Another strategy is to choose an inductor pullup (L). This allows a higher impedance without any concern of creating any DC drop across the component. Ideally, the inductor should be chosen large enough so that the impedance is high relative to the load impedance and also be operating away from its self-resonant frequency. For instance, consider a 3.3-nH pullup inductor with a self-resonant frequency of 7 GHz driving a 25-Ω spectrum analyzer input. This inductor theoretically has j50-Ω input impedance around 2.4 GHz. At this frequency, this in parallel with load is about j35-Ω, which is a 3-dB power reduction. At 1.4 GHz, this inductor has impedance of about 29 Ω. This in parallel with the 50-Ω load has a magnitude of 25 Ω, which is the same result seen with the 50-Ω pullup. The main issue with the inductor pullup is that the impedance does not look nicely matched to the load.

Figure 8-3 Inductor Pullup

As the output impedance is not so nicely matched, but there is higher output power, it makes sense to use a resistive pad to get the best impedance control. A 6-dB pad (R1 = 18 Ω, R2 = 68 Ω) is likely more attenuation than necessary. A 3-dB or even 1-dB pad might suffice. Two AC-coupling capacitors are required before the pad. In Figure 8-4, one of them is placed by the resistor to ground to minimize the number of components in the high frequency path for lower loss.

Figure 8-4 Inductor Pullup With Pad

For the resistive pad, Table 8-2 shows some common values: