7 Theory of Operation

The Theory of Operation chapter provides a technical description of

the internal components and analysis phases for the ONH836, ON836,
OH836, NH836, O836, N836, and H836 Oxygen/Nitrogen/Hydrogen
Analyzers.
Illustrations .................................................................................7–2
836 Series Oxygen/Nitrogen/Hydrogen Analyzer ..............................7–3
Calculation of Analysis Results .......................................................7–4
Calculate Area .........................................................................7–5
Adjusted Area .........................................................................7–5
Blank ................................................................................7–5
Sensitivity Factor ................................................................7–6
Adjusted Area ....................................................................7–6
Determine Analyte Mass ...........................................................7–6
Determine Range ................................................................7–6
Apply Calibration ................................................................7–7
Drift Correction ..................................................................7–7
Determine Final Analyte Mass with the Blended Approach ........7–7
Determine Concentration ..........................................................7–7
Calculate Final Concentrations ..............................................7–7
Calculation of the Calibration Equation ............................................7–8
Determine the Calibration Equation ............................................7–8
Calibration Equation ............................................................7–8
Apply Calibration ................................................................7–9
Calculate Final Concentrations ..............................................7–9
Weighting Factors ...............................................................7–9
Calibration Statistics—RMS Error ........................................7–10
Calibration Statistics—Coefficient of Determination ...............7–10
Override Conditions ................................................................7–10
Minimum Number of Replicates ..........................................7–11
Single Cluster ...................................................................7–11
Cornerstone Brand Software Rounding ..........................................7–12
Rounding Style: Fixed Decimal or Variable (n=1) ..................7–12
Rounding Style: Significant Digits .......................................7–12
Rounding Style: Variable (n>1) ..........................................7–13
Flow Diagrams ............................................................................7–15

Theory of Operation 836 Series 7–1

Illustrations

Figure 7-1 Linear Flow Diagram (ONH836) ..........................................7–15

Figure 7-2 Linear Flow Diagram (ON836) ............................................7–16
Figure 7-3 Linear Flow Diagram (OH836) ............................................7–17
Figure 7-4 Linear Flow Diagram (NH836) ............................................7–18
Figure 7-5 Linear Flow Diagram (O836) ..............................................7–19
Figure 7-6 Linear Flow Diagram (N836) ..............................................7–20
Figure 7-7 Linear Flow Diagram (H836) ..............................................7–21
Figure 7-8 Linear Flow Diagram (O836Si) ...........................................7–22
836 Series Oxygen/Nitrogen/Hydrogen Analyzer

Not all components are found in all configurations.

The 836 Series Oxygen/Nitrogen/Hydrogen Analyzer is designed for

wide-range measurement of oxygen, nitrogen, and hydrogen content
of steel, refractory metals, and other inorganic materials. The
patented detection system supports the true simultaneous analysis of
oxygen, nitrogen, and hydrogen during a single analysis, with one
crucible and no carrier gas changeover. The instrument features
custom software designed specifically for touch operation.

A pre-weighed sample is placed in a graphite crucible, which is then

heated in an impulse furnace to release analyte gases. Oxygen
present in the sample reacts with the graphite crucible to form CO and
CO2.
An inert gas carrier, typically helium, sweeps the liberated analyte
gases out of the furnace, through a Mass Flow Controller to a series
of
non-dispersive infrared (NDIR) cells. The gas then flows through a
reagent heater where the CO is oxidized to form CO2, and H2 is
oxidized to form H2O. The gas then continues through another set of
NDIR cells where H2O and CO2 are detected. These analytes are then
scrubbed out of the carrier gas stream. A patented Dynamic Flow
Compensation (DFC) system is used to add carrier gas as a make-up
for the gas lost during the scrubbing process. The final component in
the flow stream is a Thermal Conductivity (TC) detector, which is
used
to detect nitrogen.

The detection system is comprised of both NDIR and TC detectors.

NDIR cells are based on the principle that CO, CO2, and H2O absorb
infrared (IR) energy at unique wavelengths within the IR spectrum.
Incident IR energy at these wavelengths is absorbed as the gases
pass through the IR absorption cells. The complete set of CO and CO2
NDIR cells is required to give the most accurate oxygen results for a
wide range of sample types and concentrations. TC detection is based
on the principle that heated filaments within a bridge circuit are
maintained at a constant voltage in a flowing stream of carrier gas.
Changes to the composition of the gas stream will cause a change to
the resistance of the filaments. Nitrogen from the sample will cause
this type of change, which is recorded as the analytical signal. The
concentration of an unknown sample is determined relative to
calibration standards. To reduce interferences from instrument drift,
NDIR reference measurements of pure carrier gas are made prior to
each analysis, while TC reference measurements are made throughout
each analysis.
Calculation of Analysis Results

An asterisk (*) indicates variables and definitions that are

applicable only to dual-range instruments.

Table of Symbols, Constants, and Variables

Symbol Purpose (Units) Type Entry

Adjusted Area (cts) Calculated for each

Aadj, range Area after corrections have been applied for Calculated analyte for each
blank, moisture, etc. analysis

Blank Area (cts)

Calculated based on
An area representing a baseline analyte value in
Ab,range Calculated
consumables or introduced to the system via samples
trapped atmospheric gases. Default value is 0.

Raw Area (cts)

Calculated for each
The area calculated from the integration period.
Arange Calculated
No blank or drift corrections have been made to sample
this value.

Concentration (%, ppm)

Calculated for Each
C The concentration of analyte in the sample given Calculated
Sample
in units of % or ppm.

Drift Factor Method Calculated as

F Parameter Determined by User
A scalar value used to adjust for instrument drift.

Sensitivity Factor R&D

Service
Fsensitivity A scalar value used to convert area into
Constant
mass units, adjusted to make calibration and by model
slope ~ 1
Analyte Mass (g)
Calculated for Each
G The determined mass of analyte for a particular Calculated
Sample
cell or range.

Blended Analyte Mass (g)*

Calculated for Each
Gblended The resultant analyte mass after blending results Calculated
Sample
from the high and low range results.

Drifted Analyte Mass (g)

The mass of analyte for a particular cell or range Calculated for Each
Gdrift Calculated
Sample
that has been corrected for day-to-day changes in
detector signal.

Calculated Analyte Mass, High Range (g)*

Calculated for Each
Ghigh The analyte mass as calculated using the high Calculated
Sample
range cell

Calculated Analyte Mass, Low Range (g)*

Calculated for Each
Glow The analyte mass as calculated using the low Calculated
Sample
range cell

7–4 836 Series Theory of Operation

 Table of Symbols, Constants, and Variables
Symbol Purpose (Units) Type Entry

Maximum Peak Height (cts)

Determine for Each
Hp The maximum height of an analyte peak for a Measurement
Sample
given sample.

Sample Mass (g)

Sample Entered for each
m A scalar value used to record the Parameter sample
mass of the analyzed sample.

High Range Threshold (cts)*

The peak height (maximum) at which the results Method User entered for
ThH Constant each method
change from the blended region to the high cell
results.

Low Range Threshold (cts)*

The peak height (maximum) at which the Method User entered for
ThL Constant each method
results change from the low cell results to the
blended region results.

Calculate Area
The final output signals, whether baseline corrected, wedge rule
corrected, or uncorrected, are used to determine a raw peak area,
Arange, for each analyte range. This raw area is reported in the
expanded replicate information in the Cornerstone® brand software.
Further corrections, as follows, are applied to the raw total area
before they can be used with a calibration.

Adjusted Area
The adjusted area, Aadj, range, corrects the raw area for
sensitivity factors and blank values.

Blan
k
The blank value, Ab,range, is the average raw area from a set of
samples (empty crucibles, boats, etc.) determined by the user. The
blank value is used to account for analyte gases that are repeatedly
introduced through consumables or atmosphere. The default blank
value is 0
1 n
Ab,range  ∑Arange
n 1
where n is the number of blank samples as selected by the user.
 Sensitivity Factor
The sensitivity factor, Fsensitivity, is a factor that maintains a calibration
slope near 1. It is used to ensure consistent slopes across all
products, so it is easy to determine when cell sensitivity (the slope) is
incorrect. It enables better instrument troubleshooting, though the
use of the sensitivity factor means the instrument sensitivity cannot
be determined from the calibration slope.
A sensitivity factor is determined for each detector in an instrument
and may vary with software version.

Adjusted Area
The adjusted area can be calculated as follows:
Aadj,range  range ∑ ∑ sensitiviyt

Ab,range

Determine Analyte Mass

The mass of each analyte must be determined before the
concentrations can be reported. The calibration is used for
this calculation.

Determine Range
This section is applicable only to dual-range instruments.

If an instrument utilizes both a high and low range for a given analyte,
the correct mass result must be used. The result may come from the
high range, low range, or a blend of both depending on the method
settings.
If a blended result is desired, the value is determined based on the
high and low range thresholds set by the user (ThH and ThL) and the
maximum peak height of the low range result for each analysis (Hp).
If:
Hp > ThH, Use high range
Hp < ThL, Use low range
ThL < Hp < ThL, Use blended results, detailed as follows:
The peak height is reported in the expanded replicate information in
the Cornerstone brand software.

7–6 836 Series Theory of Operation

 Apply Calibration
If a calibration exists for the method in use, the areas calculated for
each range are used to determine the mass of analyte in the sample.
If no calibration exists, a default of y=1x is used. The basic
calibration equation is of the form
Aadj k 0 k1 ∑G

where G (grams) is the analyte mass for a given cell and the k values
are constants determined in the calibration. The adjusted area is
used and the mass result is calculated. For instruments with multiple
ranges, the analyte mass is usually calculated for each range.

Drift Correction
The drift factor, Fdrift, corrects for day-to-day changes in signal due to
changing environmental, matrix, or sample conditions. It is a simple
multiplicative factor applied to the final analyte mass and reduces
the frequency with which new calibrations need to be set. The drift
factor value is determined against a calibration and has a default
value of 1.
Gdrift G ∑Fdrift

Determine Final Analyte Mass with the Blended Approach

This section is applicable only to dual-range instruments.

If the analyte mass is to be calculated from a peak height that falls

between the low and high threshold values, the final value will be a
weighted average of the results from the high and low range. The final
result is
∑ 1
Gblended ∑ ∑
∑ high  ∑   lo Th ∑H .
∑G
w H
p
Th L G p

∑Th H ∑Th L ∑
Thorough calibrations are required at both the low and high ranges to
optimize signal blending.

Determine Concentration
The determined analyte mass, G or Gblended, is used in conjunction
with the sample mass to determine the analyte concentration.

Calculate Final Concentrations

The total determined concentration in weight percent is defined as
∑G ∑
C(%) 100 ∑∑ ∑
∑m ∑
The concentration will also be determined in parts per million (ppm)
6
∑G ∑
C( ppm) 10 ∑∑ ∑
∑m ∑
Calculation of the Calibration Equation

Table of Symbols, Constants, and Variables

Symbol Purpose (Units) Type Entry

Adjusted Area (cts) Calculated for each

Aadj, range Area after corrections have been applied for Calculated analyte for each
blank, moisture, etc. analysis

Concentration (%, ppm)

Calculated for Each
C The concentration of analyte in the sample given Calculated
Sample
in units of % or ppm.

Analyte Mass (g)

Calculated for Each
G The determined mass of analyte for a particular Calculated
Sample
cell or range.

Calibration Coefficient
Calculated for Each
k The best-fit coefficients determined weighted least Calculated
Method
squares

n The number of replicate samples included in the Calculated Calculated for Each
calibration. Calibration

Calibration Weighting Factor

The weighting factors can be used to increase or Sample User entered for
w decrease the relative importance of individual Parameter
replicates included in the calibration set.
The default value is 1.0.

Determine the Calibration Equation

The calibration equation is an empirical model describing how the
response of the analyzer changes with the analyte. A series of known
calibration standards, spanning the range of interest, are measured
so the response of the analyzer can be modeled as a polynomial
equation.
Contact LECO Service to enable the Higher Order Calibration Option.

Calibration Equation
A weighted least squares algorithm is used to determine
the polynomial coefficients, k.
W ∑Aadj k 0 k1 ∑G

By selecting the Regression Type option to Force Through Origin,

the operator can force the calibration equation through the origin:
k0 0

7–8 836 Series Theory of Operation

 If the high order calibration option is enabled, the operator can
select quadratic and cubic regression orders:

W ∑Aadj k0 k1 ∑G k22 k3 ∑G

3

∑G

Apply Calibration
Given the coefficients, K, the determined analyte mass, G, can be
determined for each replicate by solving the following equation:
Aadj k 0 k1 ∑G

For higher order calibrations, Cornerstone brand software finds a valid

solution for the following polynomial equation:

Aadj k0 k1 ∑G k2 2

k3 ∑G3
∑G

Calculate Final Concentrations

The total determined concentration in weight percent is defined as:
∑G ∑
C(%) 100∑∑ ∑
∑m ∑
The concentration will also be determined in parts per million (ppm):
6
∑G ∑
C( ppm) 10 ∑ ∑ ∑
∑m ∑

Weighting Factors
Weighting factors, W, are calculated as:

g ∑wi
Wi  2
0.0001 g i
where

1 n
g  ∑g i
n 1
g is the certified analyte mass for each replicate, i, included in the
calibration and w is a user entered weighting factor. By default, the
user defined weighting factors, w, are set to 1.0. If the higher order
calibration option is enabled, the operator can view and edit the
weighting factors.
 Calibration Statistics—RMS Error
The Root Mean Square Error (RMS) is a measure of the accuracy of
the calibration. The accuracy of the calibration improves as the RMS
Error approaches zero.
The RMS Error is determined as:

n 2

∑ g 
RMS Error (g) ∑ gi,certified
 i,measured
1
n ∑∑

Where ∑ is the degree of freedom as determined by the combination

of Regression Order and Regression Type as follows:

Regression Order Force Full

Through Origin Regression
Linear 1 2
*Quadratic 2 3
*Cubic 3 4
*Available only with the Higher Order Calibration Option

Calibration Statistics—Coefficient of Determination

The Coefficient of Determination, R2, is a measure of how well
the measured results fit the calibration model. The fit improves
as R2 approaches 1. R2 is determined as:
2

– gi,certified
n

2
∑ g i,measure
R 1 ∑ 1
n
d
2

∑ g i,measure –g 
1 d

Override Conditions
When possible, Cornerstone brand software chooses a Regression
Order of “Linear” and a Regression Type of “Full Regression”
whenever a new calibration equation is created.
To prevent invalid calibration equations, the software will
automatically override user settings and defaults to ensure that a valid
calibration equation is determined. The following criteria are used:

7–10 836 Series Theory of Operation

Minimum Number of Replicates
The number of replicate measurements included must exceed the
degrees of freedom of the calibration. Cornerstone will automatically
reduce the regression order until the criteria in the following table are
satisfied.

Regression Order Force Full

Through Origin Regression
Linear 1 3
*Quadratic 3 4
*Cubic 4 5
*Available only with the Higher Order Calibration Option

Single Cluster
No matter how many replicates are included, calibrations that are
recognized as a single cluster of measurements will automatically be
set to a Regression Order of “Linear” and a Regression Type of
“Force Through Origin.”
The calibration data will be interpreted as a single cluster if any of the
following criteria are true:
• The range of analyte present in the calibration data is less than
three times the range of the absolute error (g) of the
calibration.
max analyte(g)cert ∑min analyte(g) cert 3 ∑max Errorg ∑min Errorg 
• The minimum amount of analyte is greater than 80% of the
maximum amount of analyte included in the calibration
data.
max analyte(g) cert  80% ∑max analyte(g) cert

• The slope is negative.
Slope0
Cornerstone Brand Software Rounding

The following section provides examples of the different methods of

rounding within the Cornerstone brand software. Refer to Analyte Units
Tab, page 4–6, and Editing Styles, page 4–12.
Meaningful digits in the following tables are underlined.

Rounding Style: Fixed Decimal or Variable (n=1)

Actual Result
Decimal
12.3456789 1.23456789 0.123456789 0.0123456789
Places*
1 12.3 1.2 0.1 0.0
2 12.35 1.23 0.12 0.01
3** 12.346 1.235 0.123 0.012
4** 12.3457 1.2346 0.1235 0.0123
5 12.34568 1.23457 0.12346 0.01235
6 12.345679 1.234568 0.123457 0.012346
*Decimal Places indicates the number of digits displayed to the right of the decimal.
**The default value is 4 when the units are % and 3 when the units are ppm,
µg/g, or mg/kg.

Rounding Style: Significant Digits

Actual Result
Significant
12.3456789 1.23456789 0.123456789 0.0123456789
Digits*
1 10. 1. 0.1 0.01
2 12. 1.2 0.12 0.012
3** 12.3 1.23 0.123 0.0123
4 12.35 1.235 0.1235 0.01235
5 12.346 1.2346 0.12346 0.012346
6 12.3457 1.23457 0.123457 0.0123457
*Significant Digits indicates the number of meaningful digits, including zero
digits except when leading (0.001) and trailing (1000). Trailing zeros will be
displayed when significant (for example: 0.0100)
**The default value is 3 for all units (%, ppm, µg/g, or mg/kg).
The software displays “0” when the rounding style is set to
Significant Digits and the value observed is between -0.0000000001
and
+0.0000000001.

7–12 836 Series Theory of Operation

 Rounding Style: Variable (n>1)
Actual Results
Significant 12.3456789 ± 0.0123456789 ±
Digits (n>1)* 0.123456789 0.000123456789
1 12.3 ± 0.1 0.0123 ± 0.0001
2 12.35 ± 0.12 0.01235 ± 0.00012
3 12.346 ± 0.123 0.012346 ± 0.000123
4 12.3457 ± 0.1235 0.0123457 ± 0.0001235
5 12.34568 ± 0.12346 0.01234568 ± 0.00012346
6 12.345679 ± 0.123457 0.012345679 ± 0.000123457
*Significant Digits (n>1) indicates the number of significant digits displayed in the
standard deviation when more than one replicate is included in the set. The
number of decimal places in the standard deviation will determine the number of
decimal places displayed for the average value.
Significant Digits (n=1) indicates the number of digits displayed to the right of
the decimal when only one replicate is included in the set.
Variable Rounding is applicable only to set results (analytes)
for which the standard deviation is determined.
This page intentionally left blank.

7–14 836 Series Theory of Operation

Flow Diagrams
MFC
55 psig
R31 SV311 SV105 NC

NC 450
2 LPM cc/min
Dose Dose B
Gas S
B V
High
NO
10
S Bypass
20 psig
NO
V NC CO2
SV312 10
55 psig R32 R
2 COM
L
NC 1
P
2 LPM
Dose A Bypass
R
G CO Legend
T
Dose Incoming Leak Check
Gas Segment
FL
A 55 psig
T
R Analyze Leak Check
20 psig Heat Segment
ed
Reag Furnace Leak Check
ent Segment (Bypassable)
FL 650°
He T C
R
Copp
20 psig er
S Electrode Oxid
V Furnace
10
NC
P
T
P

Coolant System
Bypass H2O

Inco
min
g
Bypass Low CO2
Purif
ier
650°
C PV1 Flow
Copp COM
Scru
S
V
NO
bber
10 Leco
NC

Inco sorb/
min Anhy
g
O dron
MI
Scr e
PT2
ubb P
er NC R3
S SV108
Leco NO

Exhaust
NO
V 5000
sorb/ MeFC1
10 cc/min
Anhy COM 35 cc/min
TC

Figure 7-1
Linear Flow Diagram (ONH836)
Theory of Operation 836 Series 7–15