Chapter 6
Chapter 6
Theory of Operation
Table of Contents
Section Page
THEORY OF MEASUREMENT .................................................................................................... 1
FXI OVERVIEW ............................................................................................................................ 2
APPARATUS ................................................................................................................................ 2
AUTOMATION ............................................................................................................................... 2
THE ANALYSIS METHOD ........................................................................................................... 4
DATA ACQUISITION ................................................................................................................... 8
CHROMATOGRAPHIC FEATURES ................................................................................................... 8
Time Slice Features ............................................................................................................. 14
STANDARD “PEAK” PROCESSING ................................................................................................ 17
Peak Detection Methods ...................................................................................................... 17
Forced Gate Peak Detection ............................................................................................. 17
Slope Peak Detection........................................................................................................ 18
Start Slope Detection/End Forced Gate, Peak Detection ................................................. 19
Start Forced Gate/End Slope Detection, Peak Detection ................................................. 24
Baseline Correction Methods ............................................................................................... 24
Event Baseline Correction ................................................................................................. 24
Trailing Peak Cutoff........................................................................................................... 26
Leading Peak Cutoff.......................................................................................................... 28
Auto-Zero Reference......................................................................................................... 29
Tangent Skim .................................................................................................................... 30
Peak Configuration ............................................................................................................... 30
Ghost Peaks...................................................................................................................... 31
Baseline Configuration ...................................................................................................... 31
SAVE RESULTS ........................................................................................................................ 34
STATUS INDICATORS .............................................................................................................. 34
STATUS CONDITIONS ................................................................................................................. 34
SYSTEM ALARM CONFIGURATION ............................................................................................... 35
SYSTEM WARNING CONFIGURATION........................................................................................... 35
VALVE AND OTHER RELAY FAIL-SAFE FUNCTION......................................................................... 36
THE ANALYSIS CYCLE ............................................................................................................ 36
INITIALIZATION PHASE, CYCLE SET-UP FUNCTION ....................................................................... 37
DATA ACQUISITION PHASE ......................................................................................................... 38
Cycle Timing Function .......................................................................................................... 38
Rubber Clocks ...................................................................................................................... 38
POST ACQUISITION PROCESSING PHASE .................................................................................... 39
Method Program Groups ...................................................................................................... 39
VIRTUAL DEVICES .................................................................................................................... 40
VIRTUAL DEVICE RESET............................................................................................................. 41
INSTRUMENT OUTPUTS .......................................................................................................... 41
VALUE SET ................................................................................................................................ 41
CYCLE PROGRAM GROUPS ........................................................................................................ 42
REPORTS .................................................................................................................................. 43
End of Cycle Report Configuration ....................................................................................... 43
APPLIED INSTRUMENT TECHNOLOGIES PROPRIETARY INFORMATION. USE OR DISCLOSURE OF INFORMATION ON THIS
PAGE IS SUBJECT TO THE NOTICE OR RESTRICTIONS ON THIS DOCUMENT’S FIRST OR TITLE PAGE.
6-ii
Chapter 6. Theory of Operation
List of Figures
Figure Page
Figure 6-1. Feature Extraction Table ......................................................................................... 10
Figure 6-2. Boxcar and Slope Events ........................................................................................ 13
Figure 6-3. Slice Mode ............................................................................................................... 15
Figure 6-4. Chromatograph Diagnostic Report Example ........................................................... 16
Figure 6-5. Forced Gate Peak Detection ................................................................................... 18
Figure 6-6. Slope Peak Detection .............................................................................................. 18
Figure 6-7. Illustrating Peak Drift ............................................................................................... 19
Figure 6-8. Start Slope Detection/End Forced Gate, Peak Detection ........................................ 20
Figure 6-9. Start Slope Detection/End Forced Gate, Peak Detection Applied to Two Peaks Not
Baseline Resolved ...................................................................................................................... 21
Figure 6-10. Peak Area Distortion Due to False Peak End ........................................................ 22
Figure 6-11. Start Slope Detection/End Forced Gate, Peak Detection Applied to Avoid False
Peak End .................................................................................................................................... 22
Figure 6-12. Start Slope Detection/End Forced Gate, Peak Detection Applied as Alternative to
Tangent Skim Method ................................................................................................................. 23
Figure 6-13. Unwanted Peak “Captured” Due to Early Peak Gate On Time.............................. 24
Figure 6-14. Start Forced Gate/End Slope Detection, Peak Detection ...................................... 24
Figure 6-15. Baseline-Resolved Peaks ...................................................................................... 25
Figure 6-16. Peaks Not Baseline-Resolved ............................................................................... 25
Figure 6-17. Corrected Area (Baseline-Resolved Peaks) .......................................................... 26
Figure 6-18. Trailing Peak Cutoff Baseline Correction ............................................................... 27
Figure 6-19. Leading Peak Cutoff Baseline Correction .............................................................. 28
Figure 6-20. Auto-Zero Reference Method of Correction .......................................................... 29
Figure 6-21. Tangent Skim ........................................................................................................ 30
Figure 6-22. Peak Identification and Integration ........................................................................ 32
Figure 6-23. Baseline Adjustment .............................................................................................. 33
Figure 6-24. Measurement Overview ......................................................................................... 36
Figure 6-25. Cycle Timing .......................................................................................................... 37
Figure 6-26. Multiple Calibration Streams Example ................................................................... 46
List of Tables
Table Page
Theory of Measurement
The FXi Series5 gas chromatograph (GC) is a microprocessor-controlled instrument
that is capable to sample gaseous and light boiling point liquid mixtures and measure
their component concentrations. The typical applications for the FXi GC include
monitoring and/or controlling in processes of refineries, petrochemical and chemical
plants as well as in air safety requirements of working environment. In general, each
application requires a unique arrangement of GC controlled valves, chromatographic
columns, isothermal oven, signal detectors, and proper peak detection methods.
Together, all the factors determine component concentration magnitude and accuracy.
To achieve valid and effective chromatographic analyses, chemical, electrical, and
mechanic knowledge and effort are required.
The basic principle of chromatography is to separate a gaseous mixture into its
chemical components and measure the quantity of each component. This process is
started by injecting a sample into a chromatographic column. The sample is then
transported through the column by an inert carrier gas. As the sample mixture moves
through the column, its various components travel at different rates. At the end of the
column, each component has been completely separated from the rest of the mixture.
As these individual components exit from the column, they are measured by a detector
that is sensitive to the concentration in the flowing carrier gas. A recording of the
detector’s output is called a “chromatograph.”
As a component passes through the detector, its concentration starts at zero. The
detector signal likewise starts at a base level, builds to a maximum, and falls back to the
base level. This behavior is called a “peak.” The term peak is so prevalent in
chromatography, that it is almost synonymous with component. An individual
component’s total quantity is proportional to the area inside the peak. Imagine a
chromatogram peak recorded on a chart recorder. The bottom limit of the peak is taken
as an imaginary line from the detector signal before the peak to the detector signal after
the peak. This is called the “baseline.” The individual component quantity can be
rendered into a concentration by knowing the initial sample size.
The FXi gas chromatograph automatically performs the sampling, column separation,
detection, and integration. This sequence is called a “measurement cycle.” To perform
these cycles, the analyzer requires application specific configuration parameters that
establish when to switch the valves, when each component’s peak occurs, and so on.
As the cycle proceeds, the detector signal is analyzed to detect features such as peak
starts, peak maximums, peak ends, and so forth. These features guide the integration of
the peak areas that are used to calculate component concentrations.
A chromatograph is calibrated by performing a calibration cycle that analyzes a sample
with known component concentrations. During a calibration cycle, the analyzer
determines the response of the chromatograph (mainly the detector) to each
component. A ratio between each component’s known concentration and its measured
peak area is calculated as a calibration factor. During a measurement cycle, this
calibration factor is applied to the component’s integrated peak area to calculate a
concentration.
As a chromatograph ages, its columns, flow regulators, and so forth, may change and
cause slow variations in the time when peaks occur. These timing changes are
corrected in the FXi by stretching the time-line. This clock correction is referred to by the
terms “rubber clock” and “relative retention.”
FXi Overview
Apparatus
The chromatographic analysis is performed in an isothermal oven. The oven
temperature is monitored to detect alarm conditions.
Several standard detector types are supported: a Thermal Conductivity Detector (TCD),
a Flame Ionization Detector (FID), a Flame Photometric Detector (FPD), and a Helium
Ionization Detector (HID). Flameout detection (via analog input monitoring) and auto-
light sequencing are provided for both the FID and FPD. The TCD filaments are
protected by a carrier gas supply pressure switch.
Automation
The FXi Gas Chromatograph (GC) performs analyses on a cyclical basis with each
cycle producing multiple component measurements. In addition, analysis cycles can be
performed for different sample sources. Thus, one FXi may produce a large number of
measurements.
The following factors influence the way in which the FXi is configured. A single GC can
analyze significantly different gases that require chromatograph control and detection
parameters, calibration factors, etc., to change for each sample being analyzed.
Analysis cycles may differ in their real-time chromatograph-control and peak/
component-detection strategies; parameters relating to these aspects are grouped
within a “method.”
Normal ranges of component values may vary for samples from different sources. Even
when the method for chromatograph control and peak extraction remains the same,
different calibration factors may be desired for different reporting units. The GC virtual
instrument groups all calibration factors for each analysis within a “calibration set.” For
consistency in reporting, limit thresholds and normal “full scale” range parameters (for
trending) are separately grouped within a “value set.”
Peak/component parameters appear in three places: methods, calibration sets, and
value sets. “Linking” of parameters for a specific peak is done using the name of the
component.
Specification of each analysis cycle requires, as a minimum, identification of the
following:
1. the physical source, or “stream”
2. the “method”
3. the “calibration set”
4. the “value set”
APPLIED INSTRUMENT TECHNOLOGIES PROPRIETARY INFORMATION. USE OR DISCLOSURE OF INFORMATION ON THIS
PAGE IS SUBJECT TO THE NOTICE OR RESTRICTIONS ON THIS DOCUMENT’S FIRST OR TITLE PAGE.
6-2
Chapter 6. Theory of Operation
time can be either a hard and fast clock time, or a stretched time based on the rubber
clock. The GC performs each event action within 25 ms after its configured time. The
possible event types and their parameters are listed in Table 6-1, and are used in the
CLI (Command Line Interpreter) programming used in the GC, accessed through
PC-HyperTerminal.
Table 6-1. Acquisition Commands
Command Function
Autozero Changes the state of the hardware autozero function for a given detector. The
event is configured by detector and new state. While the autozero hardware is
on, it determines an offset that will bring the detector signal slightly above zero
at the analog to digital (A/D) converter. While the autozero hardware is off, the
most recently determined offset is continuously applied to the detector signal.
This brings the detector signal into the most sensitive region of the A/D
converter. The autozero hardware should be turned on for a short period
during the chromatogram baseline.
baseline_check Checks that the sensitivity parameters (above) are set properly.
Baseline_check specifies a number of boxcar transitions to be compared
against the noiseband. If any one of these transitions exceeds the noiseband,
then a baseline noise condition is set. The baseline_check event should be
used after each slope or boxcar event, and located on a clean section of
baseline.
Basepoint Brackets a section of baseline for an accurate level determination. The
basepoint event is configured with a basepoint identifier, and an on or off state.
The specified basepoint level is determined by averaging the detector signal
between the basepoint_on and the basepoint_off events. The baseline
subtracted from each peak may be selected between these basepoints, or the
detector level at peak start or stop. Refer to the data acquisition function for
more details.
Boxcar Sets the number of 40 Hz detector samples that are averaged for slope
detection. Refer to the section on Chromatographic features. The specified
value is a power of 2; i.e. power = 4 results in 16 samples per boxcar average.
Each boxcar event should be followed by a slope event.
chrom_output Changes the reconstructed chromatogram output going to a strip chart
recorder. The chrom event is configured for a detector with a new analog
output channel, new signal offset and attenuation as well as a feature tick
option.
chrom_storage Changes the data capture into the new result chromatogram. It is configured
for a detector with the number of detector samples to be averaged into each
chromatogram point. It specifies ON/OFF of recording to chromatogram buffer.
Recording can be turned on and off many times during a cycle.
Gate Stores information used to start or stop peak integration at specified times. Its
configuration includes detector, and gate identifier.
probe Brackets a section of the chromatogram used to determine the detector noise
level. The probe event is configured for a particular detector and state (on or
off). It determines the standard deviation of the 40 Hz detector signal between
the probe_on and probe_off event.
Command Function
purge_sync Holds cycle clock until a specified time has elapsed since the last actual
change of a stream select valve. The minimum elapsed time is specified with
the purge_sync event configuration. If more than the minimum time had
already elapsed, then the cycle clock is not held. The purpose of the
purge_sync is to insure that gas from a particular stream valve has been
flowing for a sufficiently long time through the sample loop.
scan Start or stop slope/peak data collection from a chromatographic detector. Data
collection may be either slope peak detection, or time slices, but not both at
once. The scan events specify a detector and data collection state (on or off).
Refer to the data acquisition function for a description of slope scan mode.
slice The slice event starts or stops the slice mode of data collection on the
specified detector.
slope Changes the slope detection sensitivity parameters for a detector. These
include noiseband and deadband. Boxcar power must be set separately.
start_time Records the current date and time as the cycle start time. The start_time is
required when using the internal FXi peak simulator. Simulator times are all
relative to the start_time event.
stream Opens the specified sample source stream. Use of this event is discouraged
because it locks a method into a specific stream.
stream_advance Causes all stream valves to close except the one for the next analysis cycle.
tangent Modifies the normal slope peak detection to detect the small peak riding on the
tail of a large peak. This event specifies the detector, tangent skim state, and
scale. Start with a scale of 8. If a math error occurs, set the scale down. If the
results are not repeatable, raise the scale. Refer to the data acquisition
section.
tick Causes a mark on the chart recorder. Specify up or down.
time_sync Holds the cycle clock until a specified time has elapsed since the previous
time_sync event. The configuration includes the minimum elapsed time. If
more than the minimum time had already elapsed, then a cycle error is
generated, and the cycle clock is not held. The purpose of the time_sync event
is to allow analysis cycles to be strictly periodic in time.
valve Used to drive an oven valve open or closed. The valve number and new state
are configured in the valve event.
For instance, the first event of the first method can be configured in the CLI as:
DEV[1]. REAL.gc.work.setup{
Method = 1
method[1] {
event = 1
event [1] {
time = 1000 /* ms */
clock = HARD # not rubber clock as per
page 6-38 of this manual
type = valve
type.valve.id = 1
type.valve.state = ON
}
There is normally more than one event, and may be more than one method; so the
assignment of the list sizes used above are only for example. If the WOI were used, one
would have to select into each container one element at a time until the parameter
values are reached.
Proper operation of the measurement control system requires detailed knowledge of an
applied gas chromatograph configuration. Some simple guidelines are given here.
Activation of the sample injection (oven) valve should always be immediately preceded
by”purge_sync” event. The injection should be followed by a “start_time” event
scheduled or the same moment.
Proper feature and peak detection is very dependent upon the slope sensitivity
parameters; see “Data Acquisition Function, Slope Sensitivity.”
A “scan on” event should be placed just before the earliest of all the sensitivity events.
At least one to four seconds of autozero should precede the first peak, but not overlap
any other peak. The “scan on” may have to be moved back to accommodate this. A
“scan off” event should be placed about one peak width after the last peak of interest:
scan off time = last ET + ( 3.5 * last FWHM )
where FWHM is the full peak width at half the maximum peak height, and ET is the
elution time for the peak.
If a peak of interest is well resolved to baseline on either side, a peak gate should be
placed about one peak width away on the baseline:
gate times = ET + / - ( 2 * FWHM )
If this gate time is within 2 FWHM of any other peak, then split the difference, but try to
keep gates at least 1.5 FWHM from any peak.
Peak identification windows should generally be half way to the nearest peak, or 3
FWHM, whichever is smaller. The peak drift window should be about half of the peak
identification window.
The best integration method to use is gate to gate. If either gate is unavailable due to an
adjacent peak, then use slope determination for that limit. In the very special case
where a small peak is on the tail of a much larger peak, then a tangent integration limit
should be used to stop the small peak integration.
The execution of a method’s events produces a chromatogram buffer that contains the
peaks to be measured. Once this chromatogram has been obtained, the selection and
analysis of the peaks is controlled by the “peaks” portion of the method. The peaks
portion of each method is divided into a list of container elements names “detector.”
Each peak element contains a set of parameters that identify the name of the chemical
component the peak represents, a time window for the peak, the baseline determination
to use, and so forth.
Data Acquisition
The data acquisition function takes in the detector signal in real time, filters it, and
produces the chromatographic features, baseline points, and noise determinations. In
addition, a chromatogram can be accumulated for operator inspection. The inputs are
the detector signal and data events. Outputs are the chromatographic features, baseline
points, noise intensities, and saved chromatogram. Refer to Figure 6-1.
Associated with each detector is a “chromatogram buffer” that can hold data points and
a feature list. These can be uploaded for display. The chromatogram records averages
of every n points; n may be changed by the method at any point in the data acquisition
process.
During slope-triggered integration, detector output is “piecewise” integrated. Upon the
occurrence of a slope-triggered event or time-gate event, an entry is made in a Features
Extraction Table. The entry includes the type of event, the (hard) time of the event, the
area since the last event, and the height (detector output) at the event. Following entry
of the area into the table, the respective integrator is reset. The table is processed later
to obtain peak measurements; see Figure 6-1, Feature Extraction Table.
Chromatographic Features
The FXi collects chromatograms and feature lists as virtual instrument data elements.
The raw chromatogram element consists of periodic, non-overlapping time averages of
the detector signal. The raw chromatogram’s time averages do NOT correspond to the
boxcars.
Each of the chromatographic features is stamped with a cycle time, the detector signal
at that time, and the integral of the detector signal since the last feature. The types of
chromatographic features are listed below:
Scan_on and scan_off features mark the beginning and end of detector integration
periods.
Autozero_on and autozero_off features mark the beginning and end of hardware
autozero operation. The detector signal is not valid while the autozero is in operation. If
autozero commanded during “defined peak integration time,” and
AUTOZERO_IN_PEAK error results.
Gate features are used to mark the timed beginning or ending of peak integration. Refer
to the data acquisition function for more details.
Slope_on, peak_max, and slope_off features are detected using the slope of the
detector signal as filtered by the boxcars.
A tangent feature marks the point on a downslope where the detector signal’s tangent
passes through the previous signal minimum. This is useful when a small peak appears
on the tail of a much larger peak.
A peak_max feature marks the transition from upslope to downslope. A height feature
records the maximum height seen during upslope and downslope and records it upon
detection of the end of downslope.
Timeslice features are produced during timeslice data collection on a periodic basis.
Refer to the data acquisition function.
Basepoints consist of the average time and signal level of a basepoint episode. Refer
to the basepoint event in the cycle timing function.
Noise probe points consist of the average time and standard deviation of the detector
signal during a noise probe episode. Refer to the noise probe event in the cycle timing
function.
peak_max
height
slope_on slope_off
gate 1 gate 2
time
scan_off
slope_1
gate 1
gate 2
deadband ≥4 ≤8
Any baseline_check events should test a count of at least the deadband number of
boxcars.
The point at which a slope event is “detected” is generally not the effective time of the
event. The slope-triggered processing maintains information about the previous two
boxcars, which is used to find the conditions as of the “actual” event.
After slope_on, the detector signal is monitored for its highest point; when the
downslope is detected, this height feature is added to the feature table.
During downslope episodes, a tangent feature may be detected if a line tangent to the
detector signal passes through the last upslope point. This is detected by calculating a
line from the “valley” at the start of the rider peak to the current boxcar. When the line’s
slope reaches a minimum, the tangent feature is inserted. Refer to Figure 6-2.
deadband
FWHM
slope on
t1 t2 t3
t = 15 seconds
EXAMPLE:
FWHM = 15 seconds PROGRAM:
From Table 6-2: EVENT [x] boxcar
Boxcar_power = 6 time = t1
Noise_reduction = 8 boxcar power = 6
GET “stddev” EVENT [y] slope_on
stddev = DEV[1].REAL.gc.work.cycle[1].detector[x].probe.stddev time = t2
if stddev = 2.68 noiseband = 1.34
then noiseband = 4 * stddev / nr = 4 * 2.68 / 8 = 1.34 deadband = 6
slice
slice
slice
slice
slice
slice
slice
slice
slice
slice
Feature Table
scan on
scan on
scan off
slice
slice
slice
slice
slice
slice
slice
slice
slice
slice
slice
slice
START STOP
tm
second. The controller counts the time while it looks for, and saves, the maximum
height and time of maximum height of the peak. It then continues on the downslope
portion of the peak, and following the deadband period (see Deadband/Noiseband in
the section) the peak terminates and the controller saves the height and time. The
controller then first determines if the peak is a peak of interest by determining if peak
summit time was entered in the method. If so, then the peak is a peak of interest. If not,
then the controller will pass the peak information to the chromatogram diagnostic report
but otherwise go no further with the data.
If the peak is determined to be a peak of interest, then the FXi will correct the total area
by the baseline correction method specified, apply the calibration factor to compute a
concentration, use the concentration to determine if high or low alarms exist (and report
as necessary), update the trend, and save the concentration for the optional
printer/terminal and host computer.
The FXi will also determine if the peak of interest has drifted. It does this by determining
if the time of maximum height lies within the configured time window. If this is not the
case, then the controller will conclude that the peak has drifted; and indicate that
condition. (See Figure 6-7 and Figure 6-22.)
A time that occurs on the positive slope of the peak of interest and a time desired for the
peak to end have been entered into the GC from the printer/terminal.
Refer to Figure 6-8. The total area for the peak includes all of the area between peak
start and the peak gate off time. If the standard forced gate were used without moving
the gate on and off times, the area for the peak would be only the area between the
peak gate on time and the peak gate off time.
PEAK GATE OFF
FORCED GATE
TIME
FORCED GATE
TIME
Figure 6-9. Start Slope Detection/End Forced Gate, Peak Detection Applied to Two
Peaks Not Baseline Resolved
Another use for the start slope detection/end forced gate – peak area method is to avoid
a valve spike causing a false peak end point, without resorting to forced gated entirely.
Refer to Figure 6-10. In this example, the peak is set for baseline-to-baseline correction
and standard slope detection. A valve is switched just after the peak is finished giving a
negative spike. However, because the spike is so close to the peak, the dead band may
not time out before the spike occurs. The controller would then continue to integrate
until the bottom of the spike, which would result in an incorrect end of peak and an
incorrect baseline correction. Refer to Figure 6-11. In this case, the user has used the
start slope detection/end forced gate – peak area method and reset the peak off time to
the same time as the valve switch. Here the peak ends properly.
VALVE SW ITCH
PEAK END
TIME
PEAK END
TIME
baseline” correction method with the start slope detection/end forced gate – peak area
method. Refer to Figure 6-12.
CORRECTED
AREA
PEAK GATE
OFF
PEAK START
PEAK END
CAPTURED PEAK OF
PEAK INTEREST
TIME
Figure 6-13. Unwanted Peak “Captured” Due to Early Peak Gate On Time
Start Forced Gate/End Slope Detection, Peak Detection
This method of peak detection operates in the opposite manner of the start slop
detection/ end forced gate method. At the peak gate on time, the controller terminates
any existing peak and starts a new one, continuing to integrate regardless of slope
changes until anything occurs that will terminate a peak in slope detection mode. Recall
that an upslope or a down slope that flattens to a baseline after the dead band period
will terminate a peak in slope detection mode. Refer to Figure 6-14.
PEAK GATE ON
TIME
100
1 OR
LESS
100 GREATER
THAN 1
CORRECTED
AREA
START STOP
ZERO VOLTS
Corrected Area = (Total Area) – (Base Width) (Base Height Start + Base Height Stop)
(2)
As long as the base height is above zero volts, the controller always corrects for the
difference between the baseline and zero volts whether the peak is slope detected or
force gated. (If the base height is at zero volts or less, a baseline drift alarm occurs for
peaks of interest.)
Trailing Peak Cutoff
The trailing peak cutoff method (Figure 6-18) is used for peaks which are baseline-
resolved at start, but not at stop. The controller “drops a line” at peak stop to the height
of the peak at peak start. This provides a better approximation to the true area of peaks
that are not baseline-resolved at the peak stop.
STOP
CORRECTED
AREA
START
ZERO VOLTS
START
CORRECTED
AREA
STOP
ZERO VOLTS
Auto-Zero Reference
The auto-zero reference method (Figure 6-20) is used for peaks that are not baseline-
resolved at either peak start or peak stop.
CORRECTED
AREA
START STOP
AUTOZERO
OFF
Tangent Skim
The tangent skim method (refer to Figure 6-21) is used for peaks that lie on the
downslope of a much larger peak. If the leading peak cutoff method were used, the area
allocated to the peak would be too large.
CORRECTED
AREA
START
STOP
Peak Configuration
Measurements are configured within each method as a list of peak specifications for
each detector, for example:
gc.work.setup.method[X].detector[Y].peak[]
A measurement may be based upon a peak’s area or height.
A peak is identified by the time of its maximum. A measurement is valid only if its
configured time window contains exactly one maximum feature. Peak integration is
conceptually enabled/disabled by “trigger” features. There are three types of triggers:
1) (Forced) Gates,
2) Slope, and
3) Tangential (Skim).
Trigger type is specified separately for the beginning and end of integration. Integration
may be triggered on by a gate and off by a slope, etc. Integration is performed using the
APPLIED INSTRUMENT TECHNOLOGIES PROPRIETARY INFORMATION. USE OR DISCLOSURE OF INFORMATION ON THIS
PAGE IS SUBJECT TO THE NOTICE OR RESTRICTIONS ON THIS DOCUMENT’S FIRST OR TITLE PAGE.
6-30
Chapter 6. Theory of Operation
features list to sum back in time from the summit to a trigger-on event, and forward in
time to a trigger-off event. Refer to Figure 6-22. Each peak specification has complete
use of the features extraction table. This means that the same physical peak can be
“integrated” and “calibrated” without a summit if at least one of the trigger events is a
forced gate. If the expected summit time is set to zero, normal peak identification and
drift monitoring is bypassed. The “peak” is located by the specified gate(s).
Ghost Peaks
“Ghost” peaks can be configured to collect slope triggered peak areas that are
otherwise unprocessed. The peak identification window is used to specify a time
interval. All summits lying within the window and not otherwise processed are analyzed
as a group:
1) each peak feature within the window is located, if previously processed it is
ignored, otherwise
2) its area is determined using slope triggering with baseline adjustment as
configured;
3) as additional peaks are located and processed, their areas are added to the
result value.
4) The sum total of all these areas will be corrected using a single calibration
factor.
To keep ghost peaks free of slope and boxcar events, configure separate ghost peaks
for each interval between slope and boxcar events.
Baseline Configuration
All” raw” measurements (area or height) are baseline adjusted. Baseline correction of
raw area is performed by subtracting the area of a trapezoid, which is defined by actual
“detector zero,” the trigger start and stop times, and the effective baseline heights
occurring at the trigger start and stop times. Baseline correction of a height
measurement is performed by subtracting the height of the trapezoid at the peak max
time. If the area (height) is less than the baseline, it is limited to zero, and a peak area
alarm is generated.
m ax
height
gate 1 gate 2
tim e
peak
peak identification
find first peak
within id window -id +id
expected tim e
-drift +drift
from sum m it
backward to start trigger Gate 1
forward to stop trigger Gate 2
DEFAULT
PEAK_STOP
PEAK_STOP
BASE_POINT 2
BASE_PROJECT 2 BASE_PROJECT 2
BASE_POINT 2
TO PEAK START
PEAK_START
Baseline Correction Point
TO PEAK STOP
Baseline Correction Point
BASE_POINT 1 BASE_POINT 1
BASE_PROJECT 1 BASE_PROJECT 1
Save Results
The save_results function may be used to copy the new results into the save buffer for
future reference. Only one save buffer is provided. When accessing the virtual
instrument, the new results are found in:
gc.work.cycle[1],
The old results are in:
gc.work.cycle[2]
and the same buffer is:
gc.work.cycle[3].
Status Indicators
The FXi has several outputs that indicate its status:
Watchdog timer relay
System alarm relay
System warning relay
Common data alarm relay
Specific alarm relays (optional)
The watchdog timer relay indicates that the FXi controller PWA has failed. When this
relay is active, the FXi will not be able to perform any other functions, including the
activation for the system alarm relay.
The system alarm relay provides voltage free contact closure whenever the FXi has a
fault.
The system warning relay provides voltage free contact closure whenever the FXi
needs maintenance.
The common data alarm relay is a solid state relay that provides line voltage
whenever the FXi has measured an out-of-tolerance concentration from any stream.
The specific alarm relays are relays that activate for specifically configured status
conditions. These are optional, and can be ordered as either voltage free contacts, or
solid state relays that provide line voltage.
Status Conditions
The FXi status conditions are broken into several categories:
System status conditions
Cycle result conditions
Cycle system status conditions
Cycle status conditions
Detector specific status conditions
APPLIED INSTRUMENT TECHNOLOGIES PROPRIETARY INFORMATION. USE OR DISCLOSURE OF INFORMATION ON THIS
PAGE IS SUBJECT TO THE NOTICE OR RESTRICTIONS ON THIS DOCUMENT’S FIRST OR TITLE PAGE.
6-34
Chapter 6. Theory of Operation
Cycle Related
CHROMATOGRAPH & DETECTOR I/A Series Outputs
Program
Data Acquisition Tables Groups
Baseline
Points Table
Results
Standard Peak Lim its
Feature
detector valve Processing Processing
Extraction
signals controls
Table
Data Acquisition
Adjusted
and sorted
Com m and
Rubber Clock
(event) List
Factors
METHOD
Post Acquisition
Post Acquisition
Post Acquisition
Initialization
Initialization
Initialization
Processing
Processing
Processing
Start Tim e
Start Tim e
Start Tim e
Figure 6-25. Cycle Timing
Timing within the first and last phases depends upon the particular setup. This time is
relatively small unless extensive use is made of program groups during post acquisition
processing.
Rubber Clocks
Timing within the data acquisition phase is rigidly controlled to a resolution equal to the
basic data acquisition rate of 1/40 second. Multiple events can be programmed to occur
at the same time. Events can use of two kinds of clocks: either a “hard” clock or various
“rubber” clocks.
Rubber clocks are adjusted to run faster or slower to account for long term drifts in
retention time. Determination of the adjustment factor, the so-called “relative retention”
calculation, is generally based on a programmed reference peak and possibly another
event such as a valve change. Adjustment of a rubber clock can be used in the
following cycle to specify timed events such as gates and valve changes. Rubber clocks
are kept on a detector-by-detector basis in:
gc.work.setup.method[x].detector[]
These are part of the setup and should occasionally be backed up in EEPROM.
Rubber clocks should not be used unless some large, well resolved peak is present in
any sample analyzed by a method. If there is any chance that the reference peak will
not be detectable, then rubber clocks should be avoided.
Virtual Devices
In order to ease custom programming of the GC, certain internal resources are made
available as virtual devices. These resources include:
• Cycle component results,
• Analog inputs,
• Analog outputs,
• Data alarm relays, and
• I/A Series inputs and outputs.
Each virtual device has a single data value and a unique name. The CLI program
accesses virtual devices by using the functions VDIN, and VDOUT. These CLI functions
identify specific virtual devices by a unique string that specifies the type of internal
resource, and a unique name. For instance, you could move a calculated result from
variable Q into the cycle component result by using a function line that reads:
VDOUT(“CYC_VALUE, [name]”,Q)
As with the calibration factors, the component name of the virtual device must exactly
match the component name configured in the method’s pseudo-peak. It is strongly
recommended that each measurement result be given a standard component name,
and that this name be used consistently throughout all of the analyzers’ configurations.
A complete list of virtual devices is given in “Virtual Devices” in Appendix B of this
manual, which shows all the internal resource types, and their valid names.
Instrument Outputs
After all, of a cycle’s status and component results are assembled, they must be
reported and relayed to the instrument outputs. This is accomplished using an output
“value set” as defined below, the virtual devices described above, and a “custom
configuration.”
Value Set
Each GC analysis cycle results in a component concentration measurement for every
peak listed in its method. Different methods may have different peak lists, resulting in
different sets of measurements. However, the FXi allows cycles using different methods
to produce similar reports and outputs. This is accomplished through an independent
specification of the exact order in which the component concentrations are to be
reported. This reporting order is called a value set. In addition to the reporting order, the
value set specifies other output options such as the alarm limits to apply to the
component, and its analog output trend scaling. After the component concentration
measurements are made, it is the value set that determines what status indicators to
report, where each component is to appear in the report, how it is to be scaled, and
what alarm limits to apply. Please note that none of the measurement results are
available for reporting until after they have been selected by a value set. Value sets are
configured in:
gc.work.setup.value_set[].
Refer to “Virtual Devices” in Appendix B of this manual.
Each cycle must be configured to use a value set via:
gc.work.setup.sequence.number.[x].cycle[]
which, among other things, contains an element named value_set.
Reports
End of Cycle Report Configuration
End-of-cycle reporting is configured by selecting the output devices for the report and
the report trigger via:
gc.work.setup.reports
Note that any combination of output options, including none, may be selected.
• Report to printer
• Report to Modbus
The report trigger consists of one of the following options:
• End-of-cycle
- Every cycle, or
- On 1st occurrent of a TRUE froman input virtual device
- The 1st cycle after a recurring time interval
Time interval in days and hours, and
Start of time interval in hours and minutes (shift start time)
The report format to the AMS and the printer is selected as part of a cycle’s setup in:
gc.work.setup.sequence.number.[x].cycle[Y].report_type
Only one report format is specified for each cycle. The format options are listed below.
• None, no report generated
- Trend, contains the following:
- Cycle start time
- Cycle system status: okay or not okay
- Cycle status: okay or not okay
- Component information for components in the cycle’s value set:
Component status: okay or not okay
Component value
• Brief, contains the following:
- GC name
- Cycle start time
- Stream analyzed
- Cycle system status: okay or not okay
- Cycle status: okay or not okay
- Cycle serial number
- Cycle type (measurement, calibration, etc.)
- Detector status: okay or not okay
- Component information for components in the cycle’s value set:
Component name
Component units
Component status: okay or not okay
Component type (area, height, pseudo, etc.)
APPLIED INSTRUMENT TECHNOLOGIES PROPRIETARY INFORMATION. USE OR DISCLOSURE OF INFORMATION ON THIS
PAGE IS SUBJECT TO THE NOTICE OR RESTRICTIONS ON THIS DOCUMENT’S FIRST OR TITLE PAGE.
6-43
Chapter 6. Theory of Operation
Calibration Cycles
Periodically, analysis cycles should be used to re-calibrate the FXi. Re-calibration
allows accurate corrections for slight drifts in the instrument’s sensitivity. Calibration
may also detect some instrument failures.
Calibration cycles differ from measurement cycles only in the final calculation steps.
In a measurement cycle, after the peak value is obtained, it is corrected for the detector
response to that component. Calibration factors are stored in lists. The calibration factor
within the selected set is identified by name in the method’s peak configuration. The
resulting component concentration is placed in the cycle results. Normalization may
then be applied to adjust each of the component concentrations so that the total
concentration amounts to 100%. When using normalization, all the area in a
chromatogram should be accounted for; see “Ghost Peaks” on page 6-31 of this
chapter.
In a calibration cycle, after the peak value is obtained, it is combined with the known
component concentration to determine the calibration factor. The calibration cycle
results reflect the concentrations measured using the old calibration factors. The new
calibration factors are placed in temporary storage. If auto-acceptance is specified by:
gc.work.setup.sequence.number[X].cycle[Y].type= AUTOCAL
and this new calibration factor is within a specified tolerance, from:
gc.work.setup.calib_set[X].factor[Y].upper_limit
and lower_limit,
then the new value is also saved for use on future measurements. Otherwise, the new
calibration factor must be manually accepted before it is used by measurement cycles.
Each calibration factor set may be designated as a detector response (area /
concentration), or a multiplication factor (concentration / area). There are no pre-
specified relationships between calibration factor sets and streams or methods; these
links are explicitly configured as part of the cycle configuration in:
gc.work.setup.sequence.number[X].cycle[Y]
Refer to Figure 6-26, Multiple Calibration Streams Example.
Method 1 Measurem ent Cycle n+2 Method 1 Measurem ent Cycle n Method 1 Measurem ent Cycle n+1
Stream 1 Stream 2 Stream 3
cal_set(1).cal_factor()
1 2 3 4 5 6 7
METHOD 3
Mathpack
Processing
Com ponent()
"results vector"
1 2 3
Cycle Sequencing
The instrument normally executes cycles under automatic control. Measurement cycles
are organized into “sequences,” which may be run, halted, interrupted by another
sequence, continued, etc. Each cycle, or “step,” in the sequence is specified with the
following parameters:
1) method number,
2) indivisible yes/no. (Perform this cycle consecutively with the next one in this
series.)
3) stream number,
4) cycle type:
a) normal measurement, or
b) calibration, with calibration factor acceptance either:
i) “manual,” or
ii) “automatic”
5) calibration set index,
6) value set index,
7) list of program groups to execute, and
8) report configuration.
A measurement sequence is executed step-by-step until the end of the sequence is
reached. If continuous execution is needed, a “loop-back” step may be explicitly
configured. The loop can be configured so that any number of initial cycles in the
sequence is not re-executed.
For instance, a sequence might consist of three cycles, but repeat only the last two.
Refer to the Tutorial section on Sequencing.
A measurement sequence may be interrupted to perform another sequence in support
of “manual” or timed calibration requests. The CLI program can also make sequence
requests.
As an example, the CLI program could be used to detect an abnormal condition such as
a process upset and switch to a special sequence designed to monitor the upset. Other
uses include calibration, or chromatograph-health verification.
Automatic cycle sequencing leads to inter-cycle timing concerns. For instance, the
proper purge stream is automatically selected at the beginning of each cycle. Within the
cycle, sufficient purge time is assured by the purge_sync event. Additional purge time
is generated by the stream_advance event.
Inner-cycle timing is fixed by the time_sync event. Refer to “The Analysis Method” on
page 6-4 of this chapter.
Some complicated measurements may require several streams to be sampled, and
hence several cycles to be completed before useful results are obtained. Cycles related
in this way must be performed together without the insertion of cycles from other
sequences. This can be accomplished by setting the cycle’s no_interrupt flag to ON.