October 29, 2019
Filters are a key component in most RF and microwave systems, including those designed for emerging 5G commercial applications or aerospace/military environments. Design engineers select a filter for a given application based on several key specifications, including operational bandwidth, passband insertion loss and ripple response, and cutoff band performance. Vector network analyzers (VNAs) have become the test instrument of choice to characterize these parameters, so engineers should review several measurement specifications of the instrument to make sure they have selected the best VNA for their specific design.
Dynamic range is a common variable used to determine the appropriate VNA for characterizing transmission characteristics of deep cutoff filters. The analyzer must have enough system dynamic range (SDR) to properly characterize the filter cutoff performance. It is not the only key specification, however, as engineers need to also consider source harmonics when deciding if a VNA can precisely measure the cutoff performance of a target filter.
High source harmonics are an important consideration because they may interact with the out-of-band comeback characteristics of a filter under test to affect the filter transmission measurement. Because of this, engineers must be cognizant of the level of source harmonics in the VNAs they choose for testing filter dynamic range.
Impact of SDR
VNA SDR is defined as the difference between the maximum port output power and the noise floor of the receiver. The reason SDR has always been seen as an important spec is that if the VNA does not have enough dynamic range, the measurement result will show the instrument performance rather than the filter.
Most engineers view the SDR specification as a representation of the dynamic range the VNA can measure on all types of devices under test (DUTs). For many devices that are not frequency selective, such as cables, connectors, and other passive components, that assertion is correct. For certain passive DUTs that are both highly reflective and frequency selective – such as some high rejection filters – that may not be the case. There are measurement interactions with the DUT that can make the effective dynamic range of the VNA different from the specified SDR.
Why Source Harmonics are Important
A VNA is classified as a narrowband instrument and inherently has sources with various harmonic content and receivers with some level of non-linearity. The harmonic source content is usually not an issue, as those harmonics fall typically well outside the measurement band. When it comes to filter verification, however, engineers must compensate for this instrument characteristic.
Some filters have high-pass characteristics outside the normal operational frequency band. These filters create pathways for harmonic mixing products to appear at the receiver and affect the filter cutoff band measurements. This is especially true for filters that have a comeback or high-pass region that allow source signal harmonics to pass through to the receiver when the fundamental tone is in the cutoff band of the filter, as seen in figure 1. Because of this, 2nd and 3rd harmonic components can mix in the receiver to produce the fundamental frequency.
Bandpass Filter Example
To highlight the importance of limiting source harmonics, we conducted measurements on a bandpass filter (BPF) centered at 6.1 GHz using an Anritsu ShockLine MS4652xB VNA. In the transmission plot shown in figure 2, it can be seen that the filter has a sharp rejection band approximately 95 dB below the pass band with a comeback or high-pass characteristic starting at approximately 9.46 GHz, indicated by Marker 3.
The filter high side rejection band is expected to be flat to approximately 8.7 GHz (Marker 2 in figure 1) before the filter stops operating, however the plot shown has a step in the noise floor starting at 8 GHz (Marker 1 in figure 1) that is well above the expected filter cutoff level.
This measurement result is due to the high harmonic content from the source, as well as the design of the ShockLine MS4652xB. Its architecture is basically two VNAs in one – a lower frequency baseband VNA that transitions to a higher band instrument for microwave measurements. The transition between the baseband and microwave sections of the VNA occurs at approximately 8 GHz, causing the step in the noise floor. Figure 3 shows the same measurement setup as in figure 2, except the ports are terminated. Of particular note is the step in the terminated measurement in figure 3.
A BPF serves as a good example because harmonics on a VNA source can mix back into the receiver if the BPF has a high frequency comeback in the appropriate frequency range. For example, in the BPF used in the scenario above, the comeback region allows the source harmonics of the 8 GHz – 8.7 GHz sweep to pass through to the receiver and mix down to the fundamental frequency. The result is that the receiver measures a higher floor than expected. To confirm this explanation, a 10.4 GHz low-pass filter (LPF) is added with the 6.1 GHz BPF under test to attenuate the source harmonics of the ShockLine MS4652xB VNA, as the fundamental sweep frequency is from 8 GHz – 8.7 GHz.
Figure 4 shows the results before and after adding the low-pass filter to the measurement on the same plot. The blue trace is the transmission measurement without the additional low-pass filter and the brown trace is the transmission characteristic with the LPF in series. The blue trace shows a significantly higher rejection band measurement in the 8 to 8.7 GHz region than the brown trace. This shows that reducing source harmonic content with a series LPF clearly improved the cutoff band measurement between 8 GHz and 8.7 GHz.
To learn more about the importance of source harmonics when characterizing filters using a VNA, download a new application note from Anritsu.