Instrumentation & Measurement Magazine 25-8 - 32

An Improved Thru-Line
De-Embedding Technique up to
80 GHz
Shirong Lv, Fusheng Tang, Mingqi Jiao, Shengli Zhang and Liang Xie
W
ith the rapid increase of high-speed communications,
the frequency range of integrated
circuits enters the field of millimeter wave,
and the size of devices has become smaller and smaller. In the
measurement of microwave devices, the probe cannot contact
the device directly. The actual device under test (DUT) is
surrounded by metal traces called leads, which provide the interface
between the probing pads on the wafer and the DUT
itself. The probes are pressed on the pads at both ends of the
DUT. The parasitic effect of pads and transmission lines is significant
at high frequency. The established technique such
as Short-Open-Load-Thru (SOLT) and Line-Reflect-ReflectMatch
(LRRM) move the reference plane up to the tip of the
probes by means of Cascade Microtech impedance standard
substrate (ISS). To obtain accurate parameters of DUT, a deembedding
method must be implemented to eliminate the
influence of pads and interconnect transmission lines.
Several de-embedding methods based on an equivalent
lumped-element model have been proposed previously.
Open-Short de-embedding is a typical method based on a
lumped circuit model which takes the impedance Z and admittance
Y elements between test pad and interconnect
transmission lines into account, as indicated in Fig. 1. Based on
open-short de-embedding, a series of methods such as openshort-thru
and open-thru were proposed. The open and short
standards should be manufactured with high precision; otherwise,
it may lead to over de-embedding, and the lumped
circuit model is not accurate at frequencies over 20 GHz.
According to [4], the Thru-only de-embedding is an elegant
method in terms of design and mathematical implementation
since it uses a single de-embedding structure and saves
silicon area compared to other methods. The parasitic effects
described above are simplified to a series impedance Z=R+jɷL
and a shunt admittance Y=jɷC, as shown in Fig. 2. The Thruonly
method has some problems, since the length of the Thru
has to be as short as possible to fit the model. It will cause
crosstalk between probes if the length of standard is too short,
such as Y3
in Fig. 1, and the result of de-embedding will be
inaccurate.
In [5]-[9], a distributed de-embedding technique called
L-2L based on transmission line devices of different length
is discussed. As shown in Fig. 3, the PAD contains the parasitic
effects caused by probes, pads and transmission lines
between the pad and DUT. The core idea of L-2L is to equate
the parasitic effect to error box, by regarding the measurement
Fig. 2. Thru network.
Fig. 1. Equivalent circuits of de-embedding patterns for open-short method.
32
Fig. 3. A cascade network of L-2L de-embedding.
IEEE Instrumentation & Measurement Magazine
1094-6969/22/$25.00©2022IEEE
November 2022
Instrumentation & Measurement Magazine 25-8

Table of Contents for the Digital Edition of Instrumentation & Measurement Magazine 25-8
