Filtering Modules for AWGN / IMD Test Setups for 4G LTE Bands
By Rafi Hershtig, VP of Engineering and Applied R&D, K&L Microwave
The rollout of 4G LTE networks, characterized by faster data rates than cellular 3G systems, utilizes overlapping frequency bands and presents increasingly challenging interference issues due to new intermodulation (IM) sources. Whereas existing test protocols have focused on IM products related to two CW signals, in reality, there are multiple modulated Tx carriers.
For a modulated bandwidth of X[MHz], the third-order IM product exhibits a modulated bandwidth of 3X[MHz], spread across the receiver noise floor. As a rule of thumb, for each 20dB rise in a receiver’s noise floor, the effective covered area is reduced by a factor of 10! Obviously, there are serious implications for ROI and quality of service (QoS).
A certain amount of interference is generated by LTE networks themselves (inter-cell interference), but 4G testing protocols have expanded to take into account the contribution of coexistence interference generated by “Blocker” signals from non-LTE networks (GSM, UMTS, etc.) Since network coverage and data throughput are severely impacted by signal to noise plus interference ratio (SNIR), LTE protocols are demanding stringent measurements throughout the component supply chain. From power amplifiers and Tx/Rx chipsets to tunable capacitors for impedance matching networks and various passive components, suppliers that comply with IMD specs and provide IMD data have a strategic competitive advantage. That said, lack of equipment for making such measurements forces engineers from R&D through production to construct their own setups. The degree of nonlinearity of each component impacts the IMD level of the system. When measured and held in check, the performance of the system to be deployed or maintained can be simulated and optimized system-to-system through spatial solutions, equipment isolation, frequency guard bands, component selection, and so forth.
The classic test setup of equal power two tones generated at the Transmit (Tx) side, producing third-order IMD in the Receive (Rx) band, is designed to test inter-cell interference. For 4G networks, major IMD sources include Blocker signals such as second harmonics of single tones generated at lower frequencies (UHF transmitters, TV, GPS, etc.), as well as second- or third-order products of strong inter-cell Tx signals mixing with interferer signals produced by non-LTE communication platforms.
In this article, building blocks of RF filtering solutions for test systems are described. As bands proliferate, these building blocks can be switched in and out to provide measurements over multiple LTE bands. Customizable building blocks and flexible design of the overall test setup enable measurements of IMD products from combinations of two or more signals.
PIM Versus IM
Although PIM and IM are sometimes used interchangeably, Passive Intermodulation (PIM) products are caused by faulty manufacturing and installation processes, such as bad solder joints, metal-to-metal junctions, ferromagnetic materials, flawed/substandard surface finishes, etc. These causes, by their nature, are only weakly dependent on frequency, so it is accepted practice to perform PIM testing of passive products (cables, connectors, etc.) at a single frequency band, rather than over multiple bands. Target PIM levels for passive devices by industry consensus are between -156dBc to -169dBc for two +43dBm tones. The dynamic range required by a suitable test setup must avoid any reflections of the carriers back to the sources, which can hinder establishing the baseline.
Intermodulation (IM) products, by contrast, are caused by non-linear devices, such as PIN diodes, transistors, tunable BAW/SAW filters, MEMS capacitors, etc. Frequency dependency is significantly higher, such that IM testing must be performed over multiple bands, particularly over the designated operational bands. Required Intermodulation Difference (IMD) levels specified by industry are currently numerically less severe than PIM levels, but the gap between them is narrowing with the incremental rise of Tx tone power levels.
IP3 and IM Measurements
Reasons for non-linearity causing IM products for a general DUT are extensively covered in the literature. From the point of view of implementing a test setup, the deeper the desired dynamic range (usually, IMD + 10dB), the more attention must be paid to preventing generated IM and reflected energy from reaching the DUT. The following are typical IMD levels over LTE bands as recognized by industry:
1. High Power Passive Devices for Wireless Base Station Applications:
-113dBm for two +43dBm tones
2. High Power Passive Devices for Distributed Antenna Systems (DAS):
-118dBm for two +43dBm tones
3. High Power Passive Devices for PIM Analyzer Applications:
-127dBm for two +43dBm tones
4. Low Power Broadband Switches, MEMS Capacitors, Tx/Rx Chipsets and Tunable Devices:
-140dBm for two +26dBm tones (In many cases, the first tone is in the Tx band,
and the second tone is a Blocker.)
IP3 and IMD are related by the well-known formula (in dBm):
IP3 = P+IMD/2
where IMD is the difference between the third-order IM product and the two tones each
at power P.
A commonly used test setup (Figure 1) consists of two Tx signals from the download (DL) band, creating a third-order IM product in the Rx band. While convenient and affordable because of the availability of amplifiers, low PIM broadband 3dB hybrids, and relatively inexpensive isolators, this well-known and well-used arrangement has a dynamic range limitation. The amplifiers are isolated by about 50dB by the isolators and 3dB coupler.
Depending on the return loss of the DUT, the two Tx signals will be reflected back to the coupler and split evenly between the Tx1 and Tx2 ports. Since the isolators are ferromagnetic devices, new IMD products will be generated, making the iso-coupler area high in IMD content. As a consequence, the Tx filter will have to exhibit at least 100dB rejection at the Rx band to block these IMD products from traveling to the Rx filter, and the amplifiers will have to generate 3.5dB more gain in order to compensate for the iso-coupler unit. The poorer the return loss of the DUT, the more the dynamic range of the setup is limited.
In Figure 2, the amplifiers are isolated by the two bandpass filters by approximately 75dB. While the DUT has a low PIM termination, the third-order IM product is measured by the spectrum analyzer through the receive filter. The setup can be controlled such that one tone is stationary and the second tone sweeps for gathering additional data points. For maintenance, the setup can be periodically checked by terminating the common port of the triplexer and measuring the third-order IM product through the Rx filter, verifying the existence of the baseline. As a best practice, RF ports that are frequently mated should be protected by low PIM connector savers, as number of uses tends to create PIM sources. Further, it is advisable to place a DIN connector at the common port for better durability and lower surface currents. The three other connectors are not as critical; they do not affect the PIM level of the setup, since the two strong carriers are not simultaneously present at any time.
In Figure 3, Tx1 represents a strong signal from the Transmit band, and Tx2 represents any signal that can be mixed with Tx1 to produce an IM product in the Receive band, such as in the second-order case, where IM = Tx1-Tx2.
The output duplexer covers the entire DL and UL bands. This setup (Figure 4) works only if the IM products are in the Receive band.
Figure 5B shows a test setup for IM measurements in the Rx band using a 4-port device with block diagram depicted in Figure 5A. The triplexer enables two (or more) carriers to be combined, where one is generated in the Tx band and the second is a Blocker frequency anywhere from DC to about 2.5 times center frequency (Fo). This limitation is due to the natural periodic re-entry of the bandreject filter representing pronounced degradation of the upper-side passband. Mirror imaging of the input filters to the output side of the DUT enables testing of the DUT from both input and output.
In Figure 6, the upper passband of the bandreject filter must be extended to pass higher Blocker frequencies. To overcome the natural degradation of the bandreject filter, a new duplexer is introduced, passing higher Blocker frequencies. This duplexer consists of a lowpass (LP) filter with a cut-off frequency below 2.5 times Fo and one or more bandpass (BP) filters matching the higher Blocker frequencies. As an example, for LTE Band 8, the DL band is from 925 MHz to 960 MHz, defined as the Receive band for handheld devices, the UL band is from 880 MHz to 915 MHz, defined as the Transmit band for handheld devices, and relevant Blockers are 45 MHz, 835 MHz to 870 MHz, 1805 MHz to 1875 MHz, and 2685 MHz to 2790 MHz. In order to test the IMD products in the Receive band for the handheld device, one signal is produced in the Transmit band, and the second signal is a Blocker. For bandreject filters covering 880 MHz to 960 MHz, about 2200 MHz is a limit for upper passband. Various techniques for extending upper passband described in the literature have not been found suitable for low PIM designs. The LP/BP duplexer is introduced such that the cut-off frequency of the lowpass filter (LPF) is around 2000 MHz and the bandpass filter (BPF) passes the 2685 MHz to 2790 MHz range. It is also possible to introduce a highpass filter (HPF) in place of the BPF to allow Blocker frequencies up to 12.75 GHz. The Blocker signals produced at the generator pass through a “non-reflective” switch, so that forward and reverse IMD products can be measured.
For more information, please contact the author at firstname.lastname@example.org.
this article to a friend!