Operation
40
© 2017 Pico Technology
pv106ug r1
It is important to be aware of the difference between ‘insertable’ and ‘non-insertable’ devices and the impact
on measurement and calibration technique. The figure above illustrates the difference between an insertable
device and non-insertable devices. The key issue is that with non-insertable devices an adaptor may be
required during the measurement as shown. In order to obtain accurate measurement data, the effect of the
adaptor needs to be removed from the measurements. Some of the possible ways of doing this with the
PicoVNA 106 are as follows:
Calibration used
Adaptor removal method
Comments
(A) 12-term (non-insertable DUT)
Calibration removes adaptor
effect
Accurate but requires kit with through
adaptor standard. Only one calibration kit is
required.
(B) 8-Term (unknown through method)
Calibration automatically
removes adaptor effect
Accurate and easiest method as it does
not require a characterized through
standard
(C) Any except 12-term or 8-term
De-embedding
Accurate but requires prior knowledge of
adaptor characteristics
(D) Any except 12-term or 8-term
Reference plane extension
Quick and easy but errors due to loss of
adaptor remain
Table:
Techniques for dealing with non-insertable DUTs
The preferred method is method (B) in the table above. This requires only a reciprocal through adaptor.
Note
that with this method the
isolation calibration
is done automatically by the software using measurements
taken during the
short
,
open
and
load
steps.
Note that only the 12-term non-insertable DUT calibration and 8-term unknown through methods support a
non-zero length through connection. All other calibrations (S
21
or S
11
+S
21
or full 12-term insertable DUT)
require a through connection of zero length. In effect, this means that the calibration port terminals should
be of opposite sexes. In other words, it should be possible to connect the terminals together without the use
of an adaptor. Consequently, the DUT must be an
insertable
device. If this is not the case, an adaptor will be
needed during the measurement and its effect will need to be removed by either moving the reference plane
or by using the de-embedding facility as indicated in the table above.
When performing just an S
21
calibration it is possible to complete the calibration without doing the isolation
calibration. Simply click
Apply Cal
after performing the
through
calibration. The isolation calibration corrects
errors due to crosstalk (see
Calibration and error correction
) and should be used when measuring insertion
losses larger than about 40 dB. The terminations to use during the isolation calibration can, as a guide, be
50 Ω loads. In some circumstances, such as when testing a highly reactive device (e.g. filter beyond cut-off),
a short or an open circuit may be more appropriate or for best results two actual DUTs with 50 Ω loads at
their unused ports.
6.2.1
Changing the frequency sweep settings without
recalibrating
If the start or stop frequency or number of sweep points is changed when the instrument has a valid
calibration, the user is given the choice to either keep or delete the existing calibration. If the user chooses to
keep the calibration, a new set of calibration error terms will be automatically generated by
interpolation
to
fit in with the new sweep parameters. In this case, a “?” is added to the calibration status bar to indicate that
operating parameters have changed from those used in calibration.
If it is required to change the frequency sweep parameters without recalibrating, then simply enter the new
values (see
Set Sweep Frequency
values under
Calibration
) and click
Apply
. Once the new values are sent to
the instrument, just close the window (click
Close Window
) to exit. Note that this process will
delete all
memory traces
and the display data may be invalid until a fresh measurement sweep is performed.
Some care needs to be exercised when using interpolation, particularly at low frequencies, say below 20
MHz. For example, if starting with a wide band sweep with large frequency steps, say 500 kHz or larger, then
interpolating down to a narrow band of, for example, 1MHz to 10MHz may lead to larger than expected
errors.
