Feature Article - Time Domain Reflectometry - By Christopher Henderson

In this section we will discuss time domain reflectometry. Time Domain Reflectometry, or TDR, is a technique finding increased use as a fault localization tool for devices with more complex packaging, such as wafer chip scale packaging, stacked packages, and devices with substrates or interposers. First we’ll discuss the hardware and components associated with Time domain reflectometry. Next, we’ll discuss the waveforms and how one can calculate a distance to an open or resistive change. We’ll then discuss how to interpret the waveforms produced by TDR. Last, we’ll introduce electro-optical TDR and describe how it can improve interpretation and localization of failures.

Time domain reflectometry is a technique that is becoming more common as a failure analysis technique. It is an electrical technique and is non-destructive. In time domain reflectometry, or TDR, an oscilloscope with a pulse shaper generates an electrical pulse. This pulse travels through the cabling to the package of the device under test. At the various material interfaces and corners, there will be some amount of reflection and transmission. A well-matched interface should have very little reflection, while a poorly matched interface would have a large reflection. In theory, the return waveform from an open solder bump, for instance, should show up as a marked change from the waveform through a good solder bump. The key challenge with this technology is resolution, interpretation, and integration with the design files.

Figure 1 shows an example of a time domain reflectometry setup for failure analysis work. The computer on the right drives the Tektronix digital sampling oscilloscope, shown on the lower left. The module with the silver-colored cable connects to the small blue box to generate the picosecond pulse width needed for the generated pulse, while the module connected to the copper-colored cable is the sampling head for the reflected signal. The yellow cable propagates the signal to the device under test.

Figure 2 shows a closer view of the TDR probe. In this configuration, the probe is mounted to a micromanipulator head for precise positioning over a lead, pad, or bump. The device under test is mounted to a glass slide to provide isolation from the stage, leading to a more accurate signal. The two black heads near the microscope objective are simply for lighting the sample more clearly.

Before discussing TDR waveforms let’s briefly review the electrical waveform behavior at interfaces. If you recall from basic electronics or electricity and magnetism, a waveform will react at an interface between two different impedances by generating a transmitted component and a reflected component. The closer the impedances between the two materials, the larger the transmitted component will be. The more different the impedances, the larger the reflected component will be. For instance, at the cable to solder bump interface the match is only fair, so there will be a transmitted component and a reflected component. At the solder bump to pad interface the match might be better, so there is less of a reflected component. If, however, there is an open between the solder bump and the pad trace, the impedance is much different, leading to a mostly reflected component. This shows up as a difference in the waveform like we see on the right in Figure 3.

Let’s discuss the waveforms in more detail. Figure 4 (left) is an example of the waveform from a good device. The red trace represents the waveform from the probe when disconnected from the device. We see a waveform that steps up abruptly at just under 100 psec. This indicates that the pulse is completely reflected and returns to the oscilloscope around 100 psec after launch. The blue and green waveforms show the device under test and a comparison or golden unit. Note that the waveforms are very similar. There is an initial sharp rise that matches the red trace — this is related to the mismatch at the solder bump/probe interface. The slower rise associated with blue and green waveforms comes from multiple minor reflections at various interfaces and corners within the package up to the transistors in the I/O circuitry.

In Figure 4 (right) we show a comparison device, the blue signal, and a failing device, the brown signal. We position the first cursor just beyond the minimum signal point after the initial reflection from the probe/device package interface in the comparison device. We then position the second cursor just beyond the minimum signal point in the blue waveform. There is obviously some guesswork to this placement; it is not an exact science. The difference in the times will be used to calculate the distance to the defect.

Once the engineer captures the signal and establishes a time difference, he or she must relate that time difference back into a distance to establish where the open or change in resistance might be. The distance to the defect in millimeters can be calculated by

where c is the speed of light in meters per second, Δt is the estimated time difference calculated from the waveforms, and ε is the dielectric constant of the material. Different materials have different dielectric constants, so one needs to take this variable into account. We show some example materials and their dielectric constants in Table 1. As an example calculation, a Δt of 50 psec gives a distance to the open of 3.71mm in a fiberglass substrate.

In next month’s article, we’ll discuss how to interpret TDR signals, and a new form of TDR called Electro-Optical TDR.