AN_99_06 电路图.pdf

Accurate RF Power Measurements of Second and Third Generation Wireless Communication Signals Notes Slide 1: In this seminar, we will be discussing different methods used for measuring power of modulated communication systems. While some of our examples will deal with cellular thermal sensors and diode sensors. Thermal sensors use either a thermocouple or a thermistor to measure the RF power. Both use the properties of heat to integrate the RF power over time. Due to the nature of thermal RadioFans.CN 收音机爱 好者资料库 characteristics, thermal sensors provide a slow response to changes in power, such as when you have a modulated condition. Diode sensors rectify the RF energy to a dc voltage. The advantage of this method is that they are able to track very rapid changes in power if the sensor is designed properly. We will discuss the difference between tracking the average power of a modulated signal and the peak power. For optimum performance, a diode sensor must have enough video bandwidth to track the peak envelope of the modulated signal. Notes Slide 5: When considering power measurement accuracy, or more correctly, measurement uncertainty, a number of factors must be taken into account. The 8 factors listed here are the more significant ones. Mismatch uncertainty due to the interaction between the sensor and source is usually the most dominant. Then, factors such as instrumentation linearity, calibrator uncertainty and calibrator/sensor mismatch should be included. Finally, if the measurement takes place within the last 15 dB of dynamic range, zero and noise will become a major contributor to measurement uncertainty. Notice that these terms will change whenever there is a change in frequency, testport match, or power level. So in order to determine total measurement uncertainty, a complete analysis of each of these terms is required whenever power or frequency is changed. Giga-tronics has a paper that discusses how to calculate each of these terms manually. In addition, we have developed an Accuracy Audit program which takes the measurement information and automatically converts the information into an uncertainty number. The program is a stand alone executable and is available on CD. If you would like a copy of the paper or the program, send us an email either directly to me or through our web site, and we will send you a CD with the materials. During this seminar, I will spend time discussing application concerns when measuring modulated signals rather than show how to calculate the individual uncertainties. Notes Slide 6: Here we have three classes of signals identified; Constant wave, or CW, where the amplitude is constant over time, Modulated Signals and Digital Modulation. Digital communication signals use a combination of phase and amplitude modulation to code the carrier and have a more complex nature. Before talking about modulated measurements, lets first review how power meters measure CW signals. Notes Slide 7: We will first consider how thermal sensors measure CW power and then compare that to a diode sensor. As mentioned previously, thermal sensors depend on the relationship between heat and power to determine the power level of the signal. The thermal sensor will therefore rise to the level of the power detected and eventually settle to the final reading. Notes Slide 8: When measuring CW power with a diode sensor, the sensor samples the power at a given interval which is determined by the meter. The meter stores each of these samples in a bucket, averages all of the samples in the bucket and provides a measurement reading. Averaging is the process of accumulating a number of buckets, averaging the readings and providing a single answer of all averaged buckets. So, if we were to set up the meter for an average of one, the meter would provide a reading from each bucket. When we talk about measurement speed, we are referring to how fast the meter can provide measurements over the GPIB bus. Measurement speed therefore is determined by the speed of processing of measurements by the microprocessor. We can also see that the number of samples included in each measurement is determined by the sample rate. In Giga-tronics meters all of the samples are included in a measurement. If the RadioFans.CN 收音机爱 好者资料库 measurement speed decreases due to operating conditions, then the number of samples accumulated will increase. No samples are lost. Notes Slide 9: Lets review the operating characteristics of diode sensors. Dynamic range is a key feature for power meters. As mentioned during the introduction, the handset must be characterized over a wide dynamic range in order to optimize system efficiency. A thermal sensor will have a dynamic range of 40 to 50 dB. A diode sensor starts out with a linear dynamic range of 50 dB, from -70 dBm to around -20dBm. This is known as the square-law region of the sensor. It is also called the linear region since there is a linear relationship between the power in and the voltage out. Between -20 dBm and +20 dBm the diode sensor will operate, but it is not inherently linear. Giga-tronics achieves a 90 dB dynamic range by characterizing the non-linear properties of the sensor during the front panel calibration. By storing the non-linear characteristics in the meter and recalling the correction data during measurements, the meter is able to provide a very linear dynamic range from -70 to +20 dBm. Notes Slide 10: This diagram explains further how Giga-tronics achieves a 90 dB dynamic range which is NIST traceable. We use a Wheatstone bridge which contains a thermistor as the reference. The thermistor is also called a bolometer and is a device whose measurement properties can be measured by NIST. A Wheatstone bridge with a thermistor is the method used by all power meters to establish a traceable 0 dBm reference for calibrating power sensors. What we have done in the Giga-tronics meters is included a patented process which incorporates stepped attenuators into the loop. Doing this allows us to step the power level from -30 to +20 dBm in 1 dB steps. This provides a very linear power sweep which is our reference for calibrating the sensor throughout the non-linear range and gives us a 90 dB dynamic range capability. Notes Slide 11: Lets return to a CW measurement except this time we will measure a low level signal that is close to the noise floor of the meter. Notice that we no longer have a constant level power. Average power in this case is the mean of the power variations over time. We obtain the average power by accumulating a number of readings and using the average function of the meter. Both the thermal and diode sensors will settle to the average power. The diode sensor has the potential to settle much faster with a fast sample rate. A fast sample rate provides a large number of readings very quickly allowing the meter to average the measurements and settle to the reading. Notice that one of the characteristics of noise is the normal distribution around the mean. Notes Slide 12: We will now start to evaluate modulation measurements by first examining a two tone, AM modulated signal. Notice that we are skipping FM modulation measurements. Frequency modulation implies that only the frequency, or phase, of the signal is changing. The amplitude of a FM modulated signal does not vary. It therefore can be treated as a CW signal. During modulation measurements, our primary concern is the variation of the amplitude of the signal. Notice that an AM modulated signal varies power level over time and that the average power is the mean of the power variations. Peak power is a term used to describe the power level of the modulated waveform at a specific point in time. We obtain the average power by averaging the peak power envelope. This is somewhat similar to the previous condition where we were measuring a CW signal with noise. The difference is that the modulation pattern is not normally distributed. Lets look further at the process of measuring power of an AM modulated signal. Notes Slide 13: We will first consider how a thermal sensor measures a modulated signal. We see that when measuring modulated signals using a thermal sensor, we must wait for the sensor to settle to the RadioFans.CN 收音机爱 好者资料库 average power level of the modulated waveform. This can often take multiple cycles of the modulation cycle. A diode sensor samples the pattern and begins to provide an average of the samples accumulated. Notice that if the sample rate is fast enough, it is possible to obtain the average reading after one cycle of modulation. One of the requirements of the diode sensor is that it be able to track the power envelope. To do this, it must have enough amplitude, or video, bandwidth to track the envelope accurately. In other words, the sensor must have enough rise and fall time to track the waveform. If it does not, the sensor will provide up to a few tenths dB positive offset error. Another important requirement for fast accurate measurements is the sampling method used. Notice that the sample rate can be below the modulation frequency. Since power is a scalar measurement, we do not need to obtain phase information in order to calculate the average power. Remember that the power measured is an integrated measurement which is phase independent. A consideration when undersampling is the potential for aliasing. In order to avoid aliasing, Giga-tronics power meters use asynchronous sampling. This technique minimizes aliasing and provides fast accurate power measurements. There is the possibility of measuring higher modulation rate signals which are above the bandwidth of the sensor. It is possible to make these measurements using diode sensors by staying within the square-law region of the sensor. When making measurements this way the sensor will integrate the waveform and provide average power without having to track the actual power envelope. This is similar to a thermal sensor only much faster. Notes Slide 14: Here we have a more complex modulated signal which might represent a real world signal. Notice there is a repetitive waveform characteristic to the signal. On the left, we are showing the average power of the signal as it is integrated over time. Due to the nature of the modulation, it takes a number of cycles before the reading settles to the average power of the waveform. This gives us an indication why we sometimes need to use a high averaging number in order for the reading to settle. On the right we are setting the meter to read the signal for only one modulation cycle. This is similar to what we were seeing in a two-tone measurement. If we can set the meter to a time value, rather than an averaging number where there is limited control of measurement time, we can minimize the fractional N contributions of the signal and achieve a faster measurement. Notes Slide 15: (No Notes.) Notes Slide 16: The terms Peak sensors and Modulation sensors are commonly used to describe sensors that are designed to measure modulated power. Peak sensors in this case would refer to the fact that the sensor is able to track the peak power envelope. Giga-tronics uses the term Peak Pulse sensor to describe the sensors used to measure the peak pulse power of a pulse modulated signal. We use the term Modulation Sensors to describe sensors designed to perform complex modulated measurements. To review what we have discussed regarding modulation sensors, a modulation sensor has a video bandwidth that allows the meter to track the power envelope and provide the average mean power of the modulation signal. Notice that an important consideration is that the sensor will operate this way in the non-linear as well as the square-law region. This means that the meter must take into account the non-linear characteristics of the sensor and correct for them as the samples are accumulated. Giga-tronics power meters do this in the DSP before sending the measurements to the host processor. Notes Slide 17: Now lets take a look at pulse modulation. The average power of a pulse signal is similar to what we saw in a AM modulated signal. Average power in this case is the mean of the pulse waveform over many cycles. However, designers are interested in what the peak pulse power is when the power is on. If we only have a thermal or average-only sensor we need to estimate the top of the pulse by calculation using the duty cycle. If there is a slow rising and falling edge in the waveform, which is usually the case in wireless systems, then the estimate will not be accurate enough. What we need is a sensor with fast enough rise time, or bandwidth, to directly measure the top of the pulse. Notes Slide 18: The Giga-tronics Peak Pulse Sensors provide a fast rise time to easily measure the pulse power. The measurement point is set relative to a trigger point and provides a direct reading of power at the specified time. This method eliminates the error due to duty cycle estimation. Peak pulse sensors with a rise time of 100 nsec are available with the 8540C and 8650A power meters. Notes Slide 19: Now we are ready to discuss digital communication systems. This chart identifies the more popular wireless formats. The different systems can be classified into three categories, TDMA, GSM and CDMA. The different systems use different forms of digital modulation techniques. As we saw earlier, what is important in measuring modulated signals is to know the amplitude modulation rate and to verify that the sensor used has the bandwidth necessary to track the signal. In the case of TDMA and GSM, that would be the modulation rate of the vocoder. In the case of CDMA, the modulation rate is determined by the channel bandwidth. We will discuss this further in a bit. Notes Slide 20: TDMA and CDMA systems use a form of QPSK modulation for coding the RF carrier. QPSK modulation shifts the phase of the carrier to different positions within the four IQ quadrants. Although QPSK is a phase shift technique, the process of transitioning from one quadrant to another results in a change in amplitude. Notes Slide 21 : Here we see the result of QPSK modulation. The amplitude varies as the signal is modulated. Because of the nature of the modulation, the signal takes on a psudo-random quality that looks similar to noise. The rate of modulation depends on the vocoder or, in the case of CDMA systems, channel width. Notes Slide 22: Time Division Multiple Access uses time division to multiplex multiple callers. In this system, up to 8 callers may occupy the same frequency. Each caller is assigned a specific time slot within the burst. The carrier is modulated using QPSK techniques. Notes Slide 23: GSM systems use GMSK, or Gaussian Minimum Shift Keying, modulation. GMSK also uses a phase shift method to code the carrier in the four quadrants. However, instead of transitioning the carrier through or near the origin as in QPSK s