Signal Integrity Analysis Part 3: De-Embedding

Introduction

No test equipment is perfect, including a vector network analyzer (VNA). However, using a powerful toolbox of special operations enables the measurement errors to be dramatically minimized. These operations improve the laboratory measurement data turning it into an excellent representation of the device under test (DUT). While calibration is a pre-measurement step that minimizes errors, the most important operation to reveal information about just the DUT is a post-measurement process called de-embedding. Traditionally, de-embedding has been used only by experienced users. By understanding the principles and how to use the new generation of built-in de-embedding features of VNAs, this powerful technique can be leveraged by all users, and the quality of information extracted from all measurements is dramatically improved. This practical guide to de-embedding will enable all users to take advantage of this important feature.

Table of Content

• Why De-Embedding?
• Principles of De-Embedding
• Obtaining the S-Parameter of the Fixture
• Direct Measurement of Fixture S-Parameters
• Building the Fixture S-Parameter by Fitting a Model to Measurement Data
• Simulating the Fixture S-Parameter with a 3D Field Solver

Why De-Embedding?

Although a VNA is a powerful tool for component characterization, it can only measure between well-calibrated reference planes. Often there is some type of fixture that makes the physical connection between the reference plane of the VNA to the ends of the device under test (DUT).

Figure 1 shows an example from Altera Corporation. The DUT is a ball grid array (BGA) component on the backside of the board, which is accessed from the edge where there are SMA launches and traces on the circuit board. The SMA and traces on the board contribute a larger measured impact than the DUT itself.

How do you isolate the DUT performance when all you have are the DUT and the fixtures?

This is the value of de-embedding techniques. When you have a composite measurement of a DUT/fixture combination, you can isolate the performance of the fixture and use de-embedding to extract or de-embed the fixture from the measurements.

Between the sources and receivers at the core of the VNA are directional couplers, switches, and connectors, all designed to make the measurement of the S-parameters of the DUT effortless and transparent to the user. Each of these internal components contributes to measurement artifacts that hide and obscure the intrinsic measurement of the DUT.

For example, an ideal short on a Smith chart should be a dot on the left-hand side of the chart. The actual measured Smith chart of a short connected directly to the front connector of the VNA is shown on the left of Figure 2. This measurement is basically on the left side of the Smith chart, but other than that is nothing like an ideal short. Clearly, the internal interconnects of the VNA are hiding the true nature of the device under test.

It is virtually impossible to connect a DUT directly to the front panel of test instrumentation. To further complicate measurements, cables and other mounting fixtures are almost always used to interface the DUT to the test instrumentation. Even the most precise interconnects will dramatically distort the measured response of the DUT.

The Smith chart on the right of Figure 2 shows the measured S response from a short located at the end of a meter long, precision 50 ohm, low-loss cable. Deciphering any information about the DUT is virtually impossible from this measured response.

Over the years, many different approaches have been developed for removing the effects of the internal VNA features and the test fixtures from measurement to reveal the behavior of just the DUT. They fall into two fundamental categories: pre-measurement and post-measurement operations.

Pre-measurement operations require specialized calibration standards that are inserted at the ends of the test fixture and measured. This process moves the calibration plane to the end of the fixture. All the effects of the fixture are calibrated up to this plane. The accuracy of the subsequent device measurement relies on the quality of these physical standards. This is why most VNA calibration kits include very precise air dielectric coaxial standards with calibration coefficients of inductance and capacitance that are read into the VNA firmware. Using any 50-ohm loads out of lab stock is not recommended.

Post-measurement operations involve taking a measurement of the DUT and all the fixturing leading up to it, then mathematically removing the fixturing leaving only the DUT behavior. Of course, the essential ingredient to a post-measurement operation is to have accurate information about the fixture. This process is called de-embedding. The intrinsic DUT behavior is embedded in the total measurement and de-embedding removes the fixture effects leaving just the DUT behavior.

This powerful technique can be used when the DUT is remote from the calibration plane or when there are non-coaxial connections from the VNA cables to the DUT. De-embedding is commonly used with circuit board traces, backplane channels, semiconductor packages, connectors, and discrete components.