Lakeshore 330 User Manual page 101

Lake Shore Model 330 Autotuning Temperature Controller User's Manual
There are two simple techniques which can be used to test whether these errors might be present in a measuring
system. The first is to connect a capacitor (about 10 µF) in parallel with the diode to act as a shunt for any ac noise
currents. The capacitor must have low leakage current so as not to alter the dc current through the diode. The capacitor
may also alter the time response of the measurement system, so allow sufficient time for the capacitor to charge and for
the system to equilibrate. If the dc voltage reading across the diode increases with the addition of the capacitor, there are
probably ac noise currents present. The second method simply involves measuring the ac voltage signal across the
diode. Although an oscilloscope is often the logical choice for looking at ac signals, many do not have the sensitivity
required and they often introduce unwanted grounds into the system and compound the problem. Most testing can be
performed with the same digital voltmeter used to measure the dc voltage by simply selecting the ac voltage function.
There should be no ac voltage across the diode. If there is, the data presented in the following sections can be used to
estimate the potential error in the temperature measurement.
II. EXPERIMENTAL
In order to quantify the effects of induced currents on silicon diode
temperature sensors, the circuit of Fig. 3 was used to superimpose an
ac current on the dc operating current. The dc current source was
battery powered with currents selectable from < 1 µA to > 1 mA.. The
signal generator could be varied in both amplitude and frequency. All
voltage measurements were made with a Hewlett-Packard 3456A
voltmeter in either the dc voltage mode or the ac (rms) voltage mode.
The dc measurements were taken with an integration time of 10
power line cycles without using the filtering options available on the
voltmeter. The average of several readings was taken to reduce the
measurement uncertainty. An oscilloscope was also used to double-
check and monitor signal frequency, shape, and distortion, but the
FIGURE 3. Measurement circuit schematic diagram.
oscilloscope was removed from the circuit when actual data were
recorded.
Data were recorded at the three dc current values of 1, 10, and 100 µA with the temperature stabilized at 305, 77, or 4.2
K. At each temperature and dc current value, the dc voltage and the ac voltage across the diode were recorded as the
amplitude and frequency of the signal generator were varied. The dc voltage reading across the 10 k Ω standard resistor
was also monitored to verify that the dc component of the current remained constant to within 0.05%. In addition, the IV
characteristic of the diode was measured at each temperature from 0.1 to 150 µA.
Although detailed measurements were taken on only one diode, other diodes were randomly selected and spot checked
at all three temperatures and frequencies to verify consistency with the measured data. The diodes tested were of the
DT-500 series of Lake Shore Cryotronics, Inc. and have been in production long enough to have a substantial reliability
and calibration history.
III. RESULTS AND DISCUSSION
The data were analyzed by calculating a voltage offset ∆ V. This offset is defined as the difference between the dc voltage
reading across the diode when operated with an ac + dc current and the dc voltage reading when operated with a pure dc
current (see Fig. 2). At first glance, the logical choice seems to be to examine the variation of this offset as a function of
the ac current amplitude. However, the ac (rms) voltage across the diode was chosen instead for two reasons, the first of
which is purely practical. In many circumstances, the ac voltage measurement can be made without any modifications to
existing measurement systems, so laboratory checks can be quickly taken and compared directly to the data presented
here to give an estimate of potential temperature errors. Second, in the calculations using the model presented below,
one unknown parameter could be eliminated from the calculations by using the voltage across the diode instead of the
current.
Figures 4 and 5 give the offset voltage as a function of the ac (rms) voltage across the diode for dc currents of 1, 10, and
100 µA with the ac current modulation at 60 Hz. The equivalent temperature error corresponding to the dc offset voltage
is indicated along the right edge of the figure. Figures 6 and 7 give similar plots but at a fixed 10 µA dc current with the ac
current modulation at 60, 1000, and 20,000 Hz. The magnitude of the dc offset voltages is consistent with what has been
observed in measurement systems when corrective action has been taken to eliminate noise problems. Special note
should be taken of the dc current independence in Fig. 4 and the frequency independence in Figs. 6 and 7. The data
taken at 305 K have not been shown as the results are qualitatively very similar to the 77 K measurements and can be
adequately described by the mathematical model which is presented below.
One surprising aspect of the data acquisition was how well the signal processing in the voltmeter could hide even high ac
levels in the dc measurement modes. For example, operating at 10 µA dc and 77 K with a rms noise level of 6 mV gives
a dc voltage offset of about 1.5 mV, which is about a 0.6 K temperature error. When reading the voltage signal using the
filtering and integrating capabilities of the HP 3456A, the dc voltage reading is stable to better than ±0.02 mV (8 mK).
This stability gives a deceptive view of exactly how accurate the temperature measurement really is and emphasizes the
importance of checking all aspects of a measuring system.
Application Notes B-15