IET Labs, Inc. RLC Digibridge 1693, User And Service Manual, Page: 79 / 171

55
1693 RLC Digibridge
Operation
3.6.4 Selection of Median Value for
Better Accuracy
The Digibridge can be programmed to make mea-
surements in one or more groups of three and take
for results the median value in each group. If you
also select averaging (of 5 for example), the median
values of (5) groups will be averaged, Examples of
uses for the median-value capability are given below.
If the “median-value” capability is enabled, the
Digibridge makes three measurements , discards the
highest and lowest results, and uses the median result
for further calculations (if any), such as averaging.
This capability is a special function. To enable it, press
the following keys: [1][=][SHIFT][SPECIAL] [8].
To disable “median value”, press: [0][=][SHIFT]
[SPECIAL][8]
An example of a use for the median value capability
is to greatly reduce the likelihood of displaying an er-
roneous “transitory” measurement in CONTINUOUS
measure mode. This erroneous measurement is typi-
cally caused by insertion or switching of the DUT at
some indeterminate time during a measurement cycle.
Typically, this erroneous measurement is preceded
and followed by valid ones. (The next several mea-
surements are correct until the DUT is changed again).
If median value capability is enabled, the Digibridge
displays the median of three measurements, only
one of which is liable to be erroneous. Because the
erroneous one is commonly quite different from the
other two, the median is very likely to be one of the
correct ones. Consequently, you see only one change
in the value displayed, from “before” to “after” the
DUT change.
Another example of a use for the median value
capability is for measuring in the presence of oc-
casional noise that pollutes some measurements
- particularly noise spikes or bursts that can occasion-
ally be coupled from electrical equipment (through
power line to Digibridge circuits or via inductive or
capacitive coupling to the DUT itself). Such noise
pollutes a measurement now and then, among a ma-
jority of measurements that are correct. This noise
is non-random, i.e., not “white” noise, but it may
be repetitive. Obviously you would prefer to have
only the correct results displayed and/or output via
the interfaces to other devices. If the duration of the
noise spike is typically small compared to the length
of a measurement cycle and the noise repetition rate is
small compared to the Digibridge measurement rate,
then it is likely that any polluted measurement will
be one of three measurements in the median taking,
the other two being valid and practically identical.
The median of any three consecutive measurements
is therefore very likely to be correct.
3.6.5 Accuracy Enhancement for
Large or Small Impedances at
Particular Frequencies
Regular Zeroing at Test Frequency. When measur-
ing very large or small values of impedance, the
Digibridge will provide much better accuracy than
the specifications, if the OPEN and SHORT zeroing
procedure has been recently repeated ‘with test fre-
quency set to the actual test condition.
Examples of the accuracy that is typically obtained
with measure rate = SLOW, after using the actual test
frequency when zeroing:
At 30 Hz: R = 100 megohms ±1%
(range-1 extension, a factor of 240 over Rmax)
At 120 Hz: 0 = 0.1 farad ±1%
(range-4 extension, a factor of 480 .over Cmax)
At 10 kHz: 0 = 0.1 pF ±1%
(range-1 extension, a factor of 400 below Cmin)
At 100 kHz: L = 0.1 µH ±1%
(range-4 extension, a factor of 100 below Lmin).
NOTE: Even better accuracy is possible if several
measurements are averaged. See paragraph 3.6.3,
above. Use of the “ratio display” special function is
recommended when you measure very large values
(which otherwise cannot be displayed) or very small
values (for which ratio display can provide greatly
improved resolution). See paragraph 3.3.6.