MIMO RF Testing and Debugging Using a Multi-Channel Wideband Oscilloscope

This article mainly discusses the impact of antenna crosstalk impairments, phase noise and timing errors on MIMO downlink system performance, as well as fault diagnosis techniques using a time-coherent multi-channel oscilloscope and 89600 vector signal analyzer (VSA) software, hoping to help engineers gain a deeper understanding of the impact of error mechanisms on hardware error vector magnitude (EVM) performance and system-level RF transmitter performance. This article will use LTE as the research object, but its concepts can also be applied to other signal formats, such as Mobile WiMAX.

LTE MIMO Reference Signals and EVM

LTE MIMO cross-signaling generates a known signal throughout the frequency and time domains, called the Reference Signal (RS). This signal is fundamental to recovering the MIMO signal because it allows each receive antenna to establish a signal reference for each transmitter. Figure 1 shows how the individual symbols of the reference signal are allocated to the subcarriers of the downlink signal for both antennas.

As shown in the figure, the y-axis represents the subcarrier allocation of the reference signal (every sixth subcarrier) and the x-axis represents the time interleaving. Note that the variation of the reference signal between antenna 0 and antenna 1 is viewed in terms of both occupied subcarriers and time (symbols).

Figure 1 - Orthogonal structure of downlink reference symbols for two antennas

Error vector magnitude (EVM) is an important system metric for describing RF transmitter performance. Comparing RS EVM and composite EVM not only provides engineers with insight into transmitter hardware design impairments, but also helps diagnose specific impairments such as antenna crosstalk, amplifier gain compression distortion, phase noise, and other error mechanisms.

The following example will illustrate how RS EVM and composite EVM can be used to gain insight into the types of impairments that may affect system performance errors. This example will also focus on the impact of transmit antenna timing errors on reference signal orthogonality and show how to account for this effect when interpreting antenna crosstalk, constellation diagrams, and EVM measurements.

Case Study - MIMO Downlink RF Transmitter Measurements

The four-channel MIMO test setup used in this case study is shown on the left side of Figure 2 and consists of four Agilent signal generators with arbitrary waveform generators and an Agilent four-channel Infiniium 90000A Series oscilloscope. As shown below, multichannel oscilloscopes are well suited for two- and four-channel MIMO measurements because they provide time-coherent multichannel inputs, wide bandwidths to measure RF modulated carriers, and deeper memory to analyze multiple data frames that can be demodulated using the Agilent 89600 Vector Signal Analysis (VSA) software.

The results of a baseline four-channel MIMO measurement using the VSA software and a multi-channel wideband oscilloscope are shown on the right side of Figure 2. The left side of Figure 2 shows the 16 QAM physical downlink shared channel (PDSCH) constellation diagram for two (of four) layers of spatially multiplexed data (Layers 2 and 3 are not shown here). The RF spectrum plot is shown in the upper right of the VSA display, and the error summary table is shown in the lower right of the VSA display. Note that the residual composite EVM for the baseline test case (lower right of the VSA display) is less than 0.8%, indicating that the constellation diagrams for Layers 0 and 1 are clean (left side of the VSA display).

Figure 2 - Four-channel MIMO test setup and baseline measurement results using an Agilent Infiniium 90000A Series oscilloscope

Multichannel oscilloscopes and VSA software are typically used with two- or four-channel IF-RF transmitter/upconverter hardware devices under test (DUTs) for MIMO testing. Since DUTs are not suitable for testing, a four-channel RF transmitter with simulated design impairments is modeled using the Agilent SystemVue simulator. Each transmitter consists of an IF/RF bandpass filter, LO mixer, and power amplifier (PA). The PA specifies the LO phase noise at a 10kHz frequency offset and a 1dB gain compression point. A custom model subnet is used at the output of the transmitter to model antenna crosstalk, and then the simulated IQ waveforms (including the simulated design impairments) are downloaded to the four ESGs using the ESG receiver, as shown in Figure 3.

Figure 3 - Simulated RF transmitter design including phase noise, PA gain compression, and antenna crosstalk impairments [page]

After downloading the simulated waveforms to the ESG, the generated test signals were measured according to the test setup shown in Figure 1. The generated test signals output by the ESG were centered at 1.9 GHz. As shown in Figure 4, these signals were captured by a wideband multi-channel oscilloscope and demodulated by the VSA software.

Figure 4 - Downlink RF Transmitter MIMO Results

Note that the constellation plots for layers 0 and 1 now show severe dispersion (layers 2 and 3 also show similar dispersion, but are not shown). At first glance, this looks very similar to dispersion caused by amplifier gain compression distortion or LO phase noise.

However, the EVM peak was high (43%), so the error vector spectrum (EVM vs. subcarrier) and error vector time (EVM vs. symbol) were evaluated to produce a composite EVM result. This revealed inter-symbol variations in the reference signal, so the downlink file on the VSA was modified to show only the reference signal, as shown in Figure 5.

Figure 5 - Reference signal EVM time

RS EVM time plot shows that one antenna pair performs poorly (reference signal is transmitted on consecutive time slots between antennas 0/1, then between antennas 2/3. RS EVM values ​​are calculated for multiple subcarriers and averaged over the hop paths.)

Figure 6 - VSA MIMO Information Table

To explore this in more depth, we can look at the MIMO Information table shown in Figure 6. This MIMO Information table is very useful in showing the effects of antenna crosstalk:

o Row 1: Crosstalk from transmit antennas 1-3 on Tx1/Rx0, Tx2Rx0, and T3/Rx0 or receive antenna 0

o Row 2: Crosstalk from transmit antennas 0, 2, and 3 on receive antenna 1

o Row 3: Crosstalk from transmit antennas 0, 1, and 3 on receive antenna 2

o Row 4: Crosstalk from transmit antenna 0-2 on receive antenna 3 [page]

We see that even with the crosstalk between channels, the individual RS EVM values ​​are relatively low. As mentioned above and referring to Figure 1, the MIMO reference signals are time- and frequency-orthogonal so that RS EVM is generally not affected by antenna crosstalk, unlike composite EVM, which is affected by antenna crosstalk. However, inspection of the RS timing values ​​in the MIMO information table shows that the timing error between the antenna channel ranges is approximately 2.3µs to 3µs (Tx1/Rx1, Tx2/Rx2, Tx3/Rx3). This is a problem because the timing error can result in RS orthogonality loss when it approaches or exceeds the duration of the LTE cyclic prefix (4.69µs). RS orthogonality loss affects the measurement accuracy, such as the crosstalk values, PDSCH constellation diagrams, and EVM results shown in the MIMO information table.

Consider the effect of timing errors on antenna crosstalk measurements. As long as the delay between channels is much smaller than the duration of the cyclic prefix, the reference signals from different transmit antennas remain orthogonal. However, if this condition is not met, orthogonality is destroyed, resulting in crosstalk between channels. Looking at antenna port 0 again in Figure 1, the signal power at the R1 subcarrier position indicates the presence of crosstalk. Timing errors or delays between channels cause the R1 subcarrier position to contain the power of the previous symbol, which the VSA interprets as crosstalk between channels, resulting in an erroneous reported crosstalk value.

To check the timing error reported in the MIMO information table, an oscilloscope is used to measure the timing error between antenna channels, as shown in Figure 7. The timing error between the ESG generating the antenna 0 signal and the ESG generating the antenna 1 signal is measured to be approximately 2.35 µs, which is related to the RS timing error reported in the MIMO information table.

Figure 7 - Timing error between antenna channels 0 and 1 measured using a wideband multi-channel oscilloscope

Antenna 1, antenna 2, and antenna 3 ESGs are all triggered from antenna 0 ESG. After the oscilloscope measures the timing error, the timing error problem can be solved by adjusting the pattern trigger delay of antenna 1-3 ESGs.

The generated MIMO information table (Figure 8) shows that the timing error is now within 134nS (only 2.8% of the cyclic prefix), ensuring orthogonality between the RS signals. The antenna crosstalk values ​​are now correctly displayed, reflecting the antenna crosstalk that was modeled in Figure 3.

Figure 8 - MIMO information table including correction timing error and RS orthogonality

As shown in Figure 9, after satisfying the RS orthogonality condition, the composite EVM result is now 4.1%, which is much lower than the previously reported 12.5%.

Figure 9 - Composite EVM results including corrected timing error and RS quadrature

System engineers can compare RS EVM results with composite EVM results to determine the impact of different error mechanisms on RF transmitter EVM errors. For example, antenna crosstalk may not affect the RS EVM value, but it will affect the composite EVM. On the other hand, other RF transmitter impairments such as phase noise and PA gain compression can negatively impact both RS EVM and composite EVM.

Summarize

There are many test challenges for four-channel MIMO measurements, making troubleshooting and debugging more challenging. This article describes transmit antenna timing errors that can affect LTE MIMO reference signal orthogonality, thereby affecting measurements such as antenna crosstalk, constellation diagrams, and EVM. Multichannel wideband oscilloscopes are ideal for dual-channel or quad-channel MIMO measurements and can help diagnose timing errors that may exist between transmit antenna channels. By combining a wideband multichannel oscilloscope with VSA software, engineers can measure and analyze MIMO signals from multiple different aspects: time domain, frequency domain, and modulation domain, and troubleshoot and isolate hardware performance issues based on the measurement results. By comparing RS EVM and composite EVM, engineers can understand the impact of different error mechanisms (such as phase noise, antenna crosstalk, PA gain compression) on RF transmitter EVM errors.