Why are bias tees necessary for broadband microwave and RF amplifiers?

Marki Microwave offers a wide variety of broadband driver, low noise, and low phase noise amplifiers and broadband RF bias tees. Some amplifiers have on-chip supply networks, while others require a bias tee. Why can't every amplifier use on chip biasing? In this note we discuss the limitations of MMIC technology and why bias tees are sometimes necessary. 

What is amplifier biasing?

The term 'amplifier biasing' is a little misleading. An amplifier consists of a nonlinear gain device (typically a transistor) and a passive network. The passive network has two functions:

1. Impedance matching and tuning of the gain device for the desired amplifier specs at the required frequency (typically low noise or power, or a balance of both)
2. Separating the DC power supply and the bias voltages from the input and output RF signals. The circuit elements that achieve this are blocking capacitors that isolate the bias and power supplies from the system and RF chokes that connect the power supply or bias voltage source to the transistor.

Bias voltages are low current/low power voltages that must be applied to the transistor so that it is operating at an appropriate point on the transimpedance curve. Amplifiers designed for high linearity will be biased in the middle of the linear region. Amplifiers designed for high efficiency will be biased at a lower point, consuming much less current but also sacrificing linearity. 

The important thing to note about these bias voltages is that they are low current, and therefore low power. The RF choke that supplies the bias voltage could be an inductor or a resistor, or it could be integrated into the matching network of the amplifier. Either way it is easy to integrate this circuitry on-chip. Typically the only off-chip components necessary for bias voltages are bypass capacitors and sometimes resistive networks to set the voltages from a single supply. 

Power supply voltages, in contrast, are the source of the DC power that a transistor uses to amplify a signal. More specifically, the transistor modulates the power supply current between the output and the ground, leading to amplification. The greater the power supply current, the greater the output power (subject to impedance matching and other considerations). If this current is applied through a resistive choke, it will generally lead to undesired power losses and poor power added efficiency. With few exceptions for very broadband distributed amplifiers (discussed below), power is supplied through an inductor.

Limitations of MMIC On-Chip Inductors

On-chip inductors, while advantageous for high-frequency applications due to their compact size and integration with other circuit components, encounter significant limitations when it comes to very low frequency applications:

1. Inductance Value Limitation: The primary challenge with on-chip inductors at low frequencies is their inability to achieve high enough inductance values. High inductance values are required at low frequencies to maintain proper circuit operation and performance. On-chip inductors are typically limited in size due to the constraints of semiconductor manufacturing processes, which in turn limits the maximum inductance they can achieve. This limits the practical low-end cutoff frequency to about 2 GHz. 

2. Quality Factor: On-chip inductors generally have lower quality factors compared to their off-chip counterparts. At low frequencies, a high quality factor is essential to minimize energy loss and maintain signal integrity. Lower quality factors result in higher resistive losses, which can significantly degrade the amplifier's performance.

3. Physical Space Constraints: Increasing the inductance value of an on-chip inductor typically requires more physical space, either through larger coil dimensions or more turns in the coil. However, space on a chip is at a premium, and expanding the size of inductors can be impractical when trying to maintain the compactness of integrated circuits.

4. Self Resonant Frequency: On chip inductors have a low self resonant frequency relative to off-chip inductors. This limits the bandwidth ratio of an amplifier using an on-chip inductor to around 10:1. This is even worse for inductors with wider traces (with higher current handling).

These limitations come from the construction of MMIC inductors, which are planar, printed circuits. Two metal layers are required to create the crossover required to exit the center of a typical spiral MMIC inductor. There is a tradeoff between the current handling capability of the line (limited by electromigration) and the Q of the inductor. Narrower lines lead to higher inductance values and higher Q with lower current handling. 

Necessity of External Inductors

Given these inherent limitations with on-chip inductors at low frequencies, external bias tees become not only beneficial but necessary for several reasons:

• Achieving Required Inductance: External bias tees allow for the use of larger inductors that can achieve the high inductance values necessary for low-frequency operation without the physical and performance limitations of on-chip inductors.

• Flexibility and Tunability: Using external components provides greater flexibility in trading off size and low end performance. An amplifier user can select the external inductor based on the size available and the required low frequency performance of the application. 

• Isolation and Stability: External inductors can help improve the isolation between the power supply and the signal path. They can provide a more robust solution for high reliability applications.

• Suppressed Resonances: The ferrites used in wirewound inductors act as electrical dampers, limiting the effect of resonances where the self resonant frequency of the inductors is present. 

These benefits also come from the construction of wirewound inductors, especially conical coils. A wirewound inductor is built with wire and a magnetic core. The wire can be selected for the required current handling capacity with low resistance. The magnetic core enhances the magnetic field inside the inductor, allowing for operation to MHz frequencies or below. 

Role of Blocking Capacitors

As mentioned above, blocking capacitors are required to isolate the DC voltage from the system. A capacitor will appear as an open at low frequencies, and the cutoff frequency is inversely proportional to the capacitor value. Similar to inductors, the capacitance value of printed MMIC capacitors is limited by the size available and process technology considerations. Therefore operation of amplifiers at very low frequencies requires off-chip blocking capacitors as well.

Combining the restrictions on MMIC inductors and capacitors, a rule of thumb is that amplifiers operating below 2 GHz and with supply currents above 150 mA are likely to require off chip inductors and capacitors. 

Value of the Bias Tee

For narrowband applications, it is common to combine a low cost surface mount inductor and capacitor to provide the power supply network for an amplifier. For broadband applications, however, this is a significant challenge due to the self resonant frequency (SRF) of the inductors used. The windings of an inductor have parasitic capacitance; when combined with the inductance this creates a resonator with a specific resonant frequency (the SRF). There are techniques to avoid the SRF such as using a conical coil, but these are complicated and difficult to prototype and troubleshoot on application circuit boards. 

Marki has solved this problem with our line of pre-assembled and tested surface mount bias tees. These bias tees are carefully designed for low, flat insertion loss, excellent return loss, and predictable performance in volume production. Our latest release is the BTM-0026PSM-2, which offers a smaller size and lower cost than competing conical coil solutions with superior electrical performance. This is achieved through careful design and thorough empirical investigation of potential inductors, capacitors, and circuit layout. 

Resistive Power Supply for Distributed Amplifiers

In the case of very broadband, low power, (and low noise) distributed amplifiers (such as the AMM-9024CH) the power supply can be supplied through a resistive RF choke. This is specifically possible with distributed amplifiers since a resistive termination of close to 50 ohms is required for distributed amplifiers. With the supply current of 45 mA, that means a voltage of 2.25V is dropped across the resistor, for a power loss of 100 mW. This tradeoff is made since the low frequency operation of the resistor is unlimited, allowing this amplifier to operate down as low as the off-chip blocking capacitor will allow.

Conclusion

In conclusion, while on-chip solutions offer integration and size benefits, their limitations at very low frequencies—particularly in terms of achieving sufficient inductance values and maintaining high quality factors—make external bias tees a critical component in certain applications. Fortunately Marki offers a full line of amplifiers with on-chip circuitry where possible and a full line of bias tees where it is not.