Mitigating the Effects of The DUT Interface board and Test System Parasitics in Gigabit-Plus Measurements

Much of the focus of the application series in ITC has been on the cost issues associated with the device interface-especially in terms of yield and quality. Even as the industry moves more and more to " guaranteed by characterization " approaches, at some point a device must be placed on an interface board and a test system. Thus, the need to mitigate the effects of the measurement path is critical. Very high speed devices are best described in terms of their data eye. For a transmitter, transition time, amplitude and jitter are the most important data eye parameters. The question for transmitter test is, "How much does the measurement path degrade the device under test?" Likewise for a receiver, sensitivity to jitter, amplitude attenuation, and reduced transition time is critical. The question for the receiver is, "How much does the stimulus path change the data input?" Figure #1 shows the data eye concerns. While compensating for transition time degradation and attenuation is well understood, compensating for jitter is not. A discontinuity due to an interconnect, relay, bond wire, etc. in the measurement path causes deterministic jitter 1. The electrical distance (in time) between two discontinuities influences the amplitude of the data eye when the discontinuities are physically close together. However, when they are separated in time by more than 1/2 of the data period, the reflections/ re-reflections from the discontinuities add or subtract from subsequent incident edges, thus causing jitter. In practical test systems this is almost always the case 2. The discontinuities include: i. the device output/ input impedance itself, ii. the DUT interface board-i.e. vias and launchings, iii. the test system path-i.e. relays, bond wires, protection diodes, and terminations. If a test system is predictable and repeatable, then one often assumes that correction for the measurement system is easy and possible. This, however, is not the case because the jitter Abstract ______________________________ This paper discusses the issues associated with removing the effects of the measurement path in very high speed measurements. Of critical concern is deterministic jitter caused by the interaction between the measurement path and the device under test. While individual components causing jitter in the measurement path can be well characterized and simulated, simple methods of compensating for measurement path error cannot be applied. This paper explores this issue and suggests two complementary solutions for addressing such jitter.