While the specifications provided in data sheets seem pretty straight forward, beware of some of the potential pitfalls along the way.

By Alan Rixon and Philip Gadd, Avago Technologies

There are many RFIC amplifiers available on the market, often categorized by the process they are made on, ranging from low cost silicon parts with wideband
Figure 1. Avago Technologies’ high-performance RFIC amplifiers support a wide-range of wireless applications including CDMA, WCDMA, GSM and WiMAX.
internally matched inputs to low noise gallium-arsenide parts with high linearity. Some parts are only DC tested while others are 100% RF tested by the manufacturer. Normally the guaranteed specification (min and max) will indicate which parameters are tested in production. While there are seemingly straightforward specifications given in the datasheets, the unwary should look out for one or two potential pitfalls.
Supply Voltage and Current
Check that the amplifier delivers the required performance over the full tolerance of the available supply voltage of your system, particularly when the supply voltage is 3V or less. Be aware that the current taken by an amplifier can increase under high RF input drive, which can be significant for Transmit driver chains (TX) and Local Oscillator buffering amplifiers (LO). The typical performance graphs given in the datasheet will often include data at voltages above and below nominal and either current or efficiency against RF power levels and will enable you to assess if your power supply circuit is able to meet the current needs of the RFIC over the operating conditions of your product. If the amplifier has a shutdown capability, make sure that the control requirements are clear and that the current taken when the part is shutdown is within the supply conditions of your system.
Temperature Range
The reliability of an RFIC is reduced at high temperatures which can be an issue for higher power dissipation parts. The junction-case or channel-case thermal resistance should be used to calculate the highest junction or channel temperature the RFIC will see, given the power dissipation in the part and the maximum case temperature. Care should be taken to follow any thermal mounting suggestions given in the datasheet.

Reliability Datasheets associated with each part can usually be found either posted on a supplier’s website or obtained by contacting them directly. This should give the necessary Mean Time To Failure (MTTF) information at different junction/channel temperatures.
Frequency Range
At first sight this seems straightforward. The specified operating frequency range of an amplifier must cover the frequency band of interest; however having too wide an operating frequency can have implications elsewhere in the system, usually leading to a need for better filters (which add costs and ca have their own performance impacts). In a receiver it is often not possible to put a high rejection, narrowband filter before the receiver Low Noise Amplifier (LNA) due to the higher insertion loss of a high rejection filter adding to system noise figure. This means that out-of-band signals, and sometimes wideband signals are able to reach the LNA due to the poorer rejection of a lower loss input receive filter. These unwanted signals may lead to unwelcome in-band intermodulation products or reduced receiver sensitivity due to high level blocking signals. For a TX driver, a wideband RFIC will amplify any unwanted spurious signals or out-of-band noise. This could lead to problems meeting overall spectral emission requirements if not considered when designing the PA and TX filtering and can result in parts failing regulatory or type approval specifications.

Not all RFIC amplifiers are fully matched and therefore when short-listing devices, any available application notes or design aid material should be consulted to ensure that when externally matching is required, the required performance is obtainable over the system frequency range. One should also make sure that the costs of any external matching components are included in the bill of materials roll-up, particularly when comparing to fully matched amplifiers.
Gain (G)
It is often easy to focus on typical performance figures given in the datasheet and to neglect the minimum and maximum specifications. Most RFIC amplifiers will have a gain window of a few dB and this can lead to serious system headaches if not taken into account. Parts falling at the low end of the gain distribution may mean that overall system gain is not met or that system NF is compromised. Higher gain parts may lower system linearity by delivering too much power to subsequent stages of the circuit. In general, gain will fall as temperature increases and again the data sheet performance graphs should be checked to ensure suitability.
Noise Figure (NF)
The importance of selecting a part with the appropriate NF is well understood for those designing receiver systems; however, it is sometimes overlooked when selecting drivers in the TX chain. The complex modulation schemes and wide dynamic range needs of modern radio systems often means that TX drivers with low NF are needed to minimize increases in the noise floor. When calculating system performance, take into account that Noise Figure increases with temperature.
1dB Compression Point (P1dB)
Remember that P1dB is defined as the input power (IP1dB) or more usually the output power (OP1dB) at which the gain has reduced by 1dB. Therefore they are linked by IP1dB = OP1dB - (G - 1).

If the part is being used as an LO buffer, the amplifier will often be required to operate in saturation. Information on performance in this mode may be available in the datasheet or applications material however careful evaluation of the part is recommended as the part is no longer operating in the well defined small-signal area. For example, the LO port of a passive mixer may present a poor and varying load impedance to the part, making it difficult to drive or causing instability. This instability may be at very high or very low frequencies away from the operating band and can be easy to miss. The loss associated with adding a low value attenuator between the buffer amplifier and mixer may be very worthwhile to clean up the wideband load seen by the amplifier. Remember that the return loss is improved by twice the amount of the attenuator so that a 2 dB pad will improve the return loss by 4 dB.
Third Order Intercept (IP3)
Modern systems have increased the need for high linearity, wide dynamic range RFIC amplifiers. Conventionally third order intercept specifications have been used to benchmark amplifier linearity performance. It is the theoretical point at which the output power level of two tones at frequencies f1 and f2 is equal to that of the two third order intermodulation products (2f2-f1 and 2f1-f2) produced by the amplifier. For receivers, the input power level at which this occurs can therefore be calculated from IIP3 = OIP3 – G. For every 1dB increase in power level of the two tones, the third order products increase by 3 dB. This means that the OIP3 can be calculated from spectrum analyzer measurements of the tone and third order intermodulation power levels OIP3 = Ptone + ((Δtone-intermod)/2). So with tone power levels of 22 dBm and the third order products 30 dBc below this, the OIP3 is 37 dBm.

Likewise if the OIP3 is specified, the difference between the tone power level and third order products can be predicted using Δtone-intermod= (OIP3 – Ptone) × 2.

Modern systems have generated their own specific measures of linearity such as ACLR (Adjacent Channel Leakage Power Ratio) for WCDMA or EVM (Error Vector Magnitude) for 802.11 WiFi standards. If not already available in the datasheet or applications material, ask the supplier if they can provide measured data taken using a specific standard.

Historically a 10 dB difference between P1 dB and OIP3 would have been considered adequate linearity performance. Modern RFICs are now capable of 15 dB to 20 db difference between P1dB and OIP3 and when selecting a linear amplifier one should look for the largest delta whilst meeting the other specification considerations mentioned in this article.
Design Tools
There is often a wide range of useful design tools available from the amplifier supplier to help to select the right part and design with it. As a minimum, s-parameter files should be downloadable in .s2p format to run directly in linear simulators. There may be more s-parameter files available on a supplier’s website representing a wider range of DC bias conditions than indicated on the datasheet. Non-linear models may also be available however these are normally specific to a particular simulator such as ADS or Microwave Office. Finally demo boards can be requested from the supplier so that specific measurements can be made under your system conditions.

Avago offers a simple free tool on its website called AppCad which contains a series of useful calculators for many common electrical calculations.

About the Authors

Alan Rixon is director of applications engineering for Avago Technologies and Philip Gadd is senior director of marketing for Avago. Both work in Avago’s Wireless Semiconductor Division;