The DM-NO GAAFET proposed in [16] shows the higher values of RF performance parameters such as transconductance, output conductance, early voltage, and transforming growth factor (TGF). Therefore, this device is very much suitable for higher frequency or RF applications. In this research paper, a common source amplifier is designed by using DM-NO GAAFET. Since unity gain frequency plays an important role to calculate how much amount of frequency signal can be applied at the gate input; hence, it has been calculated for the DM-NO GAAFET device.
5.1 RF performance of DM-NO GAAFET
The unity gain frequency is the frequency at which the short circuit gain of the device became unity [24]. Since miller capacitance (Cgs and Cgd) plays an essential role in the RF performance of GAAFET hence, it is necessary to have lower values of this miller capacitance for better performance. The unity gain frequency of GAAFET also depends on the values of transconductance and miller capacitance, as shown below in Eq. (30)
|
(30)
|
From Eq. (30), it can be observed that the unity gain can be higher if DM-NO GAAFET provides a higher value of transconductance and a lower value of miller capacitance. The values of miller capacitance of DM-NO GAAFET has been excerpted by applying a 1MHz frequency source at the input gate terminal, and VGS has been swept from 0 to 1V. The total gate capacitance Cgg is the summation of the intrinsic gate to source capacitance (Cgs) and intrinsic gate to drain capacitance (Cgd). Hence, Fig. 11 depicts the variation of Cgg at VDS = 0.3, 0.6, and 1V for DM-NO GAAFET, when VGS has increased from 0 to 1V.
From Fig. 11, it can be observed that as the VGS increases, it also improves the value of Cgg for all the values of VDS. However, the value of intrinsic Cgd has not dominated by the intrinsic Cgs for the GAAFET device; hence, the value of Cgg is very smaller or in order of 10-18. Therefore, this smaller value not only increases unity gain frequency but also helps to increase other RF performance parameters. Moreover, the smaller value of Cgg is observed at VDS = 1V, which is due to the higher drain current.
Fig. 12 shows the variation in unity gain frequency at VDS = 0.3, 0.6, and 1V for DM-NO GAAFET when VGS is swept from 0 to 1V. From Fig. 12, it can be seen that the maximum unity gain frequency is calculated by using Eq. 30 and is found to be 2.340 THz at VGS = 0.7V, due to the lower value of Cgg at VDS = 1V (as per Eq. (30)). Moreover, unity gain frequency is also higher at VDS = 0.3, and 0.6V which shows that the gate can easily be subjected to higher frequency signals, and the device will depict better performance even for lower VDS.
Moreover, transit time is also an important parameter that shows the time taken by electrons to cross the channel region [24]. The lower value of transit time shows the fastest device speed. The expression for transit time is given as
|
(31)
|
As per Eq. 31, the transit time of the device depends on the unity gain frequency; however, the unity gain frequency depends on the intrinsic capacitances; hence Cgg directly affects the transit time in DM-NO GAAFET. Fig. 13 shows the variation in transit time at VDS = 0.3, 0.6, and 1V. From Fig. 13, it can be observed that the transit time is lower at VDS of 1V, which is due to the higher unity gain frequency, higher drain current, and lower Cgg. Furthermore, transit times of 41.837, 127.31, and 345 ps have been observed at VDS of 0.3, 0.6, and 1V, respectively, which show the fastest speed of DM-NO GAAFET even at lower VDS.
5.2 Verilog A model based on look up table approach
The Verilog A model, which is formed by a look up table approach, has been used to perform SPICE circuit simulation with DM-NO GAAFET device [25]. Cadence virtuoso circuit simulator is considered for all circuit simulations. Firstly, the DM-NO GAAFET device has been simulated for different values of VGS and VDS. Furthermore, the transfer and capacitance characteristics are used by look up table based Verilog A model to capture DM-NO GAAFET device performance. Mainly, three tabular files, namely IdVg.tbl, Cgs.tbl and, Cgd.tbl are called by Verilog A model during circuit simulation. The DC analysis in circuit simulation takes place by using the tabular parameters given in the IdVg.tbl file, whereas the transient analysis takes place by using drain charge (use VGD and CGD), source charge (use VGS and CGS), and the terminal voltage of VGD and VGS. The Verilog A model behaves like the real device during circuit simulation and provides good performance for DM-NO GAAFET.
5.3 Common source amplifier by using DM-NO GAAFET
As DM-NO GAAFET shows better RF performance parameters hence, this device can be used for analog applications. A resistive load common source amplifier is designed by a DM-NO GAAFET device, which is important for analog integrated circuits. Fig. 14 depicts a resistive load common source amplifier circuit in which DM-NO GAAFET is used as the active device.
From Fig. 14, it can be seen that the sinusoidal input signal, which has to be amplified, will be applied at the gate terminal of DM-NO GAAFET, followed by the series resistance R2, which in general used to limits the current. Voffset is used in series with input sinusoidal signal, which adds with the ac signal. The load resistance RL and load capacitance CL is connected in parallel to observe the amplified gain. Moreover, an additional capacitance of C1 is also connected with drain and load resistance to provide isolation of DC components from the amplified output. Fig. 15 shows the equivalent circuit of resistive load common source amplifier in which DM-NO GAAFET device is replaced by small-signal model. If device output resistance is RO, from Fig. 15, it can be observed that the equivalent resistance (Req) at the output is the parallel combination of device output resistance and load resistance. For FET, the amplification factor Av depends on equivalent output resistance and device transconductance. The expression for gain is given as
|
(32)
|
From the above expression, it can be observed that the gain is directly affected by transconductance and equivalent output resistance; hence, to increase the gain in the amplifier, it is necessary to have a higher value of transconductance and equivalent output resistance. Since DM-NO GAAFET provides a higher drain current; therefore, the transconductance became high.
To evaluate the performance of the designed amplifier, DM-NO GAAFET based common source amplifier is compared with the MOSFET based amplifier. The UMC CMOS based 45nm NMOS is used as the active device in MOSFET based common source amplifier. The DM-NO GAAFET based amplifier and MOSFET based amplifier are compared for voltage transfer characteristics (VTC), gain margin, phase margin, and transient response to evaluate the performance. Fig. 16 depicts the VTC of DM-NO GAAFET based amplifier and MOSFET based amplifier. Moreover, from Fig. 16, it can be observed that the VTC is steeper for DM-NO GAAFET compare to MOSFET, which shows the higher value of output resistance for DM-NO GAAFET, so that the voltage gain can be increased (as per Eq. 32) [25]
The transient analysis for DM-NO GAAFET based common source amplifier and MOSFET based common source amplifier has been compared and plotted in Fig. 17. A sinusoidal signal with the operating frequency of 1GHz and 100mV peak to peak voltage has been applied. Fig. 17 shows that the output waveform has been amplified by 2.51 gain for DM-NO GAAFET based common source amplifier, whereas it has amplified by 1.50 gain for MOSFET based amplifier; hence this clearly shows that the DM-NO GAAFET device provides a good amount of amplification due to higher transconductance. Moreover, to perform the transient simulation, the load resistance is kept fixed at 1MΩ while load capacitance is kept fixed at 10pF. The resistance connected between drain and supply voltage is kept fixed at 25KΩ, and the input offset is considered at 1V. Since the amplifier is a very important building block of analog circuits hence, DM-NO GAAFET is a more suitable device for analog application even at higher frequencies.
Since gain margin is an important parameter for amplifier design, hence Fig. 18 shows the gain margin of DM-NO GAAFET and MOSFET based common source amplifier when the frequency is varying from 10Hz to 50GHz. As per the Fig. 18, the amplifiers provide constant gain till the 1GHz, and after that, the gain has started falling. Since the unity gain frequency is also an important parameter that is used to calculate bandwidth of the amplifier, hence the frequency at which gain is zero in dB is known as the bandwidth. Therefore, DM-NO GAAFET based common source amplifier provides higher bandwidth of 20GHz, which is also an advantage of the DM-NO GAAFET based common source amplifier.
The leakage power and active power are also analyzed for both amplifier circuits. The active power is the power consume by the amplifier during amplification due to switching of FET device at a higher frequency. Table 1 shows the performance parameter comparison for DM-NO GAAFET and MOSFET based common source amplifier. The leakage power is calculated by applying only input signal along with 1GHz frequency and 100mV peak to peak voltage. The rest of the sources have been removed during leakage power calculation. Since DM-NO GAAFET is a low DIBL device, therefore the leakage power for DM-NO GAAFET based amplifier is reduced by 46.07% as compared to the MOSFET based amplifier. If the input frequency is increased, it also increases active power dissipation. However, the average power is reduced by 80.05% for DM-NO GAAFET based amplifier even if the higher frequency is applied at the input. Therefore, the DM-NO GAAFET based common source amplifier is also a good choice from a power consumption perspective. This not only reduces the leakage and active power but also amplifies the input voltage up to 2.51 gain along with the higher bandwidth.