Introduction to IEEE 802.11ac manufacturing test requirements
Robin Irwin, Aeroflex, Inc.- April 24, 2012
|A version of this article also appeared in the June 2012 issue of Test & Measurement World. See the PDF .|
802.11ac is a WLAN standard being developed by the IEEE 802.11ac Task Group (TGac) aimed at delivering very high throughput (VHT) local wireless connectivity supporting data rates up to 10 times those of WLAN 802.11n HT (high throughput) and a signal bandwidth up to 160MHz, 4 times that of 802.11n and MIMO with up to 8 streams. Chipset vendors are releasing reference designs that have an 80MHz bandwidth. The standard also provides an optional 160MHz bandwidth, and it is expected that this will be adopted over time in the same way that 802.11n evolved into more complex and effective MIMO (multiple-in, multiple-out) implementations.
Manufacturing Test Requirements
From a manufacturing test perspective, the evolution of WLAN standards is something that must be understood and planned for ahead of time. For example, with a mandatory 80MHz channel bandwidth, 802.11ac test equipment is clearly going to require wider bandwidth analysis and generation for testing devices. (Similarly, when draft 802.11n products were being produced, there was a need to deploy equipment that could adequately test the 40MHz bandwidth. For those familiar with 802.11n, 802.11ac has many similarities.)
The Test Setup
When 802.11n was introduced, a main area of concern was the choice of test setup to evaluate multiple antennas for MIMO. This presented manufacturing test engineers with a new challenge – how to test multiple radios while maintaining the same cost of test equipment and with minimal impact to test time/throughput. Fortunately, 802.11ac has similar MIMO requirements, so existing 802.11n test setups are applicable. For the purposes of focusing on the test items that the 802.11ac specification outlines, the test setup assumed will be as follows (Figure 1).
In this example, the signal analyzer provides the means of transmitter testing while the signal generator delivers the output required for receiver testing. For the sake of simplicity, the test equipment in this example is configured to present one RF port (TX/RX) to the device that can potentially test one of the following scenarios:
- a single 802.11ac SISO device (1x1) with one RF chain/radio
- a single 802.11ac RF chain/radio on a MIMO device (nxm), testing each chain as a separate radio on the device (n transmitters can be tested in turn, for example, by sequentially switching to each chain in turn).
The Test Plan
Similar to 802.11n test plans, an 802.11ac test plan is likely to include a range of tests that cover the expected use of the device. In the same way that 802.11n test plans contained test items that allowed a device to operate in both legacy and HT (high throughput/green-field) modes, an 802.11ac test plan will do the same to address backwards compatibility (an important feature of the specification). Engineers are likely to bias this test towards 802.11ac signals wherever possible. Clearly this is the main targeted use case in order to address new designs and the mandatory requirements of the new standard. For example, it is likely that engineers will define spectral mask requirements for at least an 80MHz transmission and verify modulation accuracy at MCS (modulation and coding scheme) 9 for 256 QAM, with receiver testing also using an 802.11ac MCS 9 signal.
The specific test items contained within a test plan are very similar to 802.11n. Table 1 includes a list of test items from the standard :
Note that adjacent-channel and nonadjacent-channel rejection tests require an interfering signal from an addition signal generator. These tests are verified prior to manufacturing and are not considered a production test requirement, so they are not covered in this analysis.
Transmit Spectrum Mask
The spectral mask test verifies that the output spectrum from the device does not interfere with other devices, and that it meets the mask requirements set in the specification. This test is typically performed at maximum power output from the device. Figure 2 and Table 2 summarize the requirements for each signal bandwidth.
The engineer should check that the spectrum meets the dBr (decibels relative to reference level) mask requirements in the above example where they are relative to the maximum spectral density of the signal. Or, depending upon the power level of the input signal, the engineer should check the dBm/MHz requirement (which dictates the highest mask value allowed).
Finally, the specification describes the procedure for the 80 + 80MHz non-contiguous case. The mask is constructed from two 80MHz masks that are then combined or overlapped. The mask limits are calculated as shown in Table 3.
In terms of making a spectral mask measurement with test equipment, it should be a matter of defining the mask required and the equipment returning (as a minimum) a pass or fail versus the mask.
Spectral flatness is a measure of the deviation of each sub-carrier from the average power. This is done using BPSK (binary phase-shift keying) modulated packets. As this is dependent on the signal bandwidth in question, the specification limits are per subcarrier. Figure 5 illustrates Ei,avg as the average constellation energy of a BPSK modulated subcarrier i in a VHT data symbol. The mask limits in each case (shown in red) are ±4 or -6 depending upon whether the reference is to the central region (-B to B) or the outer region (-B to –C & B to C) of subcarriers. Table 4 provides the subcarrier indexes depending upon bandwidth of transmission. Note that subcarrier position B in Figure 5 represents the start of the outer region (inclusive).
The 160MHz bandwidth does not use the -4 dB limit. An example of a measurement for an 80MHz signal is shown in Figure 6.
Transmit Center Frequency Tolerance
This test looks at the frequency error (with respect to the desired carrier frequency) from the transmitter, normally produced as a demodulation of the modulated signal. The criteria for a pass is < ±20 ppm (0.002%). As an example, this would be ±275 kHz @ 5500MHz.
Symbol Clock Frequency Tolerance
Symbol clock frequency tolerance is a measure of the symbol clock frequency offset from the desired symbol clock frequency. The pass criteria is < ±20 ppm. This checks for any time varying frequency changes in the local oscillator. If the frequency error is measured, there is also no need to return this measurement.
Transmit Center Frequency Leakage
The transmit center frequency leakage test is designed to check for any unwanted energy at the center frequency of a modulated signal. This leakage can sometimes cause problems for receivers.
Leakage is defined according to three conditions depending upon the position of the LO (carrier). For example, an LO would not be in the center of a transmission bandwidth if a 20 or 40MHz transmission was used in an 80MHz channel. See Table 5.
PT = total transmit power, N is data plus pilots tone. The resolution bandwidth is 312.5 kHz.
Transmitter Constellation Error
Transmitter constellation error and transmit center frequency leakage (which applies to all bandwidths) both form the requirements for testing the modulation accuracy of the transmitter. The specification states the number of spatial streams under test shall be equal to the number of antennas and also the number of test equipment input ports. Table 6 shows the relative constellation error (RCE) in dB for the different MCS. The measured result should not exceed the data-rate dependent value.
The payload data must be random and at least 16 data OFDM symbols long. The test must be performed over at least 20 frames. An example of modulation accuracy results for a MCS 8 signal returning RCE is shown in Figure 7 and the constellation in Figure 8.
Each measurement result returned in this section is reported from the device itself. The test equipment must be setup with the correct signal for stimulation of such tests.
Signal generation packages are widely available from test vendors to allow for the design of specific WLAN signals. In general, most chipsets and commercial devices are set up during the manufacturing process so that they need very little in the way of specific receiver parameters. The most common requirements are to adjust the size of the data and perhaps use a specific MAC address. The following is an example requirement for an MCS 7 64 QAM 5/6 coding rate:
- Standard: 802.11ac
- Channel BW: 80 MHz
- MCS Index: 7
- Spatial Streams: 1
- TX Antennas: 1
- Data: 400 symbols
- Idle Time: 1us
Starting with the data field, we need to understand what 400 symbols equate to for a MCS 7 signal. The number of octets to be defined is dependent on MCS. Referencing table 22-41 , for 1 stream with an 800ns guard interval, there are 1170 bits per OFDM symbol, which is 146.25 octets. For 400 symbols this equates to 58500 octets. The data source is likely to be PRBS9 as per the specification. These parameters can be configured into a signal design package (Figure 9) and the PPDU (PLCP protocol data unit) verified (Figure 10).
This analysis shows that the data field begins at 3200 and ends at 131519 chips, which are 128320 chips. For an 802.11ac VHT (very high throughput) signal with 80MHz channel bandwidth and a sampling rate of 80Msps, the duration of 1 marker chip is 0.0125 us. With 4us per symbol for OFDM, then, there are 320 chips per symbol. For 400 symbols, this works out as 128000 chips, so the value determined for the data field look correct. In fact, it is 1 symbol (320 chips) longer once designed and packaged.
A common requirement is to setup WLAN signals with a broadcast MAC header. Referring to IEEE 802.11 2007 , Figure 11 (Table 7-1) shows how to setup a beacon (broadcast), that relates to Figure 12 (Figure 7-2) .
This example shows that the bits b4 to b7 need to be configured for a beacon in order to define the subtype field in the frame control. The easiest way to do this is to use least significant bit notation and enter “0000000100000000” into the frame control field. This is shown in a signal design package (Figure 13).
By selecting and configuring any other parameter required, the signal can then be generated and packaged for playback for receiver testing.
Receiver Minimum Input Sensitivity
Minimum input sensitivity testing is a key verification test of the performance of the receiver to successfully demodulate an 802.11ac signal. The PER (packet error rate) should be less than 10% for a PSDU length of 4096 octets.
The specification suggests 4096 octets. Taking MCS 7 as an example, there are 1170 bits per OFDM symbol for an 800 ns guard interval, which is 146.25 octets, so 4096 octets is just over 28 symbols. For MCS9 this is just over 21 symbols.
Receiver Maximum Input Level
Opposite to the minimum sensitivity test, this test makes sure that the device can receive when the power level incident to the antenna is comparatively high. The specification asks for this test to be carried out with a PSDU of 4096 octets using any MCS signal but at the higher power of -30dBm. The test limit is again 10% packet error rate.
Being Test-Ready for 802.11ac
80MHz is the minimum requirement with a need to test 80 + 80 MHz and 160MHz scenarios. Some test equipment already deployed in cellular and non-cellular manufacturing lines for signal generation and signal analysis already supports a 80MHz bandwidth. In those cases, the upgrade path is much easier and capability is enabled through software upgrades and options. However, the 160MHz bandwidth presents a different challenge, and hardware upgrades are necessary.
Just like 802.11a/b/g/n, 802.11ac is unlikely to be the only technology a manufacturer is concerned with testing. Indeed, end products may have WLAN as a more complimentary technology, mobile phones being an obvious example. In the early days of WLAN, test equipment solutions were often focused specifically on WLAN testing. Over time, other technologies such as Bluetooth®, GPS, FM and WiMAX® became test requirements alongside WLAN as new chipsets combined all functionalities. A credible test equipment choice is one that can offer this capability while supporting legacy WLAN standards and a breadth of cellular formats in order to be a suitable investment for manufacturers. It is no longer acceptable to support a range of test platforms for a mix of technologies. A modular hardware and software platform can also bring further advantages.
Using capable test equipment, engineers need to know about the new 802.11ac test requirements in order to understand what potential test plans can be deployed, how to address individual test cases, and how to fulfill these tests using available test equipment.
Link to a video demonstration of an 802.11ac analysis .
Institute of Electrical and Electronics Engineers, “IEEE P802.11ac™ D1.0”, May, 2011
Institute of Electrical and Electronics Engineers, “IEEE Std 802.11™ 2007”, June, 2007
About the Author
Robin Irwin is a senior applications engineer team lead supporting the PXI 3000 Series of modular instruments for Aeroflex based in the UK. He provides global application expertise in manufacturing test for wireless multi-communications products. He joined Aeroflex with over seven years of test and measurement experience and with a first class masters degree in electrical and electronic engineering from Queens University of Belfast.