Many considerations go into the deployment and operation of long-haul microwave networks, such as capital expenditures for equipment, technology, planning, engineering, and logistics. However, the total cost of ownership (TCO) often dictates the efficacy of deployment and long-term operation.
When talking TCO, the conversation usually turns to power consumption. Power consumption not only necessitates significant outlays for electricity, it also creates heat dissipation that requires cooling, which further increases the footprint and the electricity bill.
Advanced microwave-equipment manufacturers employ many techniques to decrease the power consumption of their microwave equipment. One particular innovative technology seems to be separating from the pack, though. It reduces the power consumption of long-haul radio amplifiers and decreases their heat dissipation—even eliminating the need for cooling equipment. Called “dynamic biasing,” it not only helps shrink TCO, it also makes an important contribution to a greener environment.
Power Consumption
As expectations of higher speeds, more volume, and immediate responsiveness drive the demand for higher network capacities, operators need to look for smarter, more cost-efficient wireless solutions. Today’s transmission equipment must provide very high spectral efficiency and system gain to adapt to variations in traffic load and weather conditions while maintaining the best quality of experience. At the same time, it must minimize overall operational costs.
Long-haul microwave links run on electricity and large multichannel systems can draw power in the kilowatt range, which makes them (often prohibitively) expensive. Certain areas only have a sporadic or even non-existent supply of electricity. Moreover, some installations don’t have access to commercial power, so they rely on solar energy or diesel generators for their electricity supply.
Even in areas where electricity is abundant, it constitutes a considerable component of operating expenditure (OPEX)—as much as 15%—that bites significantly into operator profit margins. In all cases, savings in power consumption can contribute to the feasibility and attractiveness of operating a high-capacity microwave network.
If made efficient enough, high-power radios don’t require cooling. In addition, if they’re installed in split-mount configurations, where radio units are mounted in the tower close to the antenna, it could completely eliminate the heat dissipation. An additional benefit is increased system gain due to the removal of RF losses and expensive waveguides. In many cases, the added power efficiency makes the difference in whether or not an installation is viable to deploy.
Improving Class-A Amp Power Efficiency
Wireless-transmission infrastructure is used to establish high-capacity, point-to-point connections between sites in mobile backhaul and other telecommunication networks. However, limited availability of spectrum resources mandates the need for high spectral efficiency. The trend in the industry is a move toward 1024QAM modulation, which calls for highly linear power amplifiers.
It’s possible to make amplifiers with close to 100% efficiency. Such devices aren’t viable for modulated signals in modern communication equipment, though, because they can’t be operated with good linearity over the required power range. Amplifiers employed in modern communication equipment traditionally operate in Class A, albeit with poor power-added efficiency (PAE).
To address the problem of linear added power (in proportion to capacity), Ceragon developed the “dynamic biasing” power-amplifier scheme. With this technology, amplifier bias voltages are functions of the signal envelope level. Dynamic biasing adjusts the amplifier bias per symbol (signal envelope) transmitted to reduce power consumption.
Class-A Amp Operation
Class-A amplifiers are a natural choice for applications with strict linearity requirements, since they feature the best linearity properties among all amplifier types. However, Class-A amplifiers are inefficient.
The amplifier function is best described by the incoming and outgoing signals along with the added power (Fig. 1). The supplied power is the sum of PIN and PADD, while the output power is POUT. PADD represents the dc supply voltage. By energy conservation, the power is dissipated in the amplifier and causes undesirable heat:
PDISS = PIN + PADD – POUT
PAE is a figure of merit for amplifiers—the higher the better:
PAE = (POUT – PIN)/PADD
The upper theoretical limit of POUT/PADD for a Class-A amplifier is 50%. This assumes that the bias conditions are optimized for the actual output power level.
However, average PAE becomes very low when applying a modulated signal. That’s because PADD is continuously at least two times the maximum POUT, while the actual signal power level shows great variation. The peak-to-average power ratio (PAPR) of modulated signals can easily rise toward 10 dB. Subsequently, a Class-A amplifier’s average efficiency reduces by the same amount; thus, it’s expected that maximum average PAE will fall in the range of 5%. In actuality, though, a “practical” Class-A amplifier has been shown to have average PAE in the 2% range (Fig. 2).
Linearization can improve the efficiency for a Class-A-designed power amplifier to the 3% range by slightly overdriving the output signal. Dissipated power would drop to 4 W if the efficiency was increased to 20%. Enhancing the efficiency of amplifiers with Class-A linearity properties is a very attractive approach for wireless radios. As a result, it’s important to have a good grasp of Class-A operating conditions.
An amplifier runs in Class A if the output-signal amplitude is a linear function of the input-signal amplitude across the entire signal swing. This ideal situation requires that the current in the amplifier device neither saturates nor goes through zero.
Traditional implementations of Class-A amplifiers integrate constant bias-supply circuitry, which is the general rule for all amplifier types. Other amplifiers, like Class B, change power demand as a function of the actual signal level, which makes them more efficient but nonlinear.
The Class-A amplifier may be tuned toward 50% efficiency at any signal power level. Therefore, if the bias conditions can be changed as a function of power level, there’s a good possibility of dramatically improving efficiency while still operating in a Class-A-like mode.
Constant bias conditions aren’t necessary to ensure linearity in the amplifier. But if they change, they should do so as a function of the input signal. If the same input signal amplitude appears at different times, then the output signal should also be the same. If the bias signal varies independently, this linearity would generally not be possible.
Narrowband, almost linear, amplifiers can be approximated as frequency-independent transfer functions sensitive only to the envelope amplitude level. Designs where the bias conditions are functions of amplitude level have the potential to deliver more efficient, Class-A-like amplifier solutions. The potential of such concepts indicates that the maximum PAE could reach as high as 24%, compared to 3% for Class-A.
Dynamic Biasing Boosts Amp Efficiency
In dynamic biasing, amplifier bias voltages are functions of the signal envelope level that in practical implementations increase the amplifier stage’s efficiency from 3% to 10%. Applying dynamic bias to a field-effect transistor (FET) amplifier can be illustrated by relating the time development of a modulated signal with the corresponding bias point and output voltage and current swing (Fig. 3).
Operation of the FET amplifier can be viewed as consistent with Class A if the bias point allows the signal amplitude to be inside the indicated acceptable region. The acceptable region extends along the load line between saturation and pinch-off.
The upper part of Figure 3 shows the time development of the signal amplitude for the high-frequency modulated signal. The lower part shows current-voltage curves for a real FET amplifier with a superimposed illustration. The illustration depicts an acceptable region for Class-A operation together with examples of suitable bias points and actual signal swing corresponding to different signal envelopes.
In the idealized case, achieving the highest PAE for a given amplitude requires the bias point to be located where the signal swing amplitude reaches both border lines of the acceptable region. The corresponding bias currents and voltages would then be linear functions of the signal amplitude. Real amplifiers have nonlinear borders for the acceptable region and gain variations inside the region. Practical bias functions should match the properties of the actual amplifier.
This approach allows for gradual PAE improvements, because the family of acceptable bias functions spans from constant values with known Class-A linearity to some optimal functions that provide sufficient linearity and maximum PAE. In fact, Ceragon was able to make significant improvements by adopting this approach.
A point-to-point microwave radio unit that generates 1 W of output power (about 30 dBm) typically consumes 70 to 80 W—approximately half of which is used in the power amplifier. Obviously, then, improving the efficiency of amplifiers with Class-A linearity properties becomes a very attractive approach for microwave radios.
In this vein, the dynamic-bias approach can be beneficial. It requires the power supply to the amplifier work to with sufficient speed and good efficiency, since the total power-efficiency calculation must include added power dissipation in the supply circuit.
The symbol (modulation) rate of long-haul microwave radios is typically about 24 Msymbols/s (24 MHz) for a 28-MHz RF spectrum, up to about 46 Msymbols/s for a 56-MHz RF spectrum. Overall efficiency gain is roughly given by multiplying the amplifier PAE by the power supply’s efficiency relative to a traditional, fixed supply. With this supply efficiency, the upper limit for overall amplifier efficiency with dynamic biasing equates to about 20% (Fig. 4). To implement modulated supply bias, the signal amplitude envelope is derived from the modulator and feeds the amplified dynamic bias control circuit (Fig. 5).
Power Efficiency And The Environment
Dynamic-biasing technology essentially triples the efficiency of the amplifier stage from 3% to 10%. Not only does dynamic biasing educe substantial power savings, it also makes compact outdoor-unit (ODU) implementations possible without the use of cooling equipment (Fig. 6). For microwave network operators, this significantly cuts both power and expense.
Beyond cost, dynamic biasing contributes to more efficient use of energy, translating into lower carbon-dioxide (CO2) emissions. Worldwide power consumption among networks produces more than 250 million tons of CO2 emissions annually—the equivalent of 50 million automobiles. Dynamic biasing can become a dynamic tool in creating more power-efficient networks and a cleaner environment.