Libero

Class 1 Power Amplifier for BluetoothTM Applications

 

 

 

5.1      THE BFP450 AS A BLUETOOTH CLASS 1 POWER AMPLIFIER


 

In this section we will show that with only one transistor a small sized power amplifier is designed, offering high gain with 67% total efficiency and single 2.75V supply voltage. Biasing is done by a circuit around a NPN transistor BC849, which gives  better stability behaviour versus temperature and a reduction of DC current gain distribution problems. Thanks to the low components count and simple matching networks, the entire PA, including bias part, only measures 10x10 mm.

 


The power amplifier consists of the Siemens SIEGET® BFP450 wideband transistor, operating in class-AB (see Figure 5.1 and Table 5.1). The BFP450 has two emitter-leads which have to be carefully grounded to ensure stable operation and performance according to the specifications. The PCB layout (see Figure 5.2) of the amplifier results in an emitter-to-ground inductance of 130 pH (typical value).

 

 

5.2      DESIGN MODEL

 

The solver software used is Ansoft DesignerTM SV ver. 1.0.

The model is shown in Figure 5.3, as we discussed to in chapters 2 and 3 the transistor is completely characterized with S parameters and thanks these parameters is possible to find optimum source and load impedances either for simultaneous conjugate matching.

 


Thanks to BFP450 simulation model given by INFINEON and with the help of Smith Tool of Ansoft Designer SV has been possible to find optimal and very light input/output matching networks, as shown in Figure 5.1 and Table 5.1.

 


Figure 5.4 shows the source (GMS) and load (GML) reflection coefficients found by simulation. Now, all that remains is to surround the transistor with components that provide it with source and load impedances “look like” GMS and GML.

For the input/output matching network design we used Smith Tool of Ansoft Designer SV. Figure 5.5 shows the input matching network design. The object of the design is to force the 50-ohm source to present a reflection coefficient of  . With GMS

 


plotted as shown, the corresponding desidered and normalized impedance is read directly from the chart as ZS = 0.06 – j0.18 ohm. Remember, this is a normalized impedance because the chart has been normalized to 50 ohms. The actual represented by GMS is equal to 50(0.06 – j0.18) = 3 – j9 ohms. To force the 50-ohm source to actually appear as a 3 – j9 ohms impedance to the transistor, we only add a shunt and a series reactive component as shown on the chart of Figure 5.5.

 


Of course, that is very simple design, indeed, it does not consider the real models of components and the PCB. COILCRAFT® and MURATA® provide, respectively, S parameters of inductors and equivalent circuit (LCR) of capacitors. With these simulation models and with Trasmission Line Model of Ansoft Designer SV is possible to design matching networks with results close to the reality.

Well, the second step is to find the matching networks with real models. That was done with the help of Smith Tool of Ansoft Designer SV and testing more values of components in simulation. The result is shown in Figure 5.6.

 

 

5.3      THE POWER AMPLIFIER ANALYSIS

 

 


Thanks to the use of Ansoft Designer tools we found good matching networks, the S11 and S22, at 2.45 GHz, are shown in Figure 5.6.

The next step is to analyse the model shown in Figure 5.3 to have some report about gain, stability factor, S11, S22, forward gain and input output return loss.

Figure 5.7 shows the gain from 0.5 to 3 GHz. At 2.45 GHz we have a gain equal to 11.93 dB. The fact that the gain of PA at low frequency is higher than high frequency does not amaze because the gain of BFP450 decreases when the frequency increases. At 2.45 GHz we have the highest gain possible at that frequency.

The stability factor at varius frequency is shown in Figure 5.8. In bluetooth band, K is greater than 1, then the PA is unconditionally stable. From 600 to 650 MHz, K is less than 1, therefore the PA is potentially unstable and will most likely oscillate. At 2.45 GHz K is equal to 1.19.

 


Figure 5.9 shows the forward gain in the range of 0.5 to 3 GHz. As well as expected at 2.45 GHz we have the maximum (-17.29 dBm).

Figure 5.10 shows the input/output return loss. Also in this case, we have the best value at 2.45 GHz, indeed the input/output ports are tuned to 50 ohms.

Finally, Figure 5.11 shows S11 and S22 at various frequencies. Obviously, they are close to 1 at 2.45 GHz. When we move from that frequency, we lose the optimum matching (50 ohms).

 


 

5.4      THE BFP450 POWER AMPLIFIER MEASUREMENTS

 

 

Measurements were performed using a signal generetor, a spectrum analyzer and a network analyzer. The power meter was used to calibrate the cables. A photo of the test setup is shown on Figure 5.12, while Figure 5.13 shows a photo of class 1 power amplifier.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 


 

 


Figure 5.1 shows the schematic with optimum matching networks. The schematic is

 


complete of power control and PA enable. As the START499 PA, the measurements was done under a determinate range of temperature, power supply and input power. Despite the design was done for a biasing current equal to 12 mA, also other current values was investigated. Of course, the best results were found with Icq = 12 mA.

 


Figure 5.14 shows the output power as function of input power at varius Icq. We can notice that with a biasing current equal to 8, 12 and 20mA we have similar results, nevertheless, as shown in Figure 5.15, with different efficiency values.

 


Also in this case, the efficiency is high with low biasing currents. But, the efficiency with Icq = 12 mA is around 70 %, higher than with START499.

At every biasing current, the maximum 1 dB compression point value is around 21 dBm.

 


Figure 5.16 shows gain, output power, and efficiency versus input power at 27 °C. At high input power we have highest efficiency (70 %) and a good output power (20 dBm), but the gain decreases (10 dB).

Figure 5.17 shows output power and efficiency versus input power at various temperature and Vcc. As we can notice, with low temperature, the gain is higher than high temperature, like the START499, but, the efficiency is worse.

 


There are not essential differences at varius Vcc, but, the RF performance depends, heavily, on temperature.

At last, Figure 5.18 shows S11 and S22 in bluetooth band, notice the marker at 2.45 GHz. Comparing these results with those got from linear analysis, we can see analogy.

 

 

5.5      THE BFP450 POWER AMPLIFIERS DISCUSSION

 

 

Amplifier tune-up is accomplished by adjusting matching networks for maximum output power with minimum collector current. The amplifier tunes in bluetooth band
while mantaining as S11 and S22 a value close to 1 (Figure 5.18). Note from Figure 5.7 that the bandwidth is further 100 MHz with the amplifier tuned to a center frequency of 2.45 GHz.

We designed the Class1 power amplifier to be most efficient possible PA in the bluetooth broadcast marketplace with a cost lowest.

The exceptional efficiency (70 %) of PA gives a competitive advantage. It can be used in battery powered applications where the power consumption is a key factor.

The measurements were focused on characterization of the bias and temperature dependance. A comparison of gain, output power and efficiency measured at room temperature and at high temperature shows large differences as reported in Figure 5.17. At temperature below zero, with any voltage supply, the efficiency degraded. These measurements suggest the importance of temperature and voltage supply effect on the class-AB amplifier characterization. As well as the START499, changing the base resistor of bias transistor (BC849) a better temperature stabilization is possible, but this causes efficiency degradation.

To have the best RF performance from BFP450, it must work at high output power.

Finally, with BFP450 power amplifier the challenge was won, we have a small size amplifier with a light BOM,  with low power consumption and with very low cost.

 

 

 

 

 



HOME