Testing WiMedia UWB
The bandwidths and frequency-hopping behavior of ultrawideband devices impose stringent test challenges—extending from the development of software used in bench instrumentation to the layout of device interface boards used in production.
By Mike Carr, Teradyne -- Test & Measurement World, 3/1/2009 2:00:00 AM
|
See other articles from our March 2009 issue. |
WiMedia UWB (ultrawideband) constitutes a short-range wireless link that transfers data at very high rates. Employing OFDM (orthogonal frequency-domain modulation), a UWB link can transmit error-corrected data at rates between 53.3 and 480 Mbps over short distances—a capability that has already been adopted for Wireless USB and Bluetooth 3.0 applications.
 Figure 1. The spectrum allocation for UWB encompasses six band groups and extends from 3.1 to 10.6 GHz. |
The bandwidths and modulation schemes used in UWB present significant test challenges that extend from the software used to characterize UWB devices on the bench up to the layout of the DIB (device-interface board) that must route 10-GHz signals to and from a DUT (device under test) during production test. The spectrum allocated for UWB goes from 3.1 to 10.6 GHz and is divided into six frequency bands called band groups (Figure 1). Each band group has three sub-bands, with the exception of group 5, which has two. The OFDM symbols can hop across sub-bands in each band group, but they cannot hop between groups.
A hopping sequence is called a TFC (time frequency code) and occurs between OFDM symbol-to-symbol transitions in a predefined order. Each band group supports 10 TFCs (Table 1). Note that TFCs 1 through 4 hop across all three carriers in a band group, while TFCs 5 through 7 stay on one carrier, and TFCs 8 through 10 hop between two carriers. The OFDM signal consists of 128 subcarriers spaced at 4.125-MHz intervals: 100 data-bearing subcarriers, 12 pilot subcarriers, 10 guard subcarriers, and six null subcarriers (Figure 2).
The data subcarriers employ two modulation schemes (Figure 3): QPSK (quadrature phase-shift keying) and DCM (dual carrier modulation). The quality of the UWB link determines which scheme is used. A high-quality link will employ the more efficient DCM scheme with less error correction to maximize the data rate. A lower-quality link will revert to a less efficient, more redundant scheme employing QPSK with more error correction. This dynamic approach causes the data rate to vary between 53.3 Mbps and 480 Mbps.
 Figure 2. The 128 subcarriers of a UWB OFDM signal can be data carriers, pilot carriers, guard carriers, or null carriers. |
In the QPSK mode, two bits are I/Q modulated on each data carrier. For each UWB symbol, the net effect is
(2 bits/data carrier) x (100 data carriers) = 200 bits/symbol.
To improve the BER (bit error rate), UWB employs two types of redundancy: TDS (time-domain spreading) and FDS (frequency-domain spreading). TDS essentially transmits the same data over two bursts. FDS transmits the same data on two carriers within the same symbol. Each of these redundancy schemes cut the raw throughput of 200 bits/symbol in half.
 Figure 3. UWB modulation schemes include QPSK (top) and DCM (bottom). Link quality determines which scheme is used. |
The DCM mode uses a 16QAM technique to modulate four bits on two data carriers spaced 50 subcarriers apart. The net throughput of this redundancy technique is:
(4 bits/data carrier) x (50 data carriers) = 200 bits/symbol.
UWB also uses FEC (forward error correction) techniques to mitigate multipath fading effects as well as other obstacles that may affect the quality of a data transmission. Table 2 lists the data rates achieved with different combinations of modulation schemes, redundancy schemes, and FEC rates.
The duration of each OFDM symbol is 242.42 ns. Each symbol also has a 70.08-ns pad, so the total duration of the symbol plus pad is 312.5 ns. The following calculations show how the data rates are achieved.
To achieve the 53.3-Mbps rate, UWB makes use of QPSK as well as FDS, TDS, and the FEC rate of 1/3. The calculations for achieving this data rate are:
QPSK = (2 bits/data carrier) x (100 data carriers) = 200 bits/symbol,
FDS = (200 bits/symbol)/2 = 100 bits/symbol, and
TDS = 100 bits/symbol/2 = 50 bits/symbol.
FEC works on six-symbol intervals, so 6 symbols x 50 bits/symbol = 300 raw bits.
The FEC rate of 1/3 reduces the 300 raw bits to 100 coded bits. With symbol durations of 312.5 ns, six symbols take 1.875 μs. To transmit 100 coded bits in 1.875 μs, the coded data rate equals 53.3 Mbps.
To achieve the 480-Mbps rate, DCM is employed with no FDS and no TDS. The FEC factor is 3/4. The calculations for this rate are:
DCM = (4 bits/data carrier) x (50 data carriers) = 200 bits/symbol.
FEC works on six-symbol intervals, so 6 symbols x 200 bits/symbol = 1200 raw bits.
The FEC rate of 3/4 reduces the 1200 raw bits to 900 coded bits. To transmit 900 coded bits in 1.875 μs, the coded data rate equals 480 Mbps.
Figure 4 shows six consecutive UWB symbols. The frequency hopping takes place from symbol to symbol. The 70-ns pad occurs when the UWB device changes frequency for a hopping TFC. The settling time for a UWB hop is 9 ns.
 Figure 4. Shown here are six consecutive UWB symbols interspersed with 70-ns pads, during which frequency hopping takes place. |
The front end of a UWB frame consists of 24 synchronization symbols, six channel estimation symbols, and 12 header symbols. The synchronization symbols assist the receiver in timing recovery. The channel estimation symbols characterize the frequency and phase response of the transmission channel and apply that channel information to the header and payload symbols. The header has information about the payload.
Transceiver details and test challenges
The challenge for UWB test engineers is to find cost-effective methods for testing UWB transceivers in as little time as possible. UWB transceivers have a basic I/Q modulator/demodulator structure. Also included is a fast switching PLL (phase-locked loop), which supplies the internal LO (local oscillator) signal that can switch at the 9-ns rate.
Some UWB transceivers may require analog baseband I/Q symbols. Other UWB transceivers may have integrated DACs (digital-to-analog converters) and ADCs (analog-to-digital converters) that require digital baseband I/Q signals. Still other UWB transceivers may be integrated with a baseband processor and require no baseband signals. These types of integrated devices may be implemented as a SIP (system in a package).
For the I/Q type UWB transceiver device, testing can be broken up into two main parts: device transmitter testing and device receiver testing. Transmitter testing typically consists of measuring EVM (error vector magnitude), channel power, spectral mask, I/Q balance, and frequency-hopping settling time. Receiver testing typically consists of measuring gain, noise figure, I/Q balance, PER (packet error rate), and BER.
 Figure 5. When analyzing UWB signals, using the same software from the bench instrumentation on the ATE’s DSPs can speed time to market. |
The analog baseband instruments needed to test UWB transceivers require very high bandwidths. To source OFDM I/Q symbols at this rate requires at least 300 MHz of analog bandwidth. If the baseband instruments are based on DSPs (digital signal processors), the sampling rates for the I/Q sources should be greater than 1.0 Gsample/s. Similarly, baseband capture instruments should have greater than 300-MHz analog bandwidth and greater than 1-Gsample/s sampling rates.
The digital instruments needed to test I/Q transceivers with integrated DACs and ADCs also require high sampling rates. A transceiver’s embedded converters can have 8-bit parallel differential I/Os; so, a device of this type would require a minimum of sixty-four 1.0-GHz digital channels for the baseband digital I/O per DUT.
For integrated UWB devices, testing is generally limited to EVM, channel power, spectral mask, and either PER or BER. Since the testing of an integrated device does not require baseband instruments, a production ATE (automated test equipment) system used for integrated UWB devices has a simpler configuration than a test system used for I/Q-based transceivers.
Testing the EVM of a UWB I/Q modulator requires a test system to source compliant UWB I/Q symbols to ensure the device is tested under the same conditions as those under which it will be used.
The hopping commands for the fast settling PLL are usually sent on a digital bus. Precise synchronization is required between the I/Q signal sources and the digital commands used to set the PLL. This synchronization is required to make sure the frequency hops of the PLL are in lock step with the I/Q symbols being sourced from the baseband instruments.
To capture the modulated RF output of a transceiver, whether SIP or I/Q-based, an RF test instrument would need better than 10 GHz of analog bandwidth. This would ensure the instrument has enough bandwidth to measure all of the band groups (3.1 to 10.6 GHz). For example, a TFC 1 signal from band-group 6 would be centered at 8.184 GHz and have a channel bandwidth of about 1.7 GHz.
Using multisite test
You can dramatically reduce test time and the cost of test by testing UWB devices in a multisite test configuration that performs very fast background digital signal processing with multiple processors working in parallel on multiple sites. The tester should perform DC biasing, device setup, and RF measurements in parallel in order to achieve the fastest test time and the lowest cost of test.
To achieve the highest parallel efficiency, you should use a POP (pattern-orientated program) if possible. With a POP, the digital pattern can set up the instrumentation, send the digital commands to program the device, and initiate the RF capture. The pattern will ensure these commands are transmitted to each site at the same time. A high-speed bus and parallel DSPs will ensure the most cost-effective test time.
You could also reduce the overall test time for UWB devices by background processing the RF and baseband measurements. For example, the tester could send a captured RF signal to the DSPs, which could start computing the EVM. Then, while the tester is calculating the EVM, it could simultaneously set up the device for another test.
Developing device interface hardware
To implement multisite testing for UWB devices, you’ll need to design quad-site or octal-site DIBs for UWB devices, and this is not a trivial task. The baseband analog I/Q-type UWB transceivers require precisely matched transmission lines. The test system’s baseband instruments must source OFDM signals that are essentially multitones going from 4.125 MHz to 256 MHz. Differential pairs for the I and Q channels must be tightly matched. These DIB traces should be routed close to each other and should be of equal length and as short as possible. The baseband traces should be on the top layer of the DIB where the DUT is located.
The RF RX and TX DIB traces are also critical. You could design a single DIB to test UWB devices that work in different band groups. For example, you may have three different products: a band-group 1 device centered at 3.96 GHz, a band-group 3 device centered at 7.128 GHz, and a band-group 6 device centered at 8.184 GHz. If the device packages are identical and the pin-outs are the same, then you could design one DIB to test the three DUTs; just be sure that the microstrip line traces for the RX and TX can work across the full bandwidth of the device family. Keeping these traces short and picking the proper dielectric are key to designing the required wideband microstrip lines. To maintain proper signal isolation, make sure the top layer of the DIB has a flooded groundplane.
Software challenges for UWB testing
Another challenge you will face in devising a test for UWB devices is creating compliant UWB baseband and RF modulated signals as well as demodulation algorithms for measuring EVM. This is a daunting task that may require you to spend many weeks studying the WiMedia UWB standard. At the end of this time-consuming process, your math may not be 100% compliant, which will lead to miscorrelation to the bench equipment used by the designers and may delay the completion of the ATE test program.
You can shorten the time-to-market cycle if you use the compliant software from bench instrumentation on the ATE’s DSPs. Using the same software will provide direct algorithmic correlation from the bench to the ATE and will let you focus on other aspects of production test. The bench software also needs the ability to operate in the DSP engines in the background of the test program execution. This feature can essentially give the ATE system bench-type qualities that make debugging and correlation much easier than before.
When employing a test procedure that combines such software with high-bandwidth instruments that have fast DSPs, you'll find it easier to keep up with emerging communication standards such as WiMedia UWB while also keeping your test costs low. Multichannel instruments will allow you to simultaneously test multiple transceivers in parallel, and fast DSPs that can work in the background will help improve throughput. Because transferring bench software to an ATE system will shorten the development time for production test, you'll get your products into production more quickly and be able to meet the market needs.
No related content found.
- 0 rated items found.
Datasheets.com Electronic Parts & Inventory Search
185 million searchable parts
- Part Number
- Description
- Inventory
- Products
- Manufacturers
View All Webcasts
