Electronically Tunable Mixed-Mode Universal Filter Employing a Single Active Block and a Minimum Number of Passive Components

: A recently developed active building block, namely Voltage Differencing Extra X Current Conveyor (VD-EXCCII), is employed in the design of multi input single output (MISO), electronically tunable mixed-mode universal ﬁlter. The ﬁlter provides low pass (LP), high pass (HP), band pass (BP), band reject (BR) and all pass (AP) responses in current-mode (CM), voltage-mode (VM), trans-impedance-mode (TIM) and trans-admittance-mode (TAM). The ﬁlter employs a single VD-EXCCII, three resistors and two capacitors. Additionally, a CM single input multi output (SIMO) ﬁlter can be derived from the same circuit topology by only adding current output terminals. The attractive features of the ﬁlter include: (i) the ability to operate in all four modes, (ii) the tunability of the Q factor independent of pole frequency, (iii) the low output impedance for the VM ﬁlter, (iv) the high output impedance current output for CM and TAM ﬁlters and (v) no requirement for double/negative input signals (voltage/current) for response realization. The VD-EXCCII and its layout is designed and validated in Cadence Virtuoso using 0.18 µ m pdk from Silterra Malaysia with a supply voltage of ± 1.25 V. The operation of the ﬁlter is examined at the 8.0844 MHz characteristic frequency. A non-ideal parasitic and sensitivity analysis is also carried out to study the effect of process and components spread on the ﬁlter performance. with a ± 1.25 V supply. The Monte Carlo analysis shows that the frequency deviation is within acceptable limits. Furthermore, the THD is within 5% for a considerable voltage/current input signal range. The power dissipation of the ﬁlter is found to be 5.76 mW. The VD-EXCCII and the mixed mode ﬁlter are also designed and tested using the models of commercially available ICs the AD844 and LM13700 in PSpice. The simulation results are found consistent with the theoretical predictions.


Introduction
The current-mode (CM) active building blocks (ABBs) are widely employed in designing universal frequency filters. The CM ABBs exhibits greater linearity, wide bandwidth, simple structure, low power consumption and enhanced dynamic range [1][2][3][4][5]. Extensive number of filter topologies using CM ABBs can be found in the literature. However, the majority of the previously proposed filters can work only in single mode of operation i.e., CM, voltage-mode (VM), trans-impedance-mode (TIM) or trans-admittance-mode (TAM) [1][2][3]5]. In present-day intricate signal processing systems, the interaction between CM and VM circuits is required. This task can be accomplished by TAM and TIM filters that not only perform signal processing, but also provide interfacing between VM and CM systems [6][7][8][9][10]. The development of mixed-mode universal filters that can provide low pass (LP), high pass (HP), band pass (BP), band reject (BR) and all pass (AP) filter responses in CM, VM, TAM and TIM modes of operation are best suited for the task.
In the literature, numerous CM ABBs can be found, each having its own merits. In this research, the Voltage Differencing Extra X Current Conveyor (VD-EXCCII) is introduced and utilized in the design of mixed-mode filter. The proposed VD-EXCCII can be considered a universal ABB, as it can realize many popular and widely employed CM ABBs as special case. The proposed VD-EXCCII can realize second generation current conveyor (CCII), voltage differencing current conveyor (VDCC), differential difference current conveyor (DDCC), voltage differencing transconductance amplifier (VDTA), voltage differencing buffered amplifier (VDBA), current backward transconductance amplifier (CBTA) and operational transconductance amplifier (OTA) by proper interconnection of its input and output terminals, thereby making it an inherently universal ABB. This will allow designers to employ VD-EXCCII instead of using each separate ABBs to test their designs, thus reducing the cost and time to market. The filter design requires a single VD-EXCCII, two capacitors and three resistors. The striking features of the proposed filters are: (i) employment of single active block (VD-EXCCII), (ii) ability to work in all four modes of operation, (iii) provision for inbuilt tunability, (iv) the filter enjoy low active and passive sensitivities. Moreover, the filter enjoy all (iv-x) properties mentioned in Table 2. The design simulation of the VD-EXCCII is done in Cadence Virtuoso using Silterra Malaysia 0.18 µm PDK. The post-layout simulation results are in good agreement with the theoretical predictions.

Voltage Differencing Extra X Current Conveyor (VD-EXCCII)
The proposed Voltage Differencing Extra X current conveyor (VD-EXCCII) is derived by connecting extra X second generation current conveyor (EXCCII) [42] and operational transconductance amplifier (OTA). The first stage comprises of OTA followed by the CCII with two current input terminals. The developed active element has characteristics of CCII and tunable OTA in one structure. The voltage-current (V-I) characteristics of the developed VD-EXCCII are presented in Equations (1)-(4) and the block diagram is presented in Figure 1.
The CMOS implementation of VD-EXCCII is given in Figure 2. The first stage consists of OTA MOS transistors (M1-M14). The output current of the transconductor depends on the voltage difference ( − ) . Assuming that all transistors are operating in saturation region and the transistors (M1-M2) have equal width to length ratio, the output current is given by: where the transconductance parameter = µ 2 , (i = 1, 2), W is the effective channel width, L is the effective length of the channel, is the gate oxide capacitance per unit area and µ is the carrier mobility.
The second stage is made up of hybrid voltage and current followers (M15-M44). The voltage developed at node W is transferred to nodes and . In the same way the input current from node is transferred to and . Furthermore, the input current from node is transferred to and . The current flowing in the and terminals are independent of each other. The class AB output stage is utilized in the output stage, as it is suitable for low voltage operation and better dynamic range [2]. The current and voltage reference circuits available in the literature [43] can be employed to generate the IBias and VBias for the circuit.
The small signal analysis yields the expression relating V , V and V . The analysis is carried out for the differential stage formed by transistors (M − M , M − M ). The voltage transfer ratio between the W and node can be derived as given in Equation (6).  The CMOS implementation of VD-EXCCII is given in Figure 2. The first stage consists of OTA MOS transistors (M1-M14). The output current of the transconductor depends on the voltage difference (V P − V N ). Assuming that all transistors are operating in saturation region and the transistors (M1-M2) have equal width to length ratio, the output current is given by: where the transconductance parameter K i = µC ox W 2L , (i = 1, 2), W is the effective channel width, L is the effective length of the channel, C ox is the gate oxide capacitance per unit area and µ is the carrier mobility. The second stage is made up of hybrid voltage and current followers (M15-M44). The voltage developed at node W is transferred to nodes X P and X N . In the same way the input current from X P node is transferred to Z P+ and Z P− . Furthermore, the input current from X N node is transferred to Z N+ and Z N− . The current flowing in the Z N and Z P terminals are independent of each other. The class AB output stage is utilized in the output stage, as it is suitable for low voltage operation and better dynamic range [2]. The current and voltage reference circuits available in the literature [43] can be employed to generate the I Bias and V Bias for the circuit.  The small signal analysis yields the expression relating V W , V XP and V XN . The analysis is carried out for the differential stage formed by transistors (M 15 -M 24 , M 39 -M 42 ). The voltage transfer ratio between the W and X P node can be derived as given in Equation (6).
is the output resistance of the MOS transistor and transconductance of the MOS transistor . Similarly, the voltage transfer gain is com

= ≌
The current transfer ratios are derived as given in Equations (8-9): The terminal resistance of and terminals is calculated as given in Equations The , and resistance are presented in Equations (12)- (14). The , and node impedances are found to be high given by the para resistance of the MOS transistors.

Proposed Electronically Tunable Mixed-Mode Universal Filter
The proposed MISO mixed mode filter is presented in Figure 3. It provides all responses in VM, TAM, TIM and CM modes of operation. The minimum component filte a single VD-EXCCII, three resistors and two capacitors. The main attributes of the filter are a single active element, (ii) employment of only five passive components, (iii) no need for matching condition, (iv) availability of VM output from low impedance terminal, (v) ava TAM and CM output from high impedance terminals, (vi) no requirement for negative/do signals for response realization, (vii) inbuilt tunability of Q independent of frequency an TIM mode, the filter gain can be adjusted without affecting and Q.
where r 0p = r 017 //r 021 , r 0 is the output resistance of the MOS transistor and g mi is the transconductance of the MOS transistor M i . Similarly, the voltage transfer gain α N is computed as: The current transfer ratios are derived as given in Equations (8-9): The terminal resistance of and terminals is calculated as given in Equations (10)- (11). The , and resistance are presented in Equations (12)- (14). where The , and node impedances are found to be high given by the parallel output resistance of the MOS transistors.

Proposed Electronically Tunable Mixed-Mode Universal Filter
The proposed MISO mixed mode filter is presented in Figure 3. It provides all five filter responses in VM, TAM, TIM and CM modes of operation. The minimum component filter employs a single VD-EXCCII, three resistors and two capacitors. The main attributes of the filter are: (i) use of a single active element, (ii) employment of only five passive components, (iii) no need for capacitive matching condition, (iv) availability of VM output from low impedance terminal, (v) availability of TAM and CM output from high impedance terminals, (vi) no requirement for negative/double input signals for response realization, (vii) inbuilt tunability of Q independent of frequency and (viii) in TIM mode, the filter gain can be adjusted without affecting and Q.

Proposed Electronically Tunable Mixed-Mode Universal Filter
The proposed MISO mixed mode filter is presented in Figure 3. It provides all five responses in VM, TAM, TIM and CM modes of operation. The minimum component filter em a single VD-EXCCII, three resistors and two capacitors. The main attributes of the filter are: (i) u a single active element, (ii) employment of only five passive components, (iii) no need for capa matching condition, (iv) availability of VM output from low impedance terminal, (v) availabil TAM and CM output from high impedance terminals, (vi) no requirement for negative/double signals for response realization, (vii) inbuilt tunability of Q independent of frequency and (v TIM mode, the filter gain can be adjusted without affecting and Q. The current transfer ratios are derived as given in Equations (8-9): The terminal resistance of and terminals is calculated as given in Equations (10)- (11). The , and resistance are presented in Equations (12)- (14). where The , and node impedances are found to be high given by the parallel output resistance of the MOS transistors.

Proposed Electronically Tunable Mixed-Mode Universal Filter
The proposed MISO mixed mode filter is presented in Figure 3. It provides all five filter responses in VM, TAM, TIM and CM modes of operation. The minimum component filter employs a single VD-EXCCII, three resistors and two capacitors. The main attributes of the filter are: (i) use of a single active element, (ii) employment of only five passive components, (iii) no need for capacitive matching condition, (iv) availability of VM output from low impedance terminal, (v) availability of TAM and CM output from high impedance terminals, (vi) no requirement for negative/double input signals for response realization, (vii) inbuilt tunability of Q independent of frequency and (viii) in TIM mode, the filter gain can be adjusted without affecting and Q.
The terminal resistance of X N and X P terminals is calculated as given in Equations (10) and (11). The Z P+ , Z N+ and W C+ resistance are presented in Equations (12)- (14).
Appl. Sci. 2020, 10, x FOR PEER REVIEW 7 of 26 where = // , is the output resistance of the MOS transistor and is the transconductance of the MOS transistor . Similarly, the voltage transfer gain is computed as: The current transfer ratios are derived as given in Equations (8-9): The terminal resistance of and terminals is calculated as given in Equations (10)- (11). The , and resistance are presented in Equations (12)- (14). where The , and node impedances are found to be high given by the parallel output resistance of the MOS transistors.

Proposed Electronically Tunable Mixed-Mode Universal Filter
The proposed MISO mixed mode filter is presented in Figure 3. It provides all five filter responses in VM, TAM, TIM and CM modes of operation. The minimum component filter employs a single VD-EXCCII, three resistors and two capacitors. The main attributes of the filter are: (i) use of a single active element, (ii) employment of only five passive components, (iii) no need for capacitive matching condition, (iv) availability of VM output from low impedance terminal, (v) availability of TAM and CM output from high impedance terminals, (vi) no requirement for negative/double input signals for response realization, (vii) inbuilt tunability of Q independent of frequency and (viii) in 2 r 0p g m18 (g m39 + g m40 ) ( The current transfer ratios are derived as given in Equations (8-9): The terminal resistance of and terminals is calculated as given in Equations (10)- (11). The , and resistance are presented in Equations (12)- (14). where The , and node impedances are found to be high given by the parallel output resistance of the MOS transistors.

Proposed Electronically Tunable Mixed-Mode Universal Filter
The proposed MISO mixed mode filter is presented in Figure 3. It provides all five filter responses in VM, TAM, TIM and CM modes of operation. The minimum component filter employs a single VD-EXCCII, three resistors and two capacitors. The main attributes of the filter are: (i) use of 2 r 0n g m15 (g m23 + g m24 ) (11) where r 0p = r 017 //r 021 and r 0n = r 016 //r 020 .

Proposed Electronically Tunable Mixed-Mode Universal Filter
The proposed MISO mixed mode filter is presented in Figure 3. It provides all five filter responses in VM, TAM, TIM and CM modes of operation. The minimum component filter employs a single VD-EXCCII, three resistors and two capacitors. The main attributes of the filter are: (i) use of a single active element, (ii) employment of only five passive components, (iii) no need for capacitive matching condition, (iv) availability of VM output from low impedance terminal, (v) availability of TAM and CM output from high impedance terminals, (vi) no requirement for negative/double input signals for response realization,

Operation in VM and TAM
In this mode of operation, all input currents ( − ) are set to zero. The input voltages ( − ) are applied as per the sequence given in Table 3. The filter transfer functions and expression for pole frequency and quality factor are given in Equations (15)-(18): The Equations (17) and (18) imply that when the frequency is varied, the quality factor of the filter will be slightly affected. The frequency can be tuned without affecting the quality factor if and are varied simultaneously such that the product (  ) remains constant. The resistor can be realized using an MOS [44] transistor, making it easily tunable. Table 3. Input excitation sequence.

Operation in VM and TAM
In this mode of operation, all input currents (I 1 -I 3 ) are set to zero. The input voltages (V 1 -V 3 ) are applied as per the sequence given in Table 3. The filter transfer functions and expression for pole frequency and quality factor are given in Equations (15)- (18):

Response Inputs Passive Matching Condition
The Equations (17) and (18) imply that when the frequency is varied, the quality factor of the filter will be slightly affected. The frequency can be tuned without affecting the quality factor if g m1 and R 1 are varied simultaneously such that the product (R 1 ·g m1 ) remains constant. The resistor can be realized using a MOS [44] transistor, making it easily tunable.

Operation in CM and TIM
In CM mode, the input voltages are set to zero. The input currents (I 1 -I 3 ) are applied according to Table 4 to obtain CM and TIM responses. Table 4. Input excitation sequence of operation in CM and TIM.

Response
Inputs Passive Matching Condition The filter transfer functions are given in Equations (19) and (20): In TIM mode, the gain of the filter can be varied without disturbing the Q and ω 0 of the filter by varying the value of R 3 , as in the transfer function, R 3 is common to all the responses: Additionally, the CM SIMO filter shown in Figure 4 is derived from the proposed mixed mode filter by adding additional current output terminals without changing the core topology. All passive components are grounded in CM SIMO configuration. The filter requires a single VD-EXCCII, two resistors and two capacitors, all grounded. The resistor R 3 is not required. since it is needed only to obtain the TIM response. The attractive features of the derived SIMO CM filter include (i) use of single active element, (ii) employment of only three grounded passive components, (iii) no need for passive components matching condition, (iv) low input impedance, (v) high output impedance and (vi) inbuilt tunability.

Operation in CM and TIM
In CM mode, the input voltages are set to zero. The input currents ( − ) are applied according to Table 4 to obtain CM and TIM responses.

Response
Inputs The filter transfer functions are given in Equations (19)(20): In TIM mode, the gain of the filter can be varied without disturbing the Q and of the filter by varying the value of , as in the transfer function, is common to all the responses: Additionally, the CM SIMO filter shown in Figure 4 is derived from the proposed mixed mode filter by adding additional current output terminals without changing the core topology. All passive components are grounded in CM SIMO configuration. The filter requires a single VD-EXCCII, two resistors and two capacitors, all grounded. The resistor is not required. since it is needed only to obtain the TIM response. The attractive features of the derived SIMO CM filter include (i) use of single active element, (ii) employment of only three grounded passive components, (iii) no need for passive components matching condition, (iv) low input impedance, (v) high output impedance and (vi) inbuilt tunability.
The filter transfer functions are given in Equations (21)- (25). Expression for pole frequency and quality factor will be same as that of mixed mode MISO filter given by Equations (17) and (18).  The filter transfer functions are given in Equations (21)- (25). Expression for pole frequency and quality factor will be same as that of mixed mode MISO filter given by Equations (17) and (18).
The BR and AP response can be obtained by simply summing the LP, HP and BP currents: I BR = I HP + I LP and I AP = I HP + I LP + I BP .

Non-Ideal Gain and Sensitivity Analysis
The non-ideal effects that influences the response of the VD-EXCCII are the frequencydependent non-ideal current (α P/N , α P/N ), voltage (β P/N ) and transconductance transfer (γ, γ ) gains. These non-ideal gains result in a change in the current and voltage signals during transfer leading to undesired response. Taking into account the non-ideal gains, the V-I characteristics of the VD-EXCCII in (1-4) will be modified as follows: Here, ε v(P,N) ε v(P,N) << 1 denote voltage tracking errors, ε iP , ε iN (|ε iP |, |ε iN | << 1) denote current tracking errors and ε g m1 , ε g m1 ε g m1 , |ε g m1 | << 1 denote transconductance errors of the VD-EXCCII.
The non-ideal analysis considering the effect of non-ideal current, voltage and transconductance transfer gains is carried out for VM, CM, TAM and TIM configurations to see its effect on the transfer function, f 0 and Q of the proposed filters. The modified expressions of the filter transfer functions, f 0 and Q for the MISO/ SIMO configurations are presented in Equations (26)-(31): The BR and AP response can be obtained by simply summing the LP, HP and BP currents: I = I + I and I = I + I + I .

Non-Ideal Gain and Sensitivity Analysis
The non-ideal effects that influences the response of the VD-EXCCII are the frequencydependent non-ideal current ( / , / ), voltage ( / ) and transconductance transfer (γ, ) gains. These non-ideal gains result in a change in the current and voltage signals during transfer leading to undesired response. Taking into account the non-ideal gains, the V-I characteristics of the VD-EXCCII in (1-4) will be modified as follows: The non-ideal analysis considering the effect of non-ideal current, voltage and transconductance transfer gains is carried out for VM, CM, TAM and TIM configurations to see its effect on the transfer function, f0 and Q of the proposed filters. The modified expressions of the filter transfer functions, and for the MISO/ SIMO configurations are presented in Equations (26)- (31): The BR and AP response can be obtained by simply summing the LP, HP and BP currents I + I and I = I + I + I .

Non-Ideal Gain and Sensitivity Analysis
The non-ideal effects that influences the response of the VD-EXCCII are the freq dependent non-ideal current ( / , / ), voltage ( / ) and transconductance transfer (γ, These non-ideal gains result in a change in the current and voltage signals during transfer lea undesired response. Taking into account the non-ideal gains, the V-I characteristics of the VD-E in (1-4) will be modified as follows: The BR and AP response can be obtained by simply summing the LP, HP and BP currents I + I and I = I + I + I .

Non-Ideal Gain and Sensitivity Analysis
The non-ideal effects that influences the response of the VD-EXCCII are the freq dependent non-ideal current ( / , / ), voltage ( / ) and transconductance transfer (γ, These non-ideal gains result in a change in the current and voltage signals during transfer lea undesired response. Taking into account the non-ideal gains, the V-I characteristics of the VDin (1-4) will be modified as follows:

Non-Ideal Gain and Sensitivity Analysis
The non-ideal effects that influences the response of the VD-EXCCII are the frequencydependent non-ideal current ( / , / ), voltage ( / ) and transconductance transfer (γ, ) gains. These non-ideal gains result in a change in the current and voltage signals during transfer leading to undesired response. Taking into account the non-ideal gains, the V-I characteristics of the VD-EXCCII in (1-4) will be modified as follows:

Non-Ideal Gain and Sensitivity Analysis
The non-ideal effects that influences the response of the VD-EXCCII are the fre dependent non-ideal current ( / , / ), voltage ( / ) and transconductance transfer (γ, These non-ideal gains result in a change in the current and voltage signals during transfer le undesired response. Taking into account the non-ideal gains, the V-I characteristics of the VD in (1-4) will be modified as follows:

Non-Ideal Gain and Sensitivity Analysis
The non-ideal effects that influences the response of the VD-EXCCII are the fre dependent non-ideal current ( / , / ), voltage ( / ) and transconductance transfer (γ, These non-ideal gains result in a change in the current and voltage signals during transfer le undesired response. Taking into account the non-ideal gains, the V-I characteristics of the VD in (1-4) will be modified as follows:

Non-Ideal Gain and Sensitivity Analysis
The non-ideal effects that influences the response of the VD-EXCCII are the frequencydependent non-ideal current ( / , / ), voltage ( / ) and transconductance transfer (γ, ) gains. These non-ideal gains result in a change in the current and voltage signals during transfer leading to undesired response. Taking into account the non-ideal gains, the V-I characteristics of the VD-EXCCII in (1-4) will be modified as follows:

Non-Ideal Gain and Sensitivity Analysis
The non-ideal effects that influences the response of the VD-EXCCII are the f dependent non-ideal current ( / , / ), voltage ( / ) and transconductance transfer (γ, These non-ideal gains result in a change in the current and voltage signals during transfer undesired response. Taking into account the non-ideal gains, the V-I characteristics of the V in (1-4) will be modified as follows:

Non-Ideal Gain and Sensitivity Analysis
The non-ideal effects that influences the response of the VD-EXCCII are the frequencydependent non-ideal current ( / , / ), voltage ( / ) and transconductance transfer (γ, ) gains. These non-ideal gains result in a change in the current and voltage signals during transfer leading to undesired response. Taking into account the non-ideal gains, the V-I characteristics of the VD-EXCCII in (1-4) will be modified as follows:

Non-Ideal Gain and Sensitivity Analysis
The non-ideal effects that influences the response of the VD-EXCCII are the frequencydependent non-ideal current ( / , / ), voltage ( / ) and transconductance transfer (γ, ) gains. These non-ideal gains result in a change in the current and voltage signals during transfer leading to undesired response. Taking into account the non-ideal gains, the V-I characteristics of the VD-EXCCII in (1-4) will be modified as follows: g m1 R 2 + α P β P α P sC 1 R 1 I 2 + α P β P sC 1 R 1 I 3 + α P β P α N

Non-Ideal Gain and Sensitivity Analysis
The non-ideal effects that influences the response of the VD-EXCCII are the dependent non-ideal current ( / , / ), voltage ( / ) and transconductance transfer (γ These non-ideal gains result in a change in the current and voltage signals during transfer undesired response. Taking into account the non-ideal gains, the V-I characteristics of the V in (1-4) will be modified as follows:

Non-Ideal Gain and Sensitivity Analysis
The non-ideal effects that influences the response of the VD-EXCCII are the frequencydependent non-ideal current ( / , / ), voltage ( / ) and transconductance transfer (γ, ) gains. These non-ideal gains result in a change in the current and voltage signals during transfer leading to undesired response. Taking into account the non-ideal gains, the V-I characteristics of the VD-EXCCII in (1-4) will be modified as follows:

Non-Ideal Gain and Sensitivity Analysis
The non-ideal effects that influences the response of the VD-EXCCII are the frequencydependent non-ideal current ( / , / ), voltage ( / ) and transconductance transfer (γ, ) gains. These non-ideal gains result in a change in the current and voltage signals during transfer leading to undesired response. Taking into account the non-ideal gains, the V-I characteristics of the VD-EXCCII in (1-4) will be modified as follows: The sensitivities of ω 0 and Q with respect to the non-ideal gains and passive components are given below: The sensitivities are low and have absolute values not higher than unity.

Non-Ideal Parasitic Analysis
The non-ideal model of the VD-EXCCII is presented in Figure 5. As can be deduced, the various parasitic resistance and capacitance appear in parallel with the input and output nodes of the device. The low impedance X P and X N nodes have a parasitic resistance and inductance in series with them. The associated parasitics at the X nodes can be quantified as Z XP = Z XN = R X(N,P) + sL X(N,P) . However, for the frequency of interest, the inductive effect can be ignored. The parasitic resistance and capacitance associated with the P, N, R ZP+ , R ZP− , R ZN+ , R ZN− , W, W C+ , W C− and Z nodes are The sensitivities of and with respect to the non-ideal gains and passive components are given below: The sensitivities are low and have absolute values not higher than unity.

Non-Ideal Parasitic Analysis
The non-ideal model of the VD-EXCCII is presented in Figure 5. As can be deduced, the various parasitic resistance and capacitance appear in parallel with the input and output nodes of the device. The low impedance and nodes have a parasitic resistance and inductance in series with them. The associated parasitics at the X nodes can be quantified as = = ( , ) + ( , ) . However, for the frequency of interest, the inductive effect can be ignored. The parasitic resistance and capacitance associated with the P, N,  Including the VD-EXCCII parasitics, the denominator of the filter transfer function will be modified as presented in Equation (35): Including the VD-EXCCII parasitics, the denominator of the filter transfer function will be modified as presented in Equation (35): The modified expressions of the frequency and quality factor including the various parasitic effects are presented in Equations (36) and (37): To minimize the parasite effects the values of the passive components should be selected such that C 1 (C ZN− and C P ), C 2 (C ZP− and C W ). In addition, note that the resistors R 1 and R 2 are connected to the low impedance X terminals, so they will absorb the parasitic resistance present at XP and XN terminals since (R 2 R XP and R 1 R XN ).

Simulation and Validation
To validate the proposed mixed-mode filter, it was designed and simulated in Cadence Virtuoso design software. The VD-EXCCII was designed in 0.18 µm Silterra Malaysia technology at a supply voltage of ±1.25 V. The width and length of the transistors used are given in Table 5. The layout of the VD-EXCCII as presented in Figure 6 was drawn using the nhp and php high-performance MOS transistors from the Silterra library, and the layout verification was done using the Calibre tool. The layout occupied a total area of 54.28 × 22.80 µm 2 . The bias current of the OTA was fixed at 120 µA resulting in transconductance of 1.0321 mS. The important design parameters were extracted from post-layout simulations and are summarized in Table 6.
The proposed filter was tested by designing for a center frequency of 8.0844 MHz and quality factor of 1.015 by selecting the passive component as R 1 = R 2 = R 3 = 1 kΩ, C 1 = C 2 = 20 pF and g m1 = 1.0321 mS. The power dissipation of the filter was found to be 5.76 mW. The five filter responses in VM, CM, TAM and TIM modes are presented in Figures 7-10. design software. The VD-EXCCII was designed in 0.18 µm Silterra Malaysia technology at a supply voltage of ±1.25 V. The width and length of the transistors used are given in Table 5. The layout of the VD-EXCCII as presented in Figure 6 was drawn using the nhp and php high-performance MOS transistors from the Silterra library, and the layout verification was done using the Calibre tool. The layout occupied a total area of 54.28 × 22.80 µm . The bias current of the OTA was fixed at 120 µA resulting in transconductance of 1.0321 mS. The important design parameters were extracted from post-layout simulations and are summarized in Table 6. Figure 6. Layout of the VD-EXCCII. Figure 6. Layout of the VD-EXCCII.

Parameters
Silterra Technology       To examine the signal processing capability of the proposed universal filter, the transient analysis was carried out in VM mode for HP, LP and BP responses. A VM sinusoidal signal of 100 mV p-p and a frequency of 8.0844 MHz was applied at the input, and the output was analyzed as presented in Figure 11. It can be inferred from the figure that the phase relation between the input and LP, BP and HP outputs of the filter are correct. presented in Figure 11. It can be inferred from the figure that the phase relation between the input and LP, BP and HP outputs of the filter are correct.
In the presented filter, the quality factor can be set independent of the pole frequency of the filter, as is clear from Equations (17) and (18). The quality factor tunability was verified by analyzing the BP response in CM for different values of R2, as shown in Figure 12. It can be deduced from Figure 12b that the quality factor of the filter can be tuned linearly. The fitting equation using a linear regression with coefficient of determination R 2 = 0.9832, which indicates the fraction of the fitting values that are closest to the line of reference data, is given in Figure 12. The pole frequency of the proposed filter can be tuned by varying the bias current of the OTA, as can be inferred from Equation (17). The tuning property is validated by plotting the VM-AP response for the different values of the OTA bias current, as shown in Figure 13. The fitting equation using a power regression with R 2 = 0.9962 is given in Figure 13b.  In the presented filter, the quality factor can be set independent of the pole frequency of the filter, as is clear from Equations (17) and (18). The quality factor tunability was verified by analyzing the BP response in CM for different values of R 2 , as shown in Figure 12. It can be deduced from Figure 12b that the quality factor of the filter can be tuned linearly. The fitting equation using a linear regression with coefficient of determination R 2 = 0.9832, which indicates the fraction of the fitting values that are closest to the line of reference data, is given in Figure 12. The pole frequency of the proposed filter can be tuned by varying the bias current of the OTA, as can be inferred from Equation (17). The tuning property is validated by plotting the VM-AP response for the different values of the OTA bias current, as shown in Figure 13. The fitting equation using a power regression with R 2 = 0.9962 is given in Figure 13b. To examine the signal processing capability of the proposed universal filter, the transient analysis was carried out in VM mode for HP, LP and BP responses. A VM sinusoidal signal of 100 mVp-p and a frequency of 8.0844 MHz was applied at the input, and the output was analyzed as presented in Figure 11. It can be inferred from the figure that the phase relation between the input and LP, BP and HP outputs of the filter are correct.
In the presented filter, the quality factor can be set independent of the pole frequency of the filter, as is clear from Equations (17) and (18). The quality factor tunability was verified by analyzing the BP response in CM for different values of R2, as shown in Figure 12. It can be deduced from Figure 12b that the quality factor of the filter can be tuned linearly. The fitting equation using a linear regression with coefficient of determination R 2 = 0.9832, which indicates the fraction of the fitting values that are closest to the line of reference data, is given in Figure 12. The pole frequency of the proposed filter can be tuned by varying the bias current of the OTA, as can be inferred from Equation (17). The tuning property is validated by plotting the VM-AP response for the different values of the OTA bias current, as shown in Figure 13. The fitting equation using a power regression with R 2 = 0.9962 is given in Figure 13b.  To study the effect of process spread and the non-idealities of the capacitors employed on the performance of the designed filter, a Monte Carlo analysis is carried out for 200 runs. The Monte Carlo analysis results for the VM BP response are given in Figure 14. The results for CM AP configuration are given Figure 15. Corresponding histograms demonstrate the variations of the pole frequency at −180°. The results indicate that the frequency deviation of the filter is within acceptable limits. This further validates the robustness of the design.
The total harmonic distortion (THD) of the filter for LP and BP responses is plotted for different input signal amplitudes for VM as shown in Figure 16. The THD plot for CM-BP/LP is presented in Figure 17. The THD remains within acceptable limits (≤7.5%) for appreciable input range.  To study the effect of process spread and the non-idealities of the capacitors employed on the performance of the designed filter, a Monte Carlo analysis is carried out for 200 runs. The Monte Carlo analysis results for the VM BP response are given in Figure 14. The results for CM AP configuration are given Figure 15. Corresponding histograms demonstrate the variations of the pole frequency at −180 • . The results indicate that the frequency deviation of the filter is within acceptable limits. This further validates the robustness of the design.
The total harmonic distortion (THD) of the filter for LP and BP responses is plotted for different input signal amplitudes for VM as shown in Figure 16. The THD plot for CM-BP/LP is presented in Figure 17. The THD remains within acceptable limits (≤7.5%) for appreciable input range. To study the effect of process spread and the non-idealities of the capacitors employed on the performance of the designed filter, a Monte Carlo analysis is carried out for 200 runs. The Monte Carlo analysis results for the VM BP response are given in Figure 14. The results for CM AP configuration are given Figure 15. Corresponding histograms demonstrate the variations of the pole frequency at −180°. The results indicate that the frequency deviation of the filter is within acceptable limits. This further validates the robustness of the design.
The total harmonic distortion (THD) of the filter for LP and BP responses is plotted for different input signal amplitudes for VM as shown in Figure 16. The THD plot for CM-BP/LP is presented in Figure 17. The THD remains within acceptable limits (≤7.5%) for appreciable input range.    The decrease in pole frequency of the filter due to rise in temperature can be attributed to the decrease in OTA transconductance. The main factors that influence the transconductance are the threshold voltage ( ) and carrier mobility.
can be approximated as a linear function of temperature [45,46] given by Equation (38):   The decrease in pole frequency of the filter due to rise in temperature can be attributed to the decrease in OTA transconductance. The main factors that influence the transconductance are the threshold voltage ( ) and carrier mobility.
can be approximated as a linear function of temperature [45,46] given by Equation (38):   The decrease in pole frequency of the filter due to rise in temperature can be attributed to the decrease in OTA transconductance. The main factors that influence the transconductance are the threshold voltage ( ) and carrier mobility.
can be approximated as a linear function of temperature [45,46] given by Equation (38): The decrease in pole frequency of the filter due to rise in temperature can be attributed to the decrease in OTA transconductance. The main factors that influence the transconduc-tance are the threshold voltage (V t ) and carrier mobility. V t can be approximated as a linear function of temperature [45,46] given by Equation (38): (38) here, α Vt denotes the threshold voltage temperature coefficient which, varies from −1 mV/ • C to −4 mV/ • C and T O is the reference temperature (300 K).
The dependence of carrier mobility on temperature is modelled by [46]: where α µ is the mobility temperature exponent considered as a constant approximately equal to 1.5. The Equations (38) and (39), show that the threshold voltage (V t ) and mobility (µ N ) exhibit a negative temperature dependence which explains the decrease in frequency with temperature, as shown in Figure 18 for CM AP response.
here, denotes the threshold voltage temperature coefficient which, varies from −1 mV/°C to −4 mV/°C and is the reference temperature (300 K). The dependence of carrier mobility on temperature is modelled by [46]: where µ is the mobility temperature exponent considered as a constant approximately equal to 1.5. The Equations (38) and (39), show that the threshold voltage ( ) and mobility (µ ) exhibit a negative temperature dependence which explains the decrease in frequency with temperature, as shown in Figure 18 for CM AP response.  To validate the proposed CM-SIMO filter, it is designed for a center frequency of 6.4 MHz and quality factor of 1.015 by selecting passive component as R 1 = R 2 = 2 kΩ, C 1 = C 2 = 20 pF and g m1 = 1.0321 mS. The five filter responses in CM mode are presented in Figure 19.
here, denotes the threshold voltage temperature coefficient which, varies from −1 mV/°C to −4 mV/°C and is the reference temperature (300 K). The dependence of carrier mobility on temperature is modelled by [46]: where µ is the mobility temperature exponent considered as a constant approximately equal to 1.5. The Equations (38) and (39), show that the threshold voltage ( ) and mobility (µ ) exhibit a negative temperature dependence which explains the decrease in frequency with temperature, as shown in Figure 18 for CM AP response.  The time domain and Monte Carlo analysis results of the filter are presented in Figures 20 and  21, which verify the correct filter operation. The histogram depicted in Figure 21 demonstrates the variations of the pole frequency at −180°.
The Q factor tunability is tested for different values of resistor , as presented in Figure 22. The fitting equation using a linear regression with R 2 = 0.9986 is given in Figure 22b. Furthermore, the total harmonic distortion for different input current amplitudes is shown in Figure 23. It can be inferred that the THD remains approximately 2.5% for a considerable signal range.    The time domain and Monte Carlo analysis results of the filter are presented in Figures 20 and  21, which verify the correct filter operation. The histogram depicted in Figure 21 demonstrates the variations of the pole frequency at −180°.
The Q factor tunability is tested for different values of resistor , as presented in Figure 22. The fitting equation using a linear regression with R 2 = 0.9986 is given in Figure 22b. Furthermore, the total harmonic distortion for different input current amplitudes is shown in Figure 23. It can be inferred that the THD remains approximately 2.5% for a considerable signal range.   The Q factor tunability is tested for different values of resistor R 2 , as presented in Figure 22. The fitting equation using a linear regression with R 2 = 0.9986 is given in Figure 22b. Furthermore, the total harmonic distortion for different input current amplitudes is shown in Figure 23. It can be inferred that the THD remains approximately 2.5% for a considerable signal range.  It can be concluded from the results that the characteristics of the AP filters have a slight imperfection, as all magnitude responses have a hump at the resonant frequency. This is caused by the frequency-dependent non-ideal current and voltage transfer gains and the parasitic resistances associated with the different nodes. All the mentioned non-idealities are discussed in detail in section 4. It is also found that the linear range (dynamic range) of the circuit is mostly affected by small supply variations; however, the filter performance is not adversely affected.
To further bring out the merits of the proposed filter, a comparative analysis of the single ABBsbased mixed mode filters is carried out. It can be inferred from Table 7 that except [40], no other filter can provide all five filter responses in all four modes of operation. The latest presented filter in [47] is not a truly mixed mode and also suffers from use of negative and double input signals for filter response realization. The designs in [14,27,40,48] suffer from passive component matching requirements. The design in [48] requires a change in circuit configuration for realizing different responses, which is impractical. Although the proposed filter consumes more power compared with a few other designs, the power consumption of the filter can be reduced by redesigning the VD-EXCCII at low supply voltage and reduced bias currents.  It can be concluded from the results that the characteristics of the AP filters have a slight imperfection, as all magnitude responses have a hump at the resonant frequency. This is caused by the frequency-dependent non-ideal current and voltage transfer gains and the parasitic resistances associated with the different nodes. All the mentioned non-idealities are discussed in detail in section 4. It is also found that the linear range (dynamic range) of the circuit is mostly affected by small supply variations; however, the filter performance is not adversely affected.
To further bring out the merits of the proposed filter, a comparative analysis of the single ABBsbased mixed mode filters is carried out. It can be inferred from Table 7 that except [40], no other filter can provide all five filter responses in all four modes of operation. The latest presented filter in [47] is not a truly mixed mode and also suffers from use of negative and double input signals for filter response realization. The designs in [14,27,40,48] suffer from passive component matching requirements. The design in [48] requires a change in circuit configuration for realizing different responses, which is impractical. Although the proposed filter consumes more power compared with a few other designs, the power consumption of the filter can be reduced by redesigning the VD-EXCCII at low supply voltage and reduced bias currents. It can be concluded from the results that the characteristics of the AP filters have a slight imperfection, as all magnitude responses have a hump at the resonant frequency. This is caused by the frequency-dependent non-ideal current and voltage transfer gains and the parasitic resistances associated with the different nodes. All the mentioned non-idealities are discussed in detail in Section 4. It is also found that the linear range (dynamic range) of the circuit is mostly affected by small supply variations; however, the filter performance is not adversely affected.
To further bring out the merits of the proposed filter, a comparative analysis of the single ABBs-based mixed mode filters is carried out. It can be inferred from Table 7 that except [40], no other filter can provide all five filter responses in all four modes of operation. The latest presented filter in [47] is not a truly mixed mode and also suffers from use of negative and double input signals for filter response realization. The designs in [14,27,40,48] suffer from passive component matching requirements. The design in [48] requires a change in circuit configuration for realizing different responses, which is impractical. Although the proposed filter consumes more power compared with a few other designs, the power consumption of the filter can be reduced by redesigning the VD-EXCCII at low supply voltage and reduced bias currents.

Filter Realization Using Macro Models of Commercially Available Integrated Circuits AD844 and LM13700
The proposed VD-EXCCII can be easily realized by commercially available ICs, the current feedback amplifier (AD844) and OTA (LM13700). The VD-EXCCII and the proposed filter circuit are realized using the PSpice macro model of the ICs to further test the feasibility of the proposed filter circuit. The setup for realizing the VD-EXCCII and the VM filter circuit depicted in Figure 3 is presented in Figure 24.

and LM13700
The proposed VD-EXCCII can be easily realized by commercially available ICs, the current feedback amplifier (AD844) and OTA (LM13700). The VD-EXCCII and the proposed filter circuit are realized using the PSpice macro model of the ICs to further test the feasibility of the proposed filter circuit. The setup for realizing the VD-EXCCII and the VM filter circuit depicted in Figure 3 is presented in Figure 24.
The OTA transconductance is fixed at 2 mS by selecting +VC = 10 V and R = 178.6 kΩ. The capacitors are selected equal to 1 nF and the resistors values are fixed as R = 1 kΩ and R = 500 Ω, resulting in = 225 kHz and = 0.707. The AC analysis results of the filter are presented in Figure   25. The measured frequency is found to be 220.4 kHz, which translates into 2% error. A time domain analysis is also carried out for the VM-BP configuration. A sinusoidal signal of 40 mVp-p at the 225 kHz frequency is applied at the input of the filter and the corresponding BP output is analyzed as shown in Figure 26, which establishes the correct functioning of the filter.  The OTA transconductance is fixed at 2 mS by selecting +V C = 10 V and R Bias = 178.6 kΩ. The capacitors are selected equal to 1 nF and the resistors values are fixed as R 1 = 1 kΩ and R 2 = 500 Ω, resulting in f 0 = 225 kHz and Q = 0.707. The AC analysis results of the filter are presented in Figure 25. The measured frequency is found to be 220.4 kHz, which translates into 2% error. A time domain analysis is also carried out for the VM-BP configuration. A sinusoidal signal of 40 mV p-p at the 225 kHz frequency is applied at the input of the filter and the corresponding BP output is analyzed as shown in Figure 26, which establishes the correct functioning of the filter.

Filter Realization Using Macro Models of Commercially Available Integrated Circuits AD844 and LM13700
The proposed VD-EXCCII can be easily realized by commercially available ICs, the current feedback amplifier (AD844) and OTA (LM13700). The VD-EXCCII and the proposed filter circuit are realized using the PSpice macro model of the ICs to further test the feasibility of the proposed filter circuit. The setup for realizing the VD-EXCCII and the VM filter circuit depicted in Figure 3 is presented in Figure 24.
The OTA transconductance is fixed at 2 mS by selecting +VC = 10 V and R = 178.6 kΩ. The capacitors are selected equal to 1 nF and the resistors values are fixed as R = 1 kΩ and R = 500 Ω, resulting in = 225 kHz and = 0.707. The AC analysis results of the filter are presented in Figure   25. The measured frequency is found to be 220.4 kHz, which translates into 2% error. A time domain analysis is also carried out for the VM-BP configuration. A sinusoidal signal of 40 mVp-p at the 225 kHz frequency is applied at the input of the filter and the corresponding BP output is analyzed as shown in Figure 26, which establishes the correct functioning of the filter.

Conclusions
This paper presents a new single VD-EXCCII-based electronically tunable mixed-mode filter structure. The filter employs only one VD-EXCCII, three resistors and two capacitors. The mixed mode filter enjoys inbuilt tunability and can realize all five filter responses in all four modes of operation (VM, CM, TAM and TIM). A CM SIMO filter can also be derived from the presented minimum component mixed mode MISO filter topology. The detailed theoretical analysis, non-ideal gain analysis and parasitic study are given. The VD-EXCCII is designed in Cadence Virtuoso software and extensive post-layout simulations are carried out to examine and validate the proposed filter in all four modes of operation. The proposed filter has all the advantages mentioned in Section 3. The filter is designed for a frequency of 8.0844 MHz with a ±1.25 V supply. The Monte Carlo analysis shows that the frequency deviation is within acceptable limits. Furthermore, the THD is within 5% for a considerable voltage/current input signal range. The power dissipation of the filter is found to be 5.76 mW. The VD-EXCCII and the mixed mode filter are also designed and tested using the models of commercially available ICs the AD844 and LM13700 in PSpice. The simulation results are found consistent with the theoretical predictions.

Acknowledgments:
We would like to thank the anonymous reviewers for their insightful comments and suggestions.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

Conclusions
This paper presents a new single VD-EXCCII-based electronically tunable mixed-mode filter structure. The filter employs only one VD-EXCCII, three resistors and two capacitors. The mixed mode filter enjoys inbuilt tunability and can realize all five filter responses in all four modes of operation (VM, CM, TAM and TIM). A CM SIMO filter can also be derived from the presented minimum component mixed mode MISO filter topology. The detailed theoretical analysis, non-ideal gain analysis and parasitic study are given. The VD-EXCCII is designed in Cadence Virtuoso software and extensive post-layout simulations are carried out to examine and validate the proposed filter in all four modes of operation. The proposed filter has all the advantages mentioned in Section 3. The filter is designed for a frequency of 8.0844 MHz with a ±1.25 V supply. The Monte Carlo analysis shows that the frequency deviation is within acceptable limits. Furthermore, the THD is within 5% for a considerable voltage/current input signal range. The power dissipation of the filter is found to be 5.76 mW. The VD-EXCCII and the mixed mode filter are also designed and tested using the models of commercially available ICs the AD844 and LM13700 in PSpice. The simulation results are found consistent with the theoretical predictions.

Abbreviations
The following abbreviations are used in this manuscript: α P/N , α P/N Frequency dependent non-ideal current gains α Vt Threshold voltage temperature coefficient β P/N Frequency dependent non-ideal voltage gains ε g m , ε g m Transconductance errors ε iP , ε iN Current tracking errors ε v(P,N) Voltage