The effects of the test fixture on the measured data must be minimized in order to accurately determine the performance of
the device under test. If the line impedance is anything other than 50 and/or there is a discontinuity at the microstrip to
SMA interface, there will be errors in the data for the device under test. The test environment can never be “perfect”, but the
procedure used to build and evaluate the test boards (outlined below) demonstrates an attempt to minimize the errors
associated with testing these devices. The lower the signal level that is being measured, the more impact the fixture errors
will have on the data. Parameters such as Return Loss and Isolation/Directivity, which are specified as low as 27dB and
typically measure at much lower levels, will present the greatest measurement challenge.
The test fixture errors introduce an uncertainty to the measured data. Fixture errors can make the performance of the device
under test look better or worse than it actually is. For example, if a device has a known return loss of 30dB and a
discontinuity with a magnitude of –35dB is introduced into the measurement path, the new measured Return Loss data
could read anywhere between –26dB and –37dB. This same discontinuity could introduce an insertion phase error of up to
1.
There are different techniques used throughout the industry to minimize the affects of the test fixture on the measurement
data. Anaren uses the following design and de-embedding criteria:
Test boards have been designed and parameters specified to provide trace impedances of 50 1.
Furthermore, discontinuities at the SMA to microstrip interface are required to be less than –35dB and
insertion phase errors (due to differences in the connector interface discontinuities and the electrical
line length) should be less than 0.50 from the median value of the four paths.
A “Thru” circuit board is built. This is a two port, microstrip board that uses the same SMA to microstrip
interface and has the same total length (insertion phase) as the actual test board. The “Thru” board
must meet the same stringent requirements as the test board. The insertion loss and insertion phase
of the “Thru” board are measured and stored. This data is used to completely de-embed the device
under test from the test fixture. The de-embedded data is available in S-parameter form on the Anaren
website (www.anaren.com).
Note: The S-parameter files that are available on the anaren.com website include data for frequencies that are outside of
the specified band. It is important to note that the test fixture is designed for optimum performance through 2.3GHz. Some
degradation in the test fixture performance will occur above this frequency and connector interface discontinuities of –25dB
or more can be expected. This larger discontinuity will affect the data at frequencies above 2.3GHz.
Circuit Board Layout
The dimensions for the Anaren test board are shown below. The test board is printed on Rogers RO4350 material that is
0.020” thick. Consider the case when a different material is used. First, the pad size must remain the same to accommodate
the part. But, if the material thickness or dielectric constant (or both) changes, the reactance at the interface to the coupler
will also change. Second, the linewidth required for 50 will be different and this will introduce a step in the line at the pad
where the coupler interfaces with the printed microstrip trace. Both of these conditions will affect the performance of the
part. To achieve the specified performance, serious attention must be given to the design and layout of the circuit
environment in which this component will be used.
If a different circuit board material is used, an attempt should be made to achieve the same interface pad reactance that is
present on the Anaren RO4350 test board. When thinner circuit board material is used, the ground plane will be closer to
the pad yielding more capacitance for the same size interface pad. The same is true if the dielectric constant of the circuit
board material is higher than is used on the Anaren test board. In both of these cases, narrowing the line before the
interface pad will introduce a series inductance, which, when properly tuned, will compensate for the extra capacitive
reactance. If a thicker circuit board or one with a lower dielectric constant is used,
the interface pad will have less capacitive reactance than the Anaren test board. In this case, a wider section of line before
the interface pad (or a larger interface pad) will introduce a shunt capacitance and when properly tuned will match the
performance of the Anaren test board.