With BER-SNR (bit-error rate/signal-to-noise ratio) correlation, it is possible to test ACR (adjacent-channel rejection) in VHF receivers. Moving component testing for any device from the bench to production ATE (automatic test equipment) poses many challenges, but moving the testing of high-speed RF devices to a production setting can be downright daunting. To facilitate the testing of a VHF RF receiver in a production environment, you can employ a technique that correlates ACR in terms of BER to SNR. The technique, which was developed for ATE, greatly reduces test time and memory requirements while ensuring highly reliable test results.
During the development of an RF receiver (RX), our team found that the performance was close to the limits of the customer’s requirements. To make things more challenging, the performance demands on the receiver were continually increasing. We needed to accurately select devices close to the limits, or we would be rejecting otherwise good devices. This required an accurate and reliable test procedure that was also cost-effective for a production environment. We undertook an exhaustive exercise to try to achieve good correlation from the test-bench setup to the intended ATE. The major problem was that we used a baseband device on the bench to downconvert the received digital RF signal, and a device was determined to be good or bad based on a result of the BER. To compound this problem, the specification that was close to allowable tolerance was ACR, which required two RF generators for the test-bench setup. Due to this complicated test approach, we could not replicate the exact bench setup in our ATE environment, so we needed a new test technique for high-volume production testing.
Testing ACR The ACR performance of a VHF receiver is based on a number of mechanisms, including image rejection, phase noise, and intermodulation distortion performance. All of these interact and cannot be treated separately.
The target application for this VHF receiver uses COFDM (coded orthogonal frequency domain multiplexing) signals that contain multiple carriers (many components at different frequencies), and they all interact. Unfortunately, it is not possible to accurately test the effect of the interaction of these subcarriers using single or dual tones. So, we developed a technique that uses representative signals in the unwanted adjacent channels and measures the noise generated in the wanted channel as a result. We begin our ACR test by tuning the receiver to a low, wanted signal level and then having another signal present in the adjacent channel (upper or lower).
The performance limit is reached when the power in the adjacent channel degrades the signal in the wanted channel by the amount that causes the BER specification to fail. Hence, the ACR is the difference between the wanted power and the adjacent power when the BER fails. By modulating one frequency generator on the ATE with the same modulation scheme used in the bench test setup, we were able to make a true comparison of the system performance.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.