Campbell Scientific TGA100 Trace Gas Analyzer Manual - Manual (Page 65)

same group as the sample detector gain, all that is needed is to select the desired gain within the group. Therefore, the
reference detector gain parameter has a range of 0 to 7, and a gain of 1 will provide an actual gain of 278, 698, 1396,
etc., depending on the setting of the sample gain.
The detector offset provides 0 to 40 mV of offset at the input of the detector preamplifier and is used to center the
detector signal in the input range, allowing the detector gain to be maximized. It is normally controlled automatically
using the “OffsetGn” function. This function adjusts the detector offset to center the detector signal (including the zero
and ramp points, but not the high or omitted points) in the input range.
The detector gains and offsets should usually be controlled automatically, with two exceptions. First, it should be
disabled while performing some of the setup steps, such as optical alignment, laser mapping, or setting the zero current.
Second, the automatic gain algorithm will not increase the sample gain beyond gain 7. Therefore, if the detector signals
are extremely weak, it may be necessary to set the detector gains and offsets manually.
4.5.2 Detector Temperature
The reference and sample detectors are cooled thermoelectrically to improve their responsivity. Generally, a lower
detector temperature will increase the detector signal and decrease the concentration noise. However, some lasers emit
enough power to saturate the detectors if they are cooled to their lowest temperature. Early TGA100s were supplied
with an iris to reduce the laser power reaching the detectors. However, reducing the iris opening increases the effect of
the optical interference that causes a concentration offset error. Therefore it is recommended that the iris be left
completely open (on units supplied with an iris), and that the detector temperature be adjusted to give the optimal
detector response.
The user sets the detector temperatures using the dynamic parameter function (see section 3.4.4). The TGA software
automatically controls the power to the detector’s thermoelectric coolers to maintain the desired temperature. The
detector temperatures can be viewed in the upper left corner of the real time screen, as described in section 3.4.1.
To adjust the detector temperatures, observe the detector signal displays in the lower left corner of the real time screen.
These graphs are scaled to match the analog input range. The automatic gain and offset algorithm will normally adjust
the gains and offsets so that the reference detector signal fills 5 to 10% of the input range, and the sample detector
signal fills 50 to 96% of the input range. If the three ‘laser off’ points are at the bottom of the graph or if any of the
ramp points (between the vertical dotted lines) are at the top of the graph, the signal is too large. The signal may be
reduced by raising the detector’s temperature (which decreases its responsivity). Conversely, if the signals are too
small, decreasing the detector temperature will increase the signal level. The ideal detector temperature settings will
give a reference signal of approximately 10% of the input range in gain 0 and a sample signal of approximately 80% of
the input range in gain 0.
Unfortunately, in addition to improving detector responsivity, decreasing the detector's temperature also increases
detector nonlinearity. To first order, detector nonlinearity can be compensated using the detector linearity coefficients,
described in section 4.5.3. However, If detector nonlinearity is significant, or if concentration accuracy is more
important than precision, it is recommended to increase the detector temperatures (especially the sample detector) to
reduce the signal level by 20 to 50%. This may increase the concentration noise, but it will improve accuracy by
reducing the detector nonlinearity. This is especially important when using dual ramp mode to measure isotope ratios.
4.5.3 Detector Linearity Coefficients
An ideal detector would have linear response, such that any increase in the incident optical power would increase its
signal proportionally. Unfortunately, real detectors have nonlinear response. As the incident optical power is increased,
the incremental response becomes gradually lower. Detector nonlinearity is worse at lower detector temperatures and at
higher flux density (large detector signals).
The TGA software corrects detector nonlinearity using the quadratic polynomial:
2 Cr r r
l
+ =
where: r is detector response, r
l
is linearity-corrected response, and C is the linearity correction coefficient. The linearity
correction coefficients are defined separately for the reference and sample detector, and for ramp A and ramp B (if
using dual ramp mode).
The reference detector linearity coefficient should be set to a value of 0, based on the assumption that the reference
detector is perfectly linear. This is assumed because it is difficult to quantify the nonlinearity in the reference detector,
and because it generally gives good results. Although the reference detector may not be perfectly linear, it is much more
linear than the sample detector, because 1) the flux density on the reference detector is low due to the beamsplitter
65