8.1.1.1.1 Key Characteristics of RF Power Detectors

Power detectors are used to accurately measure the power of a signal inside the application. The attainable accuracy of the measurement is therefore dependent upon the accuracy and predictability of the detector transfer function from the RF input power to the DC output voltage.

Certain key characteristics determine the accuracy of RF detectors and they are classified accordingly:

• Temperature Stability
• Dynamic Range
• Waveform Dependency
• Transfer Shape

Generally, the transfer function of RF power detectors is slightly temperature dependent. This temperature drift reduces the accuracy of the power measurement, because most applications are calibrated at room temperature. In such systems, the temperature drift significantly contributes to the overall system power measurement error. The temperature stability of the transfer function differs for the various types of power detectors. Generally, power detectors that contain only one or few semiconductor devices (diodes, transistors) operating at RF frequencies attain the best temperature stability.

The dynamic range of a power detector is the input power range for which it creates an accurately reproducible output signal. What is considered accurate is determined by the applied criterion for the detector accuracy; the detector dynamic range is thus always associated with certain power measurement accuracy. This accuracy is usually expressed as the deviation of its transfer function from a certain predefined relationship, such as linear in dB for LOG detectors and square-law transfer (from input RF voltage to DC output voltage) for mean-square detectors. For LOG-detectors, the dynamic range is often specified as the power range for which its transfer function follows the ideal linear-in-dB relationship with an error smaller than or equal to ±1 dB. Again, the attainable dynamic range differs considerably for the various types of power detectors.

According to its definition, the average power is a metric for the average energy content of a signal and is not directly a function of the shape of the signal in time. In other words, the power contained in a 0-dBm sine wave is identical to the power contained in a 0-dBm square wave or a 0-dBm WCDMA signal; all these signals have the same average power. Depending on the internal detection mechanism, though, power detectors may produce a slightly different output signal in response to the aforementioned waveforms, even though their average power level is the same. This is due to the fact that not all power detectors strictly implement the definition formula for signal power, being the mean of the square of the signal. Most types of detectors perform some mixture of peak detection and average power detection. A waveform independent detector response is often desired in applications that exhibit a large variety of waveforms, such that separate calibration for each waveform becomes impractical.

The shape of the detector transfer function from the RF input power to the DC output voltage determines the required resolution of the ADC connected to it. The overall power measurement error is the combination of the error introduced by the detector, and the quantization error contributed by the ADC. The impact of the quantization error on the overall transfer's accuracy is highly dependent on the detector transfer shape, as shown in Figure 71 and Figure 72.

Figure 71. Convex Detector Transfer Function

Figure 72. Linear Transfer Function

Figure 71 and Figure 72 shows two different representations of the detector transfer function. In both Figure 71 and Figure 72 the input power along the horizontal axis is displayed in dBm because most applications specify power accuracy requirements in dBm (or dB). Figure 71 shows a convex detector transfer function, while the transfer function on the right hand side is linear (in dB). The slope of the detector transfer function — the detector conversion gain – is of key importance for the impact of the quantization error on the total measurement error. If the detector transfer function slope is low, a change, ΔP, in the input power results only in a small change of the detector output voltage, such that the quantization error is relatively large. On the other hand, if the detector transfer function slope is high, the output voltage change for the same input power change will be large, such that the quantization error is small. Figure 72 has a very low slope at low input power levels, resulting in a relatively large quantization error. Therefore, to achieve accurate power measurement in this region, a high-resolution ADC is required. On the other hand, for high input power levels the quantization error are very small due to the steep slope of the curve in this region. For accurate power measurement in this region, a much lower ADC resolution is sufficient. Figure 71 has a constant slope over the power range of interest, such that the required ADC resolution for a certain measurement accuracy is constant. For this reason, the LOG-linear curve in Figure 71 generally leads to the lowest ADC resolution requirements for certain power measurement accuracy.

8.1.1.1.2 Types of RF Power Detectors

Three different detector types are distinguished based on the four characteristics previously discussed:

• Diode Detector
• (Root) Mean Square (R)MS) Detector
• Logarithmic Detectors

8.1.1.1.2.1 Diode Detector

A diode is one of the simplest types of RF detectors. As depicted in Figure 73, the diode converts the RF input voltage into a rectified current. This unidirectional current charges the capacitor. The RC time constant of the resistor and the capacitor determines the amount of filtering applied to the rectified (detected) signal.

Figure 73. Diode Detector

The advantages and disadvantages can be summarized as follows:

• The temperature stability of the diode detectors is generally very good because they contain only one semiconductor device that operates at RF frequencies.
• The dynamic range of diode detectors is poor. The conversion gain from the RF input power to the output voltage quickly drops to very low levels when the input power decreases. Typically a dynamic range of 20 dB to 25 dB can be achieved with this type of detector.
• The response of diode detectors is waveform dependent. As a consequence of this dependency, for example, its output voltage for a 0-dBm WCDMA signal is different than for a 0-dBm unmodulated carrier. This is due to the fact that the diode measures peak power instead of average power. The relation between peak power and average power is dependent on the wave shape.
• The transfer shape of diode detectors puts high requirements on the resolution of the ADC that reads their output voltage. Especially at low input power levels a very high ADC resolution is required to achieve sufficient power measurement accuracy (See Figure 71).

Table 1. Design Parameters

8.2.1.2.1 Detector Interfacing

For optimal performance of the LMV221 device, it is important that all its pins are connected to the surrounding circuitry in the appropriate way. Starting from the Functional Block Diagram the function of each pin is elaborated in the following sections. The details of the electrical interfacing are separately discussed for each pin. Output filtering options and the differences between single ended and differential interfacing with an ADC are also discussed in detail in the following subsections.

8.2.1.2.1.1 RF Input

RF parts typically use a characteristic impedance of 50 Ω. To comply with this standard the LMV221 has an input impedance of 50 Ω. Using a characteristic impedance other then 50 Ω causes a shift of the logarithmic intercept with respect to the value given in the 2.7-V DC and AC Electrical Characteristics. This intercept shift can be calculated according to Equation 13.

Equation 13.

The intercept shifts to higher power levels for R_SOURCE > 50 Ω, and shifts to lower power levels for R_SOURCE < 50 Ω.

8.2.1.2.1.2 Output and Reference

The possible filtering techniques that can be applied to reduce ripple in the detector output voltage are discussed in Filtering. In addition, two different topologies to connect the LMV221 to an ADC are elaborated.

8.2.1.2.1.2.1 Filtering

The output voltage of the LMV221 is a measure for the applied RF signal on the RF input pin. Usually, the applied RF signal contains AM modulation that causes low frequency ripple in the detector output voltage. CDMA signals, for instance, contain a large amount of amplitude variations. Filtering of the output signal can be used to eliminate this ripple. The filtering can either be achieved by a low pass output filter or a low pass feedback filter. Those two techniques are shown in Figure 75 and Figure 76.

Figure 75. Low Pass Output Filter

Figure 76. Low Pass Feedback Filter

Depending on the system requirements one of the these filtering techniques can be selected. The low pass output filter has the advantage that it preserves the output voltage when the LMV221 is brought into shutdown. This is elaborated in Output Behavior in Shutdown. In the feedback filter, resistor R_P discharges capacitor C_P in shutdown and therefore changes the output voltage of the device.

A disadvantage of the low pass output filter is that the series resistor R_S limits the output drive capability. This may cause inaccuracies in the voltage read by an ADC when the ADC input impedance is not significantly larger than R_S. In that case, the current flowing through the ADC input induces an error voltage across filter resistor R_S. The low pass feedback filter does not have this disadvantage.

NOTE

Note that adding an external resistor between OUT and REF reduces the transfer gain (LOG-slope and LOG-intercept) of the device. The internal feedback resistor sets the gain of the transimpedance amplifier.

The filtering of the low pass output filter is achieved by resistor R_S and capacitor C_S. The −3 dB bandwidth of this filter can then be calculated by: ƒ_{−3 dB} = 1 / 2πR_SC_S. The bandwidth of the low pass feedback filter is determined by external resistor R_P in parallel with the internal resistor R_TRANS, and external capacitor C_P in parallel with internal capacitor C_TRANS (see Figure 79). The −3 dB bandwidth of the feedback filter can be calculated by ƒ_{−3 dB} = 1 / 2π (R_P//R_TRANS) (C_P + C_TRANS). The bandwidth set by the internal resistor and capacitor (when no external components are connected between OUT and REF) equals ƒ_{−3 dB} = 1 / 2π R_TRANS C_TRANS = 450 kHz.

8.2.1.2.1.3 Interface to the ADC

The LMV221 can be connected to the ADC with a single-ended or a differential topology. The single ended topology connects the output of the LMV221 to the input of the ADC and the reference pin is not connected. In a differential topology, both the output and the reference pins of the LMV221 are connected to the ADC. The topologies are depicted in Figure 77 and Figure 78.

Figure 77. Single-Ended Application

Figure 78. Differential Application

The differential topology has the advantage that it is compensated for temperature drift of the internal reference voltage. This can be explained by looking at the transimpedance amplifier of the LMV221 (Figure 79).

Figure 79. Output Stage of the LMV221

Equation 14 shows that the output of the amplifier is set by the detection current I_DET multiplied by the resistor R_TRANS plus the reference voltage V_REF:

Equation 14. V_OUT = I_DET R_TRANS + V_REF

where

• I_DET represents the detector current that is proportional to the RF input power.

Equation 14 shows that temperature variations in V_REF are also present in the output V_OUT. In case of a single ended topology the output is the only pin that is connected to the ADC. The ADC voltage for single ended is thus:

Equation 15. V_ADC = I_DET R_TRANS + V_REF

A differential topology also connects the reference pin, which is the value of reference voltage V_REF. The ADC reads V_OUT – V_REF:

Equation 16. V_ADC = V_OUT – V_REF = I_DET R_TRANS

Equation 16 no longer contains the reference voltage V_REF anymore. Temperature variations in this reference voltage are therefore not measured by the ADC.