• Keine Ergebnisse gefunden

84 5. VHF and UHF Broadband PA

η[%]

0 20 40 60 80 100

VSWR Phase [deg.]

0 30 60 90 120 150 180

VSWR=1.5 VSWR=2.0 VSWR=2.5 VSWR=3.0

@f=225 MHz

(a)

η[%]

0 20 40 60 80 100

VSWR Phase [deg.]

0 30 60 90 120 150 180

VSWR=1.5 VSWR=2.0 VSWR=2.5 VSWR=3.0

@f=312 MHz

(b)

η[%]

0 20 40 60 80 100

VSWR Phase [deg.]

0 30 60 90 120 150 180

VSWR=1.5 VSWR=2.0 VSWR=2.5 VSWR=3.0

@f=400 MHz

(c)

Figure 5.29:Simulated (solid) and measured (dashed) for the drain efficiency with different VSWR and phase values for a) 225 MHz b) 312 MHz and c) 400 MHz.

5.3 UHF Broadband Class-E PA 85

Pout Id0

PAE

Zopt

Zdesigned

(a)

Pout Id0

PAE

Zopt

Zdesigned

(b)

Pout

Id0 PAE

Zopt

Zdesigned

(c)

Figure 5.30:De-embedded impedance contour for sweeping VSWR value and phase (star black) imposed on the ideal output power contours (red) with 0.5 dB step, drain efficiency contours (green) with 2 % step and DC current contours (blue) with 0.6 mA step, The black line is the designed load impedances for a) 225 MHz b) 312 MHz and c) 400 MHz.

bandwidth designed in 5.2, i.e., 56 % at 312.5 MHz. This might make the impedance matching network easier to implement with fewer numbers of poles in the output filter.

5.3.1 Matching Network Design

Load/source-pull simulation was performed using Agilent’s Advanced Design System (ADS).

The optimum load impedances over the targeted band are shown in Fig 5.31.a. This figure shows that the magnitude of load impedance can be expressed as:

ZLopt =

{ RLopt(1+j1.33) , @f = f0

∞ , @f =n f0, n is an integer > 1 (5.18)

86 5. VHF and UHF Broadband PA

|ZL|[Ω],∠ZL[deg.]

0 20 40 60

Freq. [MHz]

600 700 800 900 1000

ZL

|ZL|

(a)

Pout[dBm],Gain[dB]

10 20 30 40 50

Efficiency[%]

70 75 80 85 90

Freq. [MHz]

600 700 800 900 1000

PAE η Pout Gain

(b)

Figure 5.31: Ideal load/source-pull a) load impedance magnitude (blue) and phase (red), and b) ideal performances for output power (red), gain (blue), drain efficiency (green) and power added efficiency (black).

WhileRLopt is the optimum real load impedance. The best function that fits the figure shown in Fig. 5.31.a is a linear function as in (5.13).

RLopt =−0.0193f+31.3113 (5.19) WhereRLopt is the optimum real load impedance. The best function that fits the figure shown in Fig. 5.31.a is a linear function as in (5.13). As can be seen, the optimum impedance has a constant phase equal to 53, where the optimum impedance magnitude is a linear function with negative slope over the bandwidth. This is the same behaviour obtained from the amplifier designed previously in section 5.2, equations (5.12) and (5.13). In both designs, the optimum impedance phase has the same value. However, the magnitude slope is lower than the optimum impedances obtained in section 5.2. This was expected because the load impedance should be inversely proportional to the output power and the operating frequency, follow (5.19).

ZLopt ∝ 1

Poutf (5.20)

Since the load impedances behave similar to the designed amplifier in the VHF band, The PA network topology in Fig. 5.13 will be used in this design also. It showed good results for this kind of power amplifier. It was also easy to implement the amplifier.

The ideal output performance is shown in Fig. 5.31.b, where the drain efficiency and power added efficiency exceed 79 % and 76 %, respectively. The output power is 46 dBm (40 W) with 14.4 dB over the entire band. The design of broadband Class-E PA requires at the output a wideband band-pass filter instead of a series resonator. This filter must pass the fundamental frequencies and reject all higher order harmonics as an open circuit termination.

The filter type is an important issue in this design. As stated previously, a broadband Class-E PA requires a constant phase and magnitude of constant slope. Hence, a Butterworth filter was chosen. It is well known that a Butterworth filter has a flat gain and constant phase. However, a

5.3 UHF Broadband Class-E PA 87 Class-E PA requires high rejection at the harmonics.

For the targeted band, the first harmonic (i.e.; 1200 MHz=2×600 MHz) is very close to the band edge frequency (i.e.; 1000 MHz). A roll-off of 60 dB/decade (i.e.; 3-poles) is implemented.

Moreover, to minimize the insertion loss of the filter, the 3 dB bandwidth is designed for the band between 470 MHz and 1.160 GHz.

The circuit topology for the used bandpass filter is a T network, where the T network behaves as an open impedance for the stop-band (i.e.; harmonics). Fig. 5.32.a shows the filter topology used in this design. The transmission loss of the filter is shown in Fig. 5.32.b. It is seen from this figure that the loss in the pass-band is less than 0.3 dB. The minimum harmonic suppression, occurs at 1.2 GHz, is equal to 4.3 dB and increases as the frequency increases. This is a good suppression value for the harmonic.

The standard filter topology as discussed in the literature is designed using a 50Ωtermination at both sides. For the presented work, usually the transistor output impedance is not 50Ω. On the other hand, the values of the elements in the filter circuit depend only on the geometric mean centre frequency (the square root of the product of the two ends of the bandwidth), as stated previously. Hence, a matching network to 50Ωimpedance over the entire band plus capacitive impedance that is equal to−jXCF1 in Fig. 5.32.a was designed. The output circuit network, which combines the filter and the matching network, is shown in Fig. 5.33. Table 5.4 presents element values of the final version of the matching network and the filter. In the output matching network, an extra element is added in the design compared to the network in Fig. 5.16.

The input matching network topology, DC-feed network and stability circuit is designed here similar to the amplifier in section 5.2. However, the stability circuit uses only one parallel network in series with the input matching network and the transistor.

5.3.2 Realization

The designed amplifier was realized using Air Core inductors from coilcraft and multilayer capacitors from ATC. The parallel networkLF2CF2shown in the filter in Fig. 5.32.a is replaced

− + Port 1

CF1 LF1

CF2 LF2

CF3 LF3

50Ω + Vo

Io(t)

(a)

S21[dB]

-50 -40 -30 -20 -10 0

Freq. [GHz]

0 0.5 1 1.5 2 2.5 3

S21=-0.04 dB @ 0.6 GHz S21=-0.30 dB @ 1.0 GHz

2ndHarmonic suppression for 1.2 GHz=4.3 dB

(b)

Figure 5.32:a) Filter topology used in output matching network showing the capacitor dashed box that will be replaced, and b) the filter transmission loss over the targeted band and the harmonic.

88 5. VHF and UHF Broadband PA with a short stubλ/4. From [23], the inductor, capacitor and total parasitic loss values of anLC network can be replaced using:

L= 4Z0

ω0π (5.21)

C= π

0Z0 (5.22)

R= Z0

αl (5.23)

whereαis the losses of the transmission line,l is the line length equal toλ/4 at geometric centre frequencyω0andZ0is the characteristic impedance of the transmission line.

The DC-feed here is not connected in series to the LM1 in Fig. 5.33 as with the previous design, it is connected in a separate path closer to the transistor. This was chosen for better control of the final matching network and to add more harmonic control for the output impedance. It is important to add a DC-block before adding the rest of the matching network. The final amplifier is implemented on Roger’s substrate withεr=3.38 and a thickness of 0.51 mm, Fig. 5.34.

The drain voltage and current of the designed PA at the centre frequency are shown in Fig. 5.35. This figure presents a high frequency operation of Class-E PA. It is clearly seen that the drain current is below the zero level which can never occur with a FET transistor. This negative region of the drain current is the region where the output FET capacitor discharges its current to the load during the OFF-region.

Pin Q

Id

CM1

LM1 LM2

LM3 LF1

CF2 LF2

CF3 LF3

50Ω + Vo

Io(t) IDC

Vdd Z= (50+jXCF1)Ω

Figure 5.33:Full output matching network including the filter used in the ideal design step.

Table 5.4:Ideal lumped elements values used in the matching network shown in Fig. 5.33.

Lumped element LM1 CM1 LM2 LM3 LF1 CF2 LF2 CF3 LF3

Value 4.4 nH 3.1 pF 51.7 nH 6.4 nH 10.6 nH 9.2 pF 5.0 nH 4.0 pF 10.6 nH

5.3 UHF Broadband Class-E PA 89

Input Output

Figure 5.34:Photo of the fabricated UHF broad-band Class-E power amplifier using 30 W GaN HEMT from Eudyna.

5.3.3 Measurements

In this section, small-signal and large-signal performances were measured using Fig. 4.7. How-ever, the measurements of power amplifier degradation is not a goal of verification here.

The small signal gain was measured at an input level of -30 dBm. Fig. 5.36 shows that the maximum gain is 19 dB with 1.2 dB flatness. The minimum measured small-signal harmonic suppression is 23.6 dB. From the figure, it is observed that the small signal gain behaves rather similar to the filter designed in Fig. 5.31.

5.3.3.1 Large Signal Performance

The large signal performance was measured from 600 MHz to 1000 MHz with step of 100 MHz.

The measurement for 800 MHz is shown in Fig. 5.37. The maximum output power was 46.9 dBm at 4 dB compression point, i.e., 12.2 dB gain. The drain efficiency is 75.4 % where the power added efficiency for this operating point is 75.4 %.

The performance across the targeted band is shown in Fig. 5.38. The power over all the bandwidth is 46.1 dBm±0.8 dB. The measured large signal gain is 11.52 dB over the entire band.

Vd[V]

0 50 100 150

Id[A]

-1 1 3 5

t [nS]

0.0 0.5 1.0 1.5 2.0 2.5

Vd Id

Figure 5.35:Simulation of final stage for the current (red) and voltage (blue) waveforms. The dashed lines shows the zero level for the current, where below the dashed line is the time for the output capacitorCdsdischarging and above the line is the charging period of the output capacitorCds.

90 5. VHF and UHF Broadband PA

Gainss[dB]

-30 -20 -10 0 10 20

Freq. [GHz]

0.2 0.4 0.6 0.8 1.0 1.2 1.4

-4.6

Figure 5.36:Measured small signal gain across the bandwidth with -30 dBm showing the gain for the second harmonic of the lowest frequency.

Over this band, the measured drain efficiency and measured power added efficiency are more than 66 % and more than 62 %. Table 5.5 summarizes the measured results in Fig. 5.38.

For the bandwidth between 600 MHz to 800 MHz, the amplifier delivers 46.3 dBm output power with±0.53 dB flatness with 11.4 dB gain. The drain efficiency in this band is more than 80.0 % where the power added efficiency is more than 75 %. The fractional bandwidth is 28.6 % centred at 700 MHz.