Explanation of De-Embedding

Question: What is de-embedding and how does it work?

Answer: Most precision network analysis tools use some form of de-embedding to remove instrumentation effects and move the reference plane of a measurement closer to the device under test. These tools incorporate a standard calibration procedure that has well-controlled physical properties. These physical properties include known impedances such as a short, open, load (reference impedance, usually 50 ohm) as well as a thru or transmission standard. This is referred to a Short – Open – Load – Thru or SOLT calibration.

An extension of de-embedding for mult-port Z- or S-Parameter measurements is to use an interconnect model (TDR or S-Parameter) to remove packaging effects, PCB trace loss or even test fixture effects when performing high speed measurements such as jitter or eye height. This simplified approach allows the designer to probe virtual test points underneath packages or 6 inches from an edge connector. Most oscilloscope software tools can import S-Parameter files and then use its frequency response data to compute an “inverse” filter to de-convolve or de-embed the interconnect effects over the frequency range of interest. An example is shown below of how one can account for marginal performance on a PCI Express transmitter.

Explore posts in the same categories: Ask Scope Guru Q/A, High Speed Buses


You can comment below, or link to this permanent URL from your own site.

2 Comments on “Explanation of De-Embedding”

  1. Paul Hale Says:

    De-embedding is commonly used in the microwave world where the output is typically viewed in the frequency domain and/or the correction is small. Corrections for cables and fixtures can be quite large and can amplify noise as well as the desired (attenuated signal) giving really poor time domain results. What itype of low-pass filtering is typically done in the inverse filters in commercial scopes to limit the noise amplification? How is the bandwidth of this filter chosen? Can you point me to any literature on this topic?

    • Randy White Says:

      Paul–thanks for your visit. Here are my thoughts on your question.

      De-embedding can be considered as a first order approach to removing test loading effects. As you mentioned, when an inverse response amplification is applied all signals are boosted, both noise and real signals, within the passband. As a rule of thumb I generally don’t recommend de-embedding effects with attenuation factors beyond 14 dB, although even this is questionable. In order to minimize the noise amplification and optimize for SNR a low pass filter needs to be applied. Some tools use a fixed loss profile or threshold to apply the -3dB filter response. For example, if a 5 Gb/s high speed serial data signal is being measured across a cable with 6 dB loss at 5 GHz, then after applying the filter the de-embedded response would have have about 3 dB of gain (-6 dB original => +6 dB from inversion => -3dB filter + +6dB inversion).

      Another approach for where to select the cutoff frequency is to use the 5th harmonic of the fundamental frequency (5 Gb/s => 2.5 GHz fundamental and 12.5 GHz 5th harmonic).

      Another factor that is critical is the assumption of the channel model used for de-embedding. It’s important to make sure the model (TDR data or Touchstone file) is well-behaved. In other words, does the data contain significant frequency nulls which, after applying an inverse function, will exceed the SNR of the scope? Is the frequency spacing and time window adequate for transient events (reflections, etc.) to settle out? Is DC included in the data set? A good place to start for reference material is an application note on Tektonix.com called “Equalization and Serial Data Link Analysis Methods (SDLA) with 80SJNB Advanced Application Note“. There were also various papers presented at DesignCon (www.designcon.com), however most of these use de-embedding in the context of TDR- or VNA-based measurements.

      Good luck!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: