PXI and FPGA: the ultimate reconfigurable instrumentation
30 January 2013
There is little doubt wireless-enabled devices are becoming more prominent in all areas of life. From the Wi-Fi or cellular capabilities of the tablet you use to check the news to the satnav that guides you around an unfamiliar city.
This dependence on wireless devices is driving the need for test equipment to evolve to meet the challenges of testing a variety of cellular and wireless standards alongside standard electronics test.
An example of this comes from Dublin-based wireless test experts Benetel, who developed a manufacturing test system for a wireless multimedia tablet device. This system needed to incorporate a wide variety of tests, including LED colour and intensity measurements, touch screen calibration, audio loopback, camera, accelerometer, charging current circuitry and USB port functionality. Add to this the variety of proprietary RF tests and Wi-Fi measurements, including error vector magnitude (EVM), bit error rate (BER), TX output power, adjacent channel power (ACP) and Spectral Emission Mask, and it becomes obvious why a versatile platform was required to incorporate all of these measurements in a single system.
On top of this variety, also consider the processing power needed to even take the individual measurements. Spectral measurements such as adjacent channel leakage ratio (ACLR) tend to be the most demanding on instrument performance and speed. Modern spectrum analysers implement ACLR measurements through the fast Fourier transform (FFT), long known as the quickest method of converting time domain data into the frequency domain. But even with efficient implementations, a software-based FFT calculation can dominate measurement time. It gets worse: to improve RF performance and increase repeatability for these demanding frequency domain measurements, test engineers often average multiple measurements. Whilst this improves the accuracy, it only exacerbates the speed problem, since the time taken to repeat the measurements soon stacks up.
To overcome the problem of integrating multiple measurement types in a single system, the most straightforward solution is to utilise modular instrumentation. Benetel put this into practice in the design of the aforementioned tablet manufacturing test system, using PXI Express and modular instrumentation. They made use of NI data acquisition, digital I/O and switch modules, and also incorporated the NI PXIe-5663E Vector Signal Analyser and NI PXIe-5673E Vector Signal Generator for the RF testing. Darragh McShane, Senior Test System Design Engineer at Benetel stated that: "The most significant advantage of PXI for our products lies in the fact that it offers a compact and reliable means of incorporating all of the diverse testing requirements of a typical wireless test system into one neat, rack-mountable package."
What might not be quite so obvious is how a modular approach can solve the second problem, of coping with the demands of strenuous RF measurements. Fortunately, the same technological advances that make our consumer devices more complex are also making the instrumentation more powerful and capable to perform the tests required. Over the last 15 years, modular instruments have evolved to include vector acquisition capabilities, fast-tuning oscillators and high-performance analogue-to-digital converters, not to mention the PXI Express bus itself for high-bandwidth data movement. These features make quicker measurements possible and they also offer the ability to take advantage of the fastest PC microprocessors, including multicore technology, which you can use to implement your measurements in software. This approach, known as virtual instrumentation, has become the standard for the highest performance automated test systems. Such advances in computation ability greatly benefit complex measurements like ACLR.
If the microprocessor initiated the virtual instrumentation revolution, then the field-programmable gate array (FPGA) is ushering in its next phase. FPGAs have been used in instruments for many years. For instance, today’s high-bandwidth oscilloscopes collect so much data, it is impossible for users to quickly analyse all of it. Hardware-defined algorithms on these devices, often implemented on FPGAs, perform data analysis and reduction (averaging, waveform mathematics and triggering), compute statistics (mean, standard deviation, maximum and minimum) and process the data for display, all to present the results to the user in a meaningful way. While these capabilities present obvious value, there is lost potential in the closed nature of these FPGAs. In most cases, users cannot deploy their own custom measurement algorithms to this powerful processing hardware.
Open, user-programmable FPGAs on measurement hardware offer many advantages over processor-only systems. Because of their immense computational capabilities, FPGAs can deliver higher test throughput and greater test coverage, which reduces test time and capital expenditures. The low latency of FPGA measurements also provides the ability to implement tests that are not possible on a microprocessor alone. Their inherent parallelism offers true multisite test, even more so than with multicore processors. And finally, FPGAs can play a key role in real-time test hardware sequencing and DUT control.
So the theory works, but in reality the implementation of open FPGAs within test systems has proved less straightforward. In practice, traditionally it has been necessary to employ an embedded designer well-versed in hardware description languages (HDL) to deploy these algorithms onto an FPGA. The unfortunate truth is that for the most part, the ability to design and architect test systems and program embedded software are not often found in the same person. This proved a stumbling block in many cases, since the risks and costs of out-sourcing the development were restrictive. Nowadays, HDL abstraction tools are alleviating some of these problems and lowering the barrier to FPGA adoption, by allowing developers to design a higher-level representation of the code, often in a schematic or graphical fashion. These tools, like Xilinx System Generator for DSP, Mentor Graphics Visual Elite HDL and the NI LabVIEW FPGA Module, preclude the need for HDL programming skills, bringing the power of FPGAs within the reach of all engineers.
In order to maximise the value of open FPGAs within automated test systems, PXI Express allows direct, point-to-point transfers between multiple instruments without sending data through the host processor or memory. This is known as peer-to-peer (P2P) streaming, and enables devices in a system to share information without burdening other system resources. NI P2P technology is supported on PXI Express NI FlexRIO FPGA modules and PXI Express digitisers and vector signal analysers (VSAs).
One common need in RF applications is a real-time frequency domain trigger. While most RF instruments trigger on a power level, this is typically independent of what frequency the newly detected signal is present at. However, with peer-to-peer data streaming and processing using the NI LabVIEW FPGA Module, you can create a frequency-domain trigger. In the application depicted in Figure 1, the NI PXIe-5663 vector signal analyser uses peer-to-peer streaming to send data to the NI FlexRIO FPGA module, where it is windowed, converted to the frequency domain and then compared against a mask. When the data exceeds this mask, the FPGA module asserts a digital trigger on the PXI backplane. Once the NI PXIe-5663 receives this trigger, it uses its normal acquisition memory to capture a record of data, including pretrigger samples. You can then access this record from the host through the NI-RFSA driver for additional processing or storage.
Another common RF application need is for real-time spectral measurements. A traditional spectrum analyser uses a swept-spectrum approach and analyses only a given band at a given time, the resolution bandwidth (RBW). If a signal isn’t present when the spectrum analyser is sweeping over its frequency, it won’t be detected. With a real-time spectrum analyser, however, you can analyse the entire band of interest at all times by performing continuous, repeated and even overlapped FFTs. Using the same hardware and initial signal processing found in the above frequency domain trigger, you can add measurements on top of the raw FFT data. Simple measurements include a continuous peak hold, which, unlike a swept-spectrum analyser, does not have the potential to miss brief or intermittent signals. More advanced measurements include multi-FFT averaging for noise reduction and measurement confidence, tone tracking for frequency-agile signals and even continuous time spectrograms to display frequency domain information across time. These are examples of real-time measurements you can make only with an FPGA.
When migrating a measurement algorithm to an FPGA, you can gain the maximum benefit for the lowest investment if you focus on the computations most amenable to FPGA acceleration. For ACLR, this includes the FFT and running sum. VSA data arrives on the FPGA from a peer-to-peer first in, first out (FIFO) buffer, which is configured from the host. Next, time domain windowing can be applied to reduce spectral leakage and then the FFT applied. After the FFT, a running sum of the magnitude of each bin is maintained. Once you have obtained the specified number of accumulations, you can use DMA to transfer the resulting data to the host for normalisation (division by the number of sums), power calculation in each band, conversion into dB and the appropriate ACLR calculations.
Whilst this article has focussed mostly on testing wireless capability, FPGA processing advantages extend beyond spectral measurements. To make testing even faster and more flexible, you can implement not only additional RF tests such as time domain averaging and occupied bandwidth, but also protocol-aware and hardware-in-the-loop testing. The implementation of custom measurements in the LabVIEW FPGA Module extends LabVIEW and PXI core benefits such as parallelism and peer-to-peer streaming for faster, more reliable and more flexible test architectures.
Figure 1 Peer-to-peer (P2P) streaming enables devices to share information without burdening other system resources
Figure 2 It is possible to analyse an entire band of interest at all times by performing continuous, repeated and overlapped FFTs
Contact Details and Archive...