(t) = Vin sin( t)

V(t)

Additiona

l

Casc

ed Stages

Fig 4 basic rectifier stage, linearised model, and input impedance [Curty et al., 1005]

Figure 4 also shows a linear model of the rectifier circuit [1] It consists of an input resistance

and an input capacitance, as well as an output voltage source and an output resistance The

antenna should be inductively matched to this input impedance of the rectifier The tag also

exhibits an input capacitance due to the above mentioned parasitic capacitances

Fig 5 multi stage voltage multiplier

At high quality factors, the bandwidth of the system is reduced, and inductive antenna matching is more difficult to achieve, so the input capacitance should be minimised In summary, the goal of the rectifier design is to minimise the required input voltage and to achieve a large input impedance and low input capacitance for a given output current consumption The major design concerns are therefore minimising the capacitance between

AC and DC nodes as well as reducing the voltage drop Figure 5 shows a multiplier stack with several stages, as well as the resulting waveforms at the AC terminal and the DC nodes Each diode reduces the achievable output voltage by one threshold voltage drop This voltage drop is a significant issue, as 12 diodes create a voltage drop of more than 2 V One way to reduce this voltage drop is to use transistor diodes with threshold voltage compensation This approach is depicted in figure 6 [Nakamoto et al, 2007] The gate of the transistor diode is forward biased by one threshold voltage, so that the device effectively acts as a diode with zero threshold voltage The threshold voltage of transistors varies with temperature and process fluctuations, and the compensation voltage needs to track this variation A compensation voltage that is too large would significantly increase reverse leakage currents, because the transistor could not close properly (see figure 7)

The DC voltage at the anode of M2 is VS When the voltage at terminal VB is larger than

VS+Vth, the compensation voltage at the gate of the rectifying diode equals one threshold voltage in a first order approximation, independent of VB When the diodes are matched, the compensation works independent of the temperature, the bulk-source voltage and process variations The compensation of PMOS transistors works in an analog fashion

MP = 0.

6V

VC

MP = 0.

9V

Fig 7 I(V) characteristic of transistor diode with Vth compensation voltage

Figure 9 shows the complete rectifier circuit [Feldengut et al., 2008] It consists of two

separate charge pumps, the first of which serves only as a compensation voltage generator

(B) for the NMOS transistors in the main rectifier (A) The second rectifier (B) consists of

eight stages with minimum area standard CMOS Schottky diodes It can generate a large

output voltage because it is almost unloaded The output of this second rectifier is applied to

the VB terminal of the compensation circuit in figure 8

DC DC AC

Fig 8 compensation voltage generator

The total current consumption of the voltage dividers for the threshold voltage generation is

less than 150 nA, while the output current of the six-stage main rectifier is typically several

To reduce the current load and the required resistor size, the voltage dividers that compensate the lower stages of rectifier (A) are connected to intermediate output stages of rectifier (B) The voltage across the resistors is therefore very small The compensation transistors have a much smaller aspect ratio, so that the voltage drop equals the threshold voltage across the rectification transistors, even when the current through the voltage dividers is several times smaller than the current flowing through the main rectifier stack (A)

Figure 11 shows the results for the conventional Schottky diode rectifier as well as for the proposed circuit The load resistance is 300 kOhm and the output capacitor is 100 pF in both cases The antenna resistance is 300 Ohm At a distance of 4.5 m between the base station and the transponder, the input power is -11.3 dBm at a transmitted power of 2 W and a carrier frequency of 868 MHz At this distance, the proposed rectifier (fig 10) can power a transponder chip with 1.5 V supply voltage and 5 µA DC current, while the output voltage

of the conventional circuit (fig 5) is close to zero in the chosen process technology

MP = 0.

6V

VC

MP = 0.

9V

Fig 7 I(V) characteristic of transistor diode with Vth compensation voltage

Figure 9 shows the complete rectifier circuit [Feldengut et al., 2008] It consists of two

separate charge pumps, the first of which serves only as a compensation voltage generator

(B) for the NMOS transistors in the main rectifier (A) The second rectifier (B) consists of

eight stages with minimum area standard CMOS Schottky diodes It can generate a large

output voltage because it is almost unloaded The output of this second rectifier is applied to

the VB terminal of the compensation circuit in figure 8

DC DC

AC

Fig 8 compensation voltage generator

The total current consumption of the voltage dividers for the threshold voltage generation is

less than 150 nA, while the output current of the six-stage main rectifier is typically several

To reduce the current load and the required resistor size, the voltage dividers that compensate the lower stages of rectifier (A) are connected to intermediate output stages of rectifier (B) The voltage across the resistors is therefore very small The compensation transistors have a much smaller aspect ratio, so that the voltage drop equals the threshold voltage across the rectification transistors, even when the current through the voltage dividers is several times smaller than the current flowing through the main rectifier stack (A)

Figure 11 shows the results for the conventional Schottky diode rectifier as well as for the proposed circuit The load resistance is 300 kOhm and the output capacitor is 100 pF in both cases The antenna resistance is 300 Ohm At a distance of 4.5 m between the base station and the transponder, the input power is -11.3 dBm at a transmitted power of 2 W and a carrier frequency of 868 MHz At this distance, the proposed rectifier (fig 10) can power a transponder chip with 1.5 V supply voltage and 5 µA DC current, while the output voltage

of the conventional circuit (fig 5) is close to zero in the chosen process technology

Fig 10 complete circuit implementation

The transient start-up behaviour of the circuit is shown for different input voltages in

figure 12 The output capacitor was reduced to only 20 pF in order to reduce the simulation

time The start-up waveform differs significantly from the typical capacitor loading

waveform of a conventional diode multiplier Once the compensation voltage has been

build up, the rectifying transistors’ efficiency is significantly increased, which changes the

conversion efficiency For very high output voltages of more than 2.5 V, some transistors

are over compensated and begin to exhibit reverse leakage current This leads to an abrupt

stop of the output voltage increase

Depending on the process technology, the number of stages may have to be reduced by one

or two in each of the two rectifier stacks in order to reduce the input capacitance The input

capacitance has two negative effects: the first is that the bandwidth of the system is

significantly reduced when the quality factor Q is high The second issue with a large input

capacitance is that it may also reduce the real part of the input impedance When a large

parasitic resistor lies in series to a large parasitic capacitance (this is often the case for

substrate parasitics of diodes and capacitors in bulk CMOS), the equivalent parallel RC tank

has a reduced resistance at the frequency of interest

Fig 11 Simulated output voltage as a function of input power under ideal matching conditions

0.00.51.01.52.02.5

Fig 10 complete circuit implementation

The transient start-up behaviour of the circuit is shown for different input voltages in

figure 12 The output capacitor was reduced to only 20 pF in order to reduce the simulation

time The start-up waveform differs significantly from the typical capacitor loading

waveform of a conventional diode multiplier Once the compensation voltage has been

build up, the rectifying transistors’ efficiency is significantly increased, which changes the

conversion efficiency For very high output voltages of more than 2.5 V, some transistors

are over compensated and begin to exhibit reverse leakage current This leads to an abrupt

stop of the output voltage increase

Depending on the process technology, the number of stages may have to be reduced by one

or two in each of the two rectifier stacks in order to reduce the input capacitance The input

capacitance has two negative effects: the first is that the bandwidth of the system is

significantly reduced when the quality factor Q is high The second issue with a large input

capacitance is that it may also reduce the real part of the input impedance When a large

parasitic resistor lies in series to a large parasitic capacitance (this is often the case for

substrate parasitics of diodes and capacitors in bulk CMOS), the equivalent parallel RC tank

has a reduced resistance at the frequency of interest

Fig 11 Simulated output voltage as a function of input power under ideal matching conditions

0.00.51.01.52.02.5