Agilent PNA-X Series Microwave Network Analyzers Active-Device Characterization in Pulsed Operation Using the PNA-X Application Note
Table of Contents Introduction ..........................................................................................................................3 Device Types ........................................................................................................................4 Pulsed-RF Measurement Types ................................................................................5 Pulsed-RF Detection Techniques..............................................................................
Introduction Vector network analyzers (VNA) are the common tool for characterizing RF and microwave components in both continuous-wave (CW) and pulsed operations. Some external equipment may be used in conjunction with a VNA to modulate the stimulus or DC bias, and to perform accurate S-parameter measurements in pulsed operation.
Device Types Figure 1 shows two types of pulse operation modes, pulsed-RF and pulsedbias. Pulsed-RF operation drives the device with a pulse-modulated RF signal while the DC bias is always on. Amplifiers in receivers used in pulsemodulated applications are typically tested under pulsed-RF operation.
Pulsed-RF Measurement Types Figure 2 shows three major types of pulsed-RF measurements. The first two are pulsed S-parameter measurements, where a single data point is acquired for each carrier frequency. The data is displayed in the frequency domain with magnitude and/or phase of transmission and/or reflection. Average pulse measurements make no attempt to position the data point at a specific position within the pulse.
Pulsed-RF Detection Techniques Figure 3 shows an important measure of a pulsed RF signal and its relationship between the time and frequency domain. When a signal is switched on and off in the time domain (pulsed), the signal’s spectrum in the frequency domain has a sin(x)/x response. The width of the lobes is inversely related to the pulse width (PW). This means that as the pulses get shorter in duration, the spectral energy is spread across a wider bandwidth.
Wideband detection Wideband detection can be used when the majority of the pulsed-RF spectrum is within the bandwidth of the receiver. In this case, the pulsed-RF signal will be demodulated in the instrument, producing baseband pulses. With wideband detection, the analyzer is synchronized with the pulse stream, and data acquisition only occurs when the pulse is in the “on” state.
Narrowband detection Narrowband detection is used when most of the pulsed-RF spectrum is outside the bandwidth of the receiver. In other words, the pulse width is narrower than the minimum data acquisition period with the widest available receiver bandwidth. With this technique, the entire pulse spectrum is removed by filtering except the central frequency component, which represents the frequency of the RF carrier. After filtering, the pulsed-RF signal appears as a sinusoid or CW signal.
Pulsed-RF S-parameter Measurements Using PNA-X This section discusses pulsed-RF S-parameter measurements using the PNA-X with wideband detection and narrowband detection techniques. PNA-X pulse system PNA-X hardware overview Figure 6 shows the PNA-X block diagram with two test ports, two internal sources, source/receiver attenuators, internal combining network, and rear access loops with mechanical path switches.
PNA-X IF paths Figure 7 shows the PNA-X receiver/IF path block diagram with internal pulse generators and modulators. The narrowband filter path employs a crystal filter with 30 kHz bandwidth centered at 10.7 MHz for better signal selectivity, and also adds receiver gating capability for narrowband pulse measurements.
Internal pulse modulators The PNA-X with the internal pulse modulator options adds a pulse modulator to the “OUT 1” of each internal source (Option 021 for Source 1 and Option 022 for Source 2). Both modulators are driven by a common pulse, which can be selected from P1 through P4 internal pulse generators, or external pulse inputs with minimum pulse width of system clock timing resolution.
Pulse system delays From pulse trigger to internal pulse generators, to pulse modulators, and then to ADC for data acquisition, there are some delays in the pulse system that need to be taken into account. Figure 9 shows the timing diagram from the pulse trigger at PULSE SYNC IN through the data acquisition with wideband detection technique. Figure 9.
Internal pulse generators start generating the pulses approximately 60 to 100 ns after the pulse trigger inputs at the PULSE SYNC IN – (denoted as pulse trigger delay). The jitter of this delay is the minimum time resolution of the system. The exact pulse trigger delay can be measured with an oscilloscope between PULSE SYNC IN and one of the PULSE OUT (P1 through P4). If the internal clock is used to trigger the pulse generators, this pulse trigger delay is very small and therefore it can be neglected.
Setting up pulsed-RF measurements continued Step 2 Synchronize data acquisition and pulses With internal pulse generator option, pulse trigger feature is added to the PNA-X trigger system (Figure 10b Pulse Trigger tab on Trigger dialog). Specify trigger source – either internal or external (incoming pulses to PULSE SYNC IN) and check “Synchronize receiver to pulse generator Pulse0”. Note that measurement trigger (MEAS TRIG IN port on PNA-X rear panel) is not used in typical setup.
Setting up with pulsed-RF measurement application Option 008 pulsed-RF application optimizes the setup through Pulse Setup dialog, and adds narrowband pulse measurements. The basic pulsed-RF measurements can be done by simply choosing measurement type (Standard or Pulse Profile), specifying the pulse width and period on Pulse Setup dialog (Figure 11a). All other necessary settings explained earlier are done automatically.
Autoselect Pulse Detection Method If selected, wideband detection method is used for pulse widths of 267 ns or wider for point-in-pulse measurements, and 1.6 us or wider pulse width for pulse profile measurements. Otherwise, narrowband detection method is used. Autoselect IF Path Gain and Loss Autoselect is used in most cases, but it may need to be adjusted for highpower configurations with user-supplied high-power couplers and attenuators instead of internal test port couplers.
PNA-X wideband pulse measurements Wideband pulse measurements are accomplished by modulating the RF source, setting the IF bandwidth wide enough to capture all pulse spectrum, and synchronizing the modulation and measurements with appropriate triggering and delay settings. The internal pulse generators and modulators can be controlled either by SCPI/COM commands or the Pulse Setup dialog added with Option 008 pulsed-RF measurements.
In wideband pulse profile measurements, all the data is acquired with a single pulse, thus the pulse width needs to be relatively wide (Figure 13). For each data point with 1 MHz and wider IF bandwidth on DSP version 4 or 10 kHz and wider IF bandwidth on DSP version 5, the latter half of samples from the previous data point are used for the first half of samples of the next data point. This means that the data acquisition timing resolution is one-half of the data acquisition time.
Table 1 shows theoretical values of the data-acquisition time-per-point and timing resolution for each IF bandwidth. These represent the minimum pulse width for wideband point-in-pulse and timing resolution for wideband pulse profile respectively. The minimum timing resolution can also be found on the PNA-X display by setting CW time sweep and dividing sweep time by the number of points.
Synchronizing pulsed-RF stimulus and measurements In wideband pulse measurements, the appropriate IF bandwidth and the pulse generators’ width/delay are set in order to acquire the data within pulses. The following measurements demonstrate the results with inappropriate IF bandwidth and delay settings, which should be avoided when customizing pulse measurement setups.
When the measurement delay (or pulse delay) is adjusted, it is crucial to keep the data acquisition window within the pulses. Unlike previous examples, the measurement errors may not be obvious due to a few ADC samples outside of the pulses. Figure 14b shows the measurement errors caused by improper measurement delay. It is recommended to have enough timing margin when adjusting IF bandwidth and/or the delays.
In pulse-to-pulse measurements, PRI has to be wide enough for the PNA-X to keep up with the data acquisition from one pulse to the next. The minimum PRI becomes slightly longer with narrower IF bandwidth as the data acquisition time becomes longer. Table 2 is the minimum PRI by IF bandwidth for 26.5 GHz PNA-X with DSP version 4 and 1.5 GHz CPU board as a reference. The minimum PRI could be improved as the data processing time becomes shorter with a faster CPU and/or DSP. Table 2.
Wideband pulse system dynamic range The wideband detection with PNA-X can be used for relatively fast pulsed-RF operation (with shorter PRI or high PRF) compared to legacy VNA’s due to a wide IF bandwidth up to 15 MHz. However, the recommended IF bandwidth should be wide enough to capture most of the pulse spectrum, while the dynamic range and trace noise performance meets the measurement requirements.
PNA-X narrowband pulse measurements The Option 008 pulsed-RF measurement uses narrowband detection method for a pulse width narrower than 267 ns in standard pulse (point-inpulse) measurements and 1.6 us pulse width or narrower in pulse profile measurements. When narrowband detection is selected, each receiver gate width and delay can be controlled independently.
Figure 18 shows how the narrowband filter path affects pulsed-RF signals by looking at the central frequency tone of the pulsed-RF signals with 200 ns pulse width and 2 us PRI. In Figure 18, the top window shows the PNAX’s 500 Hz IF filter response with the wideband IF path. The bottom window shows the narrowband filter path, which removes some of the undesired spectrum of the pulse-RF signals with 30 kHz crystal filter. Figure 18.
Software gating The IF gating in the narrowband filter path uses hardware switches driven by a pulse generator, which leaves residual noise at the gate off time caused by internally generated noise and the switch isolation. When hardware IF gating and an internal pulse generator are used, the on and off times of the gated signal can be determined precisely.
Figure 20 shows an example of the dynamic range improvement that the PNA-X has over the E836x PNA models due to the narrowband ilter path and the software gating technique. In this example of .001% duty cycle, the measurement in the PNA is very noisy, and many averages are needed to get a usable measurement. At this duty cycle, the PNA-X’s hardware gives about 20 dB of improvement and the software gating provides another 20 dB, for a total of 40 dB improvement.
Digital filter nulling The gated and filtered pulsed IF signal is digitized and sent to DSP for IF filtering. The remaining pulse spectrum is filtered using custom IF filter nulls leaving only the center spectral component to analyze. The spacing between the nulls depends on the IF filter bandwidth, which is adjusted to align with the PRF. Figure 21 illustrates the effects of digital filter nulling. In this example, the PRI is adjusted to 1.
Active-Device Measurements with Calibrated Stimulus Devices that operate under pulsed conditions are often discrete active devices or modules that consist of amplifiers and/or mixers. The performance of these devices is typically power-dependent. Therefore, they are characterized in linear and nonlinear operating conditions. Inaccurate stimulus power may introduce considerable measurement errors.
Accurate pulsed stimulus using receiver leveling Reference receivers are typically used for the receiver leveling, although any receiver or a power sensor (if added as a receiver to the PNA-X) can be used. The source level accuracy in receiver leveling mode depends on the receiver’s absolute power measurement accuracy, therefore receiver calibration is strongly recommended.
The following examples show calibration with pulse modulation off. Receiver leveling with wideband detection This section explains the basic steps to setup a leveled pulsed-RF stimulus, make calibrated swept-frequency point-in-pulse S-parameter, and absolute power measurements using wideband detection technique. Setting up wideband pulse measurements First, set up the wideband pulse measurements with the following conditions.
Performing receiver and S-parameter calibrations A calibrated receiver enables accurate and leveled pulsed-RF stimulus using receiver leveling mode. In the calibration process, use a power sensor to create source power and receiver calibration coefficients. In order for the power sensor to read correct power levels, the pulse modulation must be turned off. Source power calibration should be part of the calibration process.
Comparing the results Figure 24 also shows memory traces comparing the input match, gain, input power, and output power under pulsed conditions with open loop and receiver leveling. The difference is very small in the input match and gain measurements, but it is quite large in the absolute power measurements. If you perform only S-parameter measurements and the DUT is in the linear operation, the open loop leveling is probably acceptable.
Figure 25b. Measurements after calibration with pulse modulator off Figure 25c. Corrected pulsed-RF S-parameters and absolute power measurements using narrowband detection Note that the reference receiver measures the pulsed stimulus power at the set power minus 20*log (receiver gate duty cycle) with some additional path loss (Figure 25a). Once the reference receiver is calibrated it reads close to set power at -10 dBm (Figure 25b).
Swept-power measurement examples The setup and calibration processes described in the previous sections can be used for swept-power with either wideband or narrowband pulse measurements. Figure 26 compares the results between open loop and receiver leveling modes in swept-power measurements with the same sample 5 GHz ampliier using wideband detection technique. The following setup is used in this example.
Improving stimulus power level accuracy in pulse profile measurements In pulse proile measurements, regardless of wideband or narrowband detection techniques, receivers measure signals inside of pulses as well as noise outside of pulses. If receiver leveling is used while the receivers measure noise, the source tries to adjust the source power by referencing the noise measurements causing source unleveled errors.
Compression vs. Frequency Measurements in Pulse Mode Gain Compression (commonly specified as input/output power at 1 dB compression point: IP1dB/OP1dB) is a very popular amplifier characteristic. It is measured and calculated from either swept-power gain (S21) or amplifier output power at a single frequency point. Often this is repeated at many frequency points to characterize the amplifier compression point over its operating frequency range (Figure 28).
In the swept power measurement example shown earlier (Figure 26), found were different input powers at compression points by the leveling mode due to the errors of the input powers. In GCA, the input and output powers at compression point versus frequency also show some difference between the leveling modes. Shown in Figure 29, the maximum difference is about 1 dB. Figure 29.
Two-tone IMD measurements in Pulse Mode The PNA-X employs clean internal sources with high output power, internal combining network, and an optional IMD measurement application that makes two-tone IMD measurements over the frequencies easier and faster.
The IMD calibration includes R1 receiver calibration, source 1 and source 2 power calibration, and B receiver calibration when port 1 for the DUT input and port 2 for the DUT output are selected. If 2-port error correction is used during an IMD calibration, the power sensor mismatch during R1 receiver calibration is corrected and the transmission response term is used to transfer the R1 receiver calibration to the B receiver.
Figure 30b. Wideband pulse swept-frequency IMD measurements with receiver leveling Figure 30c shows swept-power IMD measurements with the same pulse setup as above and compares open loop and receiver leveling modes. Although IM3 traces show a few dB difference, both input and output referred IP3 are very close to each other. This is because the difference of IM products is translated to smaller changes of IP3 from the following IP3 formula.
Additional Resources www.agilent.com www.agilent.com/find/pna Application note Pulsed-RF S-Parameter Measurements Using Wideband and Narrowband Detection AN 1408-12, literature number 5989-4839EN Web More information about the PNA-X, including pulsed-RF measurement option can be found at: www.agilent.com/find/pnax Agilent pulsed-RF signal generation, power/spectrum/network measurements can be found at: www.agilent.com/find/pulsedrf Application Note 1408-21 Three-Year Warranty www.agilent.
