2012 IEEE 18th International Mixed-Signal, Sensors, and Systems Test Workshop IEEE 802.11n MIMO Radio Design Verification Challenge and a Resulting ATE Program Implemented for MIMO Transmitter and Receiver Test Neil Wu and Tai-Mo Wang Accton Technology, No.1, Creation Road 3, Hsinchu Science Park, Hsinchu 30077, Taiwan, ROC Email: neil_wu@accton.com.tw, tm_wang@accton.com.tw Abstract IEEE 802.11n has become a popular WLAN standard for the wireless networking industry since its official release in 2009. It adopts MIMO multiple spatial data streams architecture and requires significant efforts in RF radio design verification due to its multiple RF chains which demand multiple Vector Signal Generators for Receive test and multiple Vector Signal Analyzers for Transmit test in the instrument setup. This paper addresses our approach to compile an ATE program using reliable and proven tools and construct a radio test set to validate the program. LitePoint s instruments IQnxn were selected for test set and Qualcomm Atheros AR9390 SoC chip was chosen for a 3x3 MIMO radio design. The ATE main control program was compiled using VC 6.0 (build wsock32.lib for DUT communication API) and VC++ 2005 (build IQmeasure.dll for IQnxn measurement API) as development platform. IQmeasure Library API provided by LitePoint contains all the measurement functions of 802.11a/b/g/n signals on instruments IQnxn and communication API. Windows socket wsock32.lib is the library and basic element that controls the DUT from PC via Ethernet. A couple of test cases were executed to test the functionality and effectiveness of the ATE program. These test cases are also addressed in the paragraphs. Keywords IEEE 802.11n; MIMO RF Test; MIMO Radio Design Verification Test; AR9390; IQnxn; ATE Program I. INTRODUCTION Since the initial release of IEEE 802.11 [1] in 1997, wireless LAN has been increasingly popular in the wireless networking era. The data rate enhancement standards 802.11 a/b/g [2-4] were announced a few years later with highest rate 54 Mbps. However, due to vast demands on wireless data transmission, these legacy modes no longer meet the requirement shortly. Therefore a High Throughput (HT) 802.11n standard [5] was released in 2009. The 802.11n-2009 adopts a MIMO (Multiple Input Multiple Output) multiple spatial data streams approach and several enhancement actions to boost data rates enormously to 600 Mbps for a 4x MIMO case. One of the most impressive portions in 802.11n is lots of MCS (Modulation and Coding Scheme) data rates available, including HT20 MCS0~MCS31 for 20 MHz channel bandwidth and HT40 MCS0~MCS31 for 40 MHz channel bandwidth. Additionally, all the radio hardware must be backward compatible to early standards. Therefore, the 802.11n available data rates are about 9 times the number of legacy modes, for both 5 GHz and 2.4 GHz radio bands, for the highest 4x MIMO case. Table 1 shows the comparison. Because RF design verification always seeks for maximum testing coverage, including maximum number of channel and power level against each data rate, consequently required verified items grow into a tremendous number. This becomes a very heavy burden for RF hardware verification and demands an ATE program to speed the MIMO RF testing. Table 1 Contents of data rate of various 802.11 standards 802.11 802.11n (4x MIMO) 802.11n (4x MIMO) 802.11a 802.11b 802.11g standard 2.4GHz 5GHz Data Rate 6,9,12,18, 6,9,12,18, MCS0,MCS1,MCS2,, MCS0,MCS1,MCS2,, 1,2,5.5,11 (Mbps) 24,36,48,54 24,36,48,54 MCS30,MCS31 MCS30,MCS31 Number of 8 4 8 64* 64* Data Rate * Note that the numbers include both 20 MHz (HT20) and 40 MHz (HT40) channel bandwidth. 978-0-7695-4726-8/12 $26.00 2012 IEEE DOI 10.1109/IMS3TW.2012.21 55
II. MIMO RF TEST A critical feature for MIMO RF test is that multiple RF signals must be present for multiple data streams in both transmit and receive tests, i.e., it requires multiple vector signal generators (VSGs) for receive test and vector signal analyzers (VSAs) for transmit test. MIMO RF test methodology was addressed by many equipment suppliers [6-10], however, most do not have realized 802.11n MIMO test program. LitePoint provides an IQDVT test program [11] for relatively extensive design verification, but no flexibility for user to modify any contents and no match to the fast evolution of 802.11n chipset and utility tool, and also costly. Therefore the best strategy for equipment users is to compile the MIMO test program by their own specific flow and purpose. A capable MIMO RF test program must be able to verify following classified items in Table 2, which partially specified in 802.11n-2009 Sections 20.3.21 and 20.3.22 and partially not specified but frequently used in high level 802.11n Wireless LAN product definitions. The implementation of test program includes two main choices: the test equipment and the radio chip. In this test case, we adopt LitePoint equipment IQnxn [12] which is a one-box tester with VSA and VSG together (for 802.11 only). This IQnxn is suitable for a pile of up to 4x tester for a 4x4 MIMO radio test. On the 802.11n chip vendor, we pick QCA (Qualcomm Atheros) because her chips are recognized as very suitable for high level product design due to delicate RF power calibration and power setting algorithm. QCA s state-of-the-art AR9390 [13] for 3x3 MIMO is hence selected. In the later paragraphs, a 3x3 case is addressed although the principle can be applied straightforward to a 4x4 MIMO. Table 2 Testing parameters classified against 802.11n-2009 standard Transmit parameters Transmit power level Transmit power accuracy Transmit power linearity Transmit EVM (Error Vector Magnitude) Transmit spectral mask Transmit spectral flatness Transmit center frequency tolerance Transmit center frequency leakage Receive Parameters Receive sensitivity Receive maximum input power Receive adjacent channel rejection Receive nonadjacent channel rejection Inter-band signal rejection Parameter Notes Not specified in standard The required maximum power level varies from case to case, depending on the product demand (external power amplifier may be needed in the design) Not specified in standard This depends on Silicon chip s capability, including RF power calibration and RF power setting algorithm Not specified in standard This is usually required by higher level products for a wider dynamic range, This is required against desired power levels and MCS rates Test spec are usually beyond the standard due to Silicon chip s advancement Not specified in standard. This is critical in dual-band (2.4GHz /5GHz) dual radio concurrent operation 56
Figure 1 Measurement system hardware setup. This case adopts DUT (QCA AR9390) for a 3x3 MIMO and 3x LitePoint s IQnxn piled up. III. IMPLEMENTATION OF ATE PROGRAM An ATE measurement system has hardware test set and software program. The hardware test set for a 3x3 MIMO is shown in Figure 1. This measurement setup consists of PC/NB, test equipments (LitePoint s tester IQnxn), DUT (RF board of QCA AR9390), adaptor board (to PC/NB) or digital main board (of Access Point) for DUT, switch, and all necessary RF accessories. Figure 2 MIMO ATE program architecture. QCA provides a utility tool ART2 (Atheros Radio Test 2nd generation) [14] for AR9390 series chipset testing. ART2 accesses the DUT through an ART GUI. Accton s MIMO ATE program adopts the other approach and its architecture is shown in Figure 2. There are two setup files and a main control program for this ATE operation. First setup file (artsetup) is used to set test environment, including device ID, tester IP, tester configuration such as 1x1, or 2x2, or 3x3 MIMO, DUT tested through a remote (via Ethernet) or local mode (via PC/NB slot), etc. Second setup file (calsetup) is used to set DUT relevant parameters including calibration loss (insertion loss from DUT to IQnxn), test standard (a, b, g, or n), test channel, test data rate, equipment measurement option setting (phase correction, symbol timing correction, full packet channel estimate, ), tested parameter upper or lower limit, DUT transmit power setting, DUT power swept range setting, IQnxn power swept range setting for receive test, etc. The main control program was compiled using VC 6.0 (build wsock32.lib for DUT communication API) and VC++ 2005 (build IQmeasure.dll for IQnxn measurement API) as development platform, because it is fully compatible with QCA s ART software architecture. IQmeasure Library API provided by LitePoint contains all the IQsignal (an application program for LitePoint instruments) measurement functions of 802.11 a/b/g/n signals on IQnxn and 57
communication API. Windows socket wsock32.lib is the library and basic element that controls the DUT from PC via Ethernet. Non blocking mode was designated in our program, and a thread is established to watch socket feedback messages from DUT during program running time. Log files are generated to record the results and any error message. For a DUT local connection mode (RF board adapted to a PC/NB directly), the ATE program accesses the AR9390 through QCA MDK (Module Development Kit) driver accompanied with NART.exe thru Windows socket wsock32.lib. NART contains hardware specific code needed to control the QCA radio chipset and runs in a slave mode and takes its commands from a socket. On the other hand, for a DUT remote connection mode (RF board adapted on a digital main board), the AR9390 is accessed through the Ethernet by assigning a dedicated socket port from the host PC/NB. IV. VALIDATION OF ATE PROGRAM This ATE program is designed specifically to test RF specifications against radio channel, data rate, and power level. It can also be used to set AR9390 to send swept power for transmit linearity test or IQnxn to send swept power for receive sensitivity test. Test case 1: Transmit parameters of a 5GHz radio board was tested by setting AR9390 to individual target power level for a list of channel and data rate. Figure 3 is a partial portion from a test log file. This captured list includes several data rates of radio chain (i.e., physical path) 0, 1, and 2, at channel 5200 MHz. Test case 2: Transmit power linearity of a 5GHz radio board was tested by setting AR9390 to swept output power. Figure 4 is a partial list from a linearity test log file of AR9390 with MCS0 data rate (single data stream) of radio chain2 at channel 5825 MHz. Notes that the measured power versus target power. Test case 3: Rx sensitivity was obtained by forcing IQnxn to send swept power over a range pre-set by the setup file (calsetup). At each input signal strength (iss) level, IQnxn sent 100 packets. The threshold is packet success rate (psr) at 90%. Figure 5 is a partial portion of HT40 MCS22 and MCS23 (3x data streams) of AR9390 test log file at 5520 MHz. Sensitivity is the iss level right higher than the level of psr failure. Test case 4: A data parsing utility written using Microsoft Excel VBA is associated with the MIMO test program. In Figure 6, EVM of a specific radio channel (5180 MHz) and data rate (HT20 MCS7) of radio chain0, 1, and 2 versus transmit target power was analyzed. Test case 5: RxSEN (Receive Sensitivity) was analyzed by the data parsing utility (5a) In Figure 7, RxSEN sorted by a specific data rate (HT20 MCS1) of radio chain0 and 1 versus all radio channels. (5b) In Figure 8, RxSEN sorted by a specific radio channel (2412 and 5180 MHz) of radio chain0 and 1 versus all one stream data rates. V. CONCLUSION We proposed an ATE program architecture to verify MIMO RF specifications against radio channel, data rate, and power level. Essential program components, radio test setup, and DUT radio design with AR9390 are addressed in the paragraphs. A couple of test cases to demonstrate the program capabilities are also shown. In our practice, the ATE program did save radio test time from weeks with manual operation to a few overnight running, also with enhanced accuracy. In addition, data parsing is one part of ATE program and reduces post-processing efforts. Figure 3 Captured partial list of AR9390 test log file of several data rates for radio chains 0, 1, and 2, at channel 5200 MHz. 58
Figure 4 Captured partial portion of AR9390 linearity test log file with MCS0 data rate (single data stream) of radio chain2 at channel 5825 MHz. Figure 5 Captured partial portion of AR9390 sensitivity test log file at 5520 MHz for data rates HT40 MCS22 and MCS23 (3x data streams). Figure 6 EVM of radio chain0, 1, and 2 measured in channel 5180 MHz and HT20 MCS7 data rate versus transmit target power. Figure 7 RxSEN of radio chain0 and 1 sorted by HT20 MCS1 data rate versus all radio channels. 59
Figure 8 RxSEN of radio chain0 and 1 sorted by radio channels 2412 MHz and 5180 MHz, respectively, versus all one stream data rates. REFERENCES [1] IEEE Standard 802.11-1997, Wireless LAN Medium Access Control and Physical Layer specifications. [2] IEEE Standard 802.11a-1999, Wireless LAN Medium Access Control and Physical Layer specifications: High-speed Physical Layer in the 5 GHz Band. [3] IEEE Standard 802.11b-1999, Wireless LAN Medium Access Control and Physical Layer specifications: Higher-Speed Physical Layer Extension in the 2.4 GHz Band. [4] IEEE Standard 802.11g-2003, Wireless LAN Medium Access Control and Physical Layer specifications, Amendment 4: Further Higher Data Rate Extension in the 2.4 GHz Band. [5] IEEE Standard 802.11n-2009, Wireless LAN Medium Access Control and Physical Layer Specifications, Amendment 5: Enhancements for Higher Throughput. [6] Agilent, MIMO Wireless LAN PHY Layer [RF] Operation & Measurement, Application Note 1509, literature number 5989-3443EN, 2008 [7] Rohde & Schwarz, Guidelines for MIMO Test Setups Part 1, Application Note 1GP50_0E, 2010. [8] Rohde & Schwarz, Guidelines for MIMO Test Setups Part 2, Application Note 1GP51_1E, 2010. [9] Carr, M, (of Teradyne) Volume Testing Microwave Devices, pp. 101-109, IEEE Microwave Magazine, Aug. 2010. [10] Keithley, Advanced Measurement Techniques for OFDM- and MIMO-based Radio Systems, Application Handbook, 1st edition, 2009. [11] LitePoint, IQfact for Design Verification Testing (DVT), Product Brochure, 2007. [12] LitePoint, IQnxn MIMO Test Solution, Data Sheet rev2.1, 2007. [13] Atheros, AR9390 Single-Chip 3x3 Spatial Stream MIMO MAC/BB/Radio with PCI Express Interface for 802.11n 2.4 or 5 GHz WLANs, Data Sheet, Apr. 2011. [14] Atheros, AR93xx Atheros Radio Test 2 (ART2) Reference Guide, 2010. 60