Link to Document

Multiple-input, multiple-output (MIMO) radio technology is gaining momentum and it's becoming a reality with the advancement of the IEEE 802.11n WLAN standard. Even though version 2.0 of the 802.11n draft standard was approved in March, there is alot of pre-n equipment available in the market. MIMO technology significantly increases throughput without increasing power or bandwidth. The increased complexity of MIMO WLAN chipsets may increase the production line test cost, which adds cost to the final product. This is an unacceptable outcome because the WLAN consumer market is not ready to embrace higher costs. So the challenge rests with manufacturing test equipment vendors to improve MIMO test times without sacrificing performance or quality.

There are several manufacturing test equipment vendors in the market who provide MIMO WLAN solutions ranging from legacy to new test methodologies. Hence, it is important for WLAN manufacturers to select a test methodology that does not add any additional test time to the legacy system.

Four different test methodologies can be used for testing MIMO WLAN chipsets and products:

1. Multiple vector signal generators (VSGs) and vector signal analyzers (VSAs) test solution.

2. One-box VSG and VSA solution plus combiner and fast switch.

3. One-box VSG and VSA solution plus combiner.

4. One-box VSG and VSA solution plus fast switch.

Multiple VSG and VSA test solution

In the multiple VSG and VSA test solution, each receiver and transmitter pair of the device under test (DUT) is connected directly to its own VSG and VSA pair. The transmitters and receivers can be tested one at a time, all simultaneously or in any other combination. Using this configuration, key MIMO parameters can be measured, e.g., power, spectrum, transmitter impairments, including transients and transmit chain interactions, transmitter quality indicator EVM, RF chain isolation, and receiver sensitivity.

Figure 1 shows the multiple VSG and VSA test solution like the LitePoint IQnxn test solution for a 3 × 3 configuration, testing a 2 × 3 DUT. When testing the DUT transmitters, the VSAs of the test system are active. Each transmitter is connected to its corresponding VSA. Data capture is performed for all transmit chains simultaneously and the captured data from the VSAs is routed to a comprehensive software package for analysis. With this test system and a comprehensive analysis software combination, the quality of each transmitter and their interaction with each other can be analyzed in detail. The VSA analysis also includes a full data demodulator that helps verify that the transmit signal is correctly constructed, a feature that is helpful during development. Verifying the CRC checks the correctness of the demodulated packet data. The packet data can be saved to a file for comparison with the transmitted data.

Using this test system, any combination of transmitters can be active. The dynamic combination of multiple VSAs and comprehensive analysis software can measure the following transmit parameters in a single capture:

• Tx power throughout the packet;

• Tx channel response and spectral flatness of each transmitter;

• Tx isolation between transmitters;

• Tx frequency variations;

• Tx I/Q imbalance, phase and amplitude, each transmitter;

• Tx local oscillator (LO) leakage, each transmitter;

• Tx signal quality or EVM for each transmitter;

• Tx phase noise, each transmitter and combined;

• Tx compression, each transmitter, by showing the CCDF;

• Tx power variation during the packet;

• Tx symbol clock offset;

• Tx mutual packet alignment in time; and

• Tx payload verification.

When measuring the DUT receivers, the VSGs of the multiple VSG and VGA test solution are active. Each receiver is connected to its corresponding VSG. The test control software loads the waveforms into all VSGs, individually sets the RF signal level for each VSG and the RF frequency common to all VSGs. The multiple VSG and VSA test system can be set up to transmit the loaded waveform in an endless loop or for a user-specified number of transmissions. The waveforms sent by a VSG can either be from a single transmitter using an ideal channel, from a single transmitter using a multipath channel, or from multiple transmitters each with its own multipath channel and power level. This allows measuring the packet error rate (PER) of the receiver hardware under realistic MIMO and legacy multipath channel conditions.

During development it is important to assess the receiver sensitivity to transmitter impairments. Using the multiple VSG and VSA test solution, MIMO receivers can be analyzed in detail. Transmitter impairments such as LO leakage, I/Q imbalance, transmitter compression, phase noise, added noise or carrier frequency offsets can be included in the signals generated by the test system. Most impairments can be individually specified for each transmitter.

With the multiple VSG and VGA test solution, the following receiver tests can be performed:

• PER as function of signal level or SNR;

• PER for legacy and MIMO multipath channels;

• receiver sensitivity to transmitter impairments, e.g., frequency offsets, I/Q imbalance, LO leakage, etc.;

• clear channel assessment;

• RSSI calibration; and

• isolation between receiver RF paths.

This approach is used in design verification, debug and quality assurance test stages. For manufacturing lines, this solution is expensive and may not satisfy the targeted return on investment (ROI).

Single VSG/VSA plus combiner and fast switch

For production testing, a fully parameterized performance assessment is not required. The main goal of production testing is to verify that the unit has been correctly assembled and that it meets the specified performance. Even for MIMO WLAN systems, this can be achieved with a single VSG/VSA combination similar to legacy WLAN production testing. The first single VSG/VSA configuration considered here is a one-box VSG and VSA solution plus combiner and fast switch like the LitePoint IQflex/IQview plus multiport test adapter (MPTA). For convenience, the fast switch and combiner will be referred to as MPTA.

The MPTA consists of a fast switch, an attenuator in each RF signal path and an intelligence sequence controller as shown in Figure 2. This setup can be configured into a static mode with each switch closed or open and each attenuator set to a specific value. The setup can also be configured into a dynamic mode. In the dynamic mode, a series of configurations, or states, is defined. Each configuration has its own switch and attenuator settings. The transition from state to state is controlled by the intelligent MPTA itself based on its signal inputs.

For transmitter testing, the MPTA is configured to dynamically switch between the transmitters. The MPTA starts with routing the signal from T×1 to the VSA. The start of the packet will trigger the start of the capturing process in the VSA. Once sufficient data has been captured from T×1, the VSA stops capturing and the MPTA switches to T×2. The next packet will trigger the capture process in the VSA again. This capture and switch process will continue until the capture memory is filled with the predefined number of samples.

As the switching is controlled by the RF signal, the extra time required for capturing the MIMO signal compared to the IQnxn system is just the extra time to transmit two packets instead of one. Processing and analysis time for this system is the same as a multiple VSG and VSA test system.

For transmitter testing, the VSA treats these sequentially captured signals as a MIMO signal even though the captures from different transmitters are taken sequentially instead of simultaneously as in the multiple VSG and VSA test system. Most of the transmitter tests performed by the multiple VSG and VSA test system can be performed with the one-box VSG and VSA plus combiner and fast switch test solution. Specifically, the EVM, power amplifier compression, and isolation are calculated per RF chain. Still, several differences exist between two solutions. Even though performing the classic MIMO EVM calculation assesses the quality of each transmitter, the EVM will be limited by the RF chain isolation if the isolation between transmitters is poor. This limitation does not apply if the payload data is kept constant between consecutive packets. In addition, the MIMO EVM calculation does have to be adapted to track the phase trajectory of each transmitter individually in contrast with the standard MIMO EVM calculation, which tracks all the transmitters with the same phase corrections. These system measurements are not as comprehensive as IQnxn measurements. However, it provides an excellent manufacturing solution. The measurements not supported by this solution are: 1) the payload data cannot be recovered, 2) dynamic interaction between the transmit chains cannot be evaluated, and 3) packet alignment between transmitters cannot be measured.

The MPTA can also be configured to receive the signals simultaneously from multiple transmitters. This mode is discussed further in the section describing the combiner-only configuration.

For receiver testing the one-box VSG and VSA solution plus combiner and fast switch generates a single transmit signal. This signal can be a legacy or a one-stream MIMO signal and this signal can be fed into any or all receivers. When feeding the signal into one receiver at a time, the RSSI indicators of the DUTs allow measurement of the RF chain isolation.

There are two modes to test the receiver PER or sensitivity. In one mode, the switch is set into a static state. In this state, the sensitivity of each receiver can be measured individually by routing the VSG signal to one receiver at a time without having to enable or disable the receivers in the DUT. This helps to measure the isolation between receive chains. In addition, the improvement in sensitivity due to maximum ratio combining (MRC) can be verified by feeding the VSG signal into all receivers simultaneously. This sensitivity improvement verifies a substantial part of the MIMO signal processing. The driver of the DUT should be able to report the number of packets received in error or the number of packets without error.

Another mode is more comprehensive exercising the transmit/receive switching to test the receiver PER and sensitivity of the DUT. In this mode, the switch is configured to measure the acknowledgements (ACKs) sent by the receiver. These ACKs are only sent when no errors are detected. This way, the PER can be measured for different settings of the attenuators and for different receiver configurations, i.e., with one, two, three or four receivers enabled. By measuring the PER at different signal levels the sensitivity of any receiver combination can be measured accurately. This whole sequence can be carried out by the switch/attenuator combination automatically. The test software defines the test sequence and starts the testing using only this sequence. After that, the MPTA goes through the sequence automatically. Not only does this mode verify the Tx/Rx switching and the MRC sensitivity improvement, but there is no time penalty due to multiple control and DUT software interactions.

One-box VSG/VSA plus combiner

In this configuration, a passive splitter/combiner replaces the combiner and fast switch configuration as shown in Figure 3. This is the lowest-cost MIMO test solution with good MIMO manufacturing test coverage.

The signals from the transmitters are summed in the combiner for transmitter testing. The IQsignal analysis software derives some of the transmit signal properties that are common to all transmitters from the legacy or the MIMO preamble. Examples include the carrier frequency dynamics at packet start-up and frequency offset at the end of the preamble. Other properties of the transmitted signals can be derived for each transmitter from the MIMO preamble, such as the Tx power, I/Q imbalance, and spectral flatness. The overall transmitter quality indicator EVM is measured by comparing the combined signal with the ideal combined signal using the channel responses estimated from the MIMO preamble. This measurement requires the data content of the payload to be known prior to the analysis software. However, the transmitter scrambler is allowed to vary from packet to packet as the starting state is derived in the analysis software.

Any transmitter impairment that causes its signal quality to fall below its limits degrades the combined signal EVM. This includes compression and I/Q imbalance. Compared to one-box VSG and VSA solution plus combiner and fast switch, this system cannot measure the RF chain isolation nor attribute impairments to a specific transmit chain. RF chain isolation may not be a serious concern if the assembly reliably provides 20 dB isolation, which is a number often used as a specification for sufficient isolation between chains to provide good MIMO performance.

For receiver testing, the same signal is routed to all receivers in the DUT. If the DUT enables a single receiver, the single receiver sensitivity can be measured. If all receivers are enabled, the improvement in sensitivity due to maximum ratio combining (MRC) verifies a substantial part of the MIMO signal processing.

In the one-box VSG and VSA solution plus combiner configuration, transmitters and receivers are tested in the MIMO mode with quality test parameters, reasonable test time, and lower cost. LitePoint IQflex plus combiner supports this test solution methodology providing reliable and excellent performance test solution. The shortcoming of this methodology is that it requires known data for MIMO transmitter measurements and doesnot measure isolation between RF chains.

Single VSG/VSA plus fast switch

In this configuration, as illustrated in Figure 4, the MPTA is replaced with a switch without combiner. Similar configurations have been developed by ODMs using IQflex and off-the-shelf RF switches and with the switch control under direct control of the test software.

For transmit testing, this configuration cannot verify that all transmitters are transmitting in a time-aligned fashion. The main difference between this configuration and one-box VSG and VSA solution plus combiner and fast switch configuration is the receiver testing. Contrasted with the MPTA, this configuration cannot feed the VSG signal into all DUT receivers simultaneously. Therefore, it lacks the ability to verify the MRC process in the MIMO receiver of the DUT.

When this configuration is assembled from off-the-shelf components, it cannot exercise the Tx/Rx switching when measuring the Rx PER and is likely to require longer test times since the switching has to be performed by the test controller software. Especially, accurate receiver sensitivity measurements are likely to be too slow for production testing.

The advantage of this configuration compared to the combiner configuration is that the isolation between RF chains can be measured. Compared to the MPTA configuration, however, the receiver testing is severely limited.

Comparison

Table 1 lists the measurement parameters and the capabilities of each solution methodology.

Conclusion

The IQnxn system is a comprehensive test solution that addresses MIMO WLAN test challenges. It is a powerful test solution typically used during R&D, DVT, and for manufacturing QA and debug stations.

Likewise, IQflex or IQview plus combiner methodology provides the lowest-cost MIMO test solution with good test performance, low test times, and good test coverage. For improved test coverage and without test time penalty, the IQflex plus MPTA adds RF chain isolation, individual transmit impairment identification, and enhanced receiver testing.

Table 1. Capability of each test configuration

Parameter	Multiple VSGs & VSAs	Single IQflex + MPTA	Single IQflex + combiner	Single IQflex + fast switch
TX power, each	ϫ	ϫ	ϫ	ϫ
TX spectrum, combined	ϫ	ϫ	ϫ
TX spectrum, each	ϫ	ϫ	ϫ,1	ϫ
TX carrier leakage, each	ϫ	ϫ	ϫ,1	ϫ
TX EVM, composite MIMO		ϫ,2	ϫ,2
TX EVM, individual MIMO	ϫ	ϫ,3		ϫ,3
TX isolation	ϫ	ϫ		ϫ
TX spectral flatness, each	ϫ	ϫ	ϫ	ϫ
TX IQ imbalance, each	ϫ	ϫ	ϫ	ϫ
TX phase noise	ϫ	ϫ	ϫ	ϫ
TX CCDF, each	ϫ	ϫ		ϫ
TX symbol clock offset	ϫ	ϫ	ϫ	ϫ
TX mutual packet alignment	ϫ	ϫ	ϫ	4
TX PSDU data verification	ϫ
RX sensitivity, individual chains, AWGN channel	ϫ	ϫ
RX sensitivity, AWGN channel, MRC combining	ϫ	ϫ	ϫ
RX sensitivity with MIMO multipath channels	ϫ
RX isolation	ϫ	ϫ		ϫ
RX CCA	ϫ	ϫ	ϫ	ϫ
RX MIMO multistream data path verification	ϫ
TX/RX switching during PER test		ϫ
IQsignal GUI	ϫ
1 Requires DUT TX on/off control. Repeating captures for different configurations may increase test time. 2 Requires known data in PSDU to be transmitted by DUT. 3 If random data in PSDU, the EVM is limited by the TX isolation. If consecutive packets have same data in PSDU, EVM is not limited by isolation. Captures from different transmitters are separated in time. Each stream is affected by different phase trajectories, which have to be tracked individually before applying the inverse channel. For cases with substantial frequency variations during the preamble, this may cause incorrect results. 4 If transmitters have low isolation, alignment between transmitters can be verified.

ABOUT THE AUTHORS

Dick Walvis is a Fellow, Signal Analysis at LitePoint in Sunnyvale Calif. He focuses on the development of signal analysis architecture and algorithms for physical layer wireless test and measurement equipment. Walvis obtained his Ingenieurs degree (MSEE) at the Delft University of Technology, the Netherlands.

Sireesha Mallipeddi is a director of product management at LitePoint in Sunnyvale, Calif. She manages several products focusing on wireless manufacturing and R&D markets. Mallipeddi obtained her MSEE and MBA degrees from the University of Maryland.