Calibration Certification of Spectrophotometers

CALIBRATION & CERTIFICATION OF SPECTRO PHOTOMETERS
Author Manish Kapadia, JISL Ahmedabad.(JISL belongs to Jaysynth Group, Mumbai, Pioneers in Textile
colorants, and early Indian manufacturers of spectrophotometers.)
1 Introduction
Like all instruments used for measuring physical quantities, spectrophotometers
need to be calibrated and characterized periodically. The procedure itself
must be certified, also periodically, to validate the use of the instrument. ISO
9000 a series of quality management and quality assurance standards
recommends the documentation of these procedures to maintain a history
trail of a process.

Since many people are unfamiliar with ISO 9000, the original
standards in the series are outlined in a table:

1.1 Terminology
Before proceeding any further, we need to agree on the terminology. Often we
use interchangeably terms like calibration, characterization, certification,
qualification, and validation. However, these terms are all different.

Calibration. To correlate the readings of an instrument with a standard. After
a calibration procedure the output of the instrument may be different (closer
to the standard).

Characterization. To relate the readings of an instrument to a standard. The
characterization procedure does not change the instruments output. The
characterization of a (calibration) standard, e.g., a reference tile, consists in
determining a physical property of the standard based on a physical observation.
If the characterization instrument has been calibrated with respect to a
natural constant or a primary standard, then the characterization is absolute,
otherwise it is relative.

Certification. The creation of a document attesting that performance of an
instrument is within a pre-specified tolerance. The certification requires three
elements: an authority (person or apparatus skilled in the art), a characterization,
and the presence of a tolerance.

Qualification. The verification that an instrument can be used to fulfill a specified
task. The steps required to certify an instrument for a particular task.

Validation. Confirmation that an instrument or procedure performs as certi-
fied. Validation occurs through peer review or through signature of a certifi-
cate after the assertion has been verified.

Traceability. The ability to relate an instrument to a natural constant or a primary
standard by transferring a calibration from one tile to another [9].

Precision. Random error in measurement.

1.2 Manufacturing issues
Most commonly CIE colorimetry is employed to determine whether the color
of a sample is a match to the aim-color of a product for an average observer
under specified viewing conditions. The tolerance of an instrument can then
be just slightly lower than a just noticeable difference (jnd). In a different
application, some drifts in manufacturing processes can be detected more easily
with a colorimetric measurement than measuring a different parameter,
e.g., temperature* or acidity. In these cases jnds are irrelevant and precision is
much more important than accuracy.
This report describes a verification procedure for spectrophotometers used in
process control based on trend analysis. This analysis is more valuable if the
instruments drift is tightly monitored, so it can be unequivocally established
that any detected drift is in the process under observation and not in the
instrument used for the measurement.

Verification implies a two step process:
a calibration to adjust the instruments scale, and the measurement of a sample
of known color to validate the calibration within a predetermined tolerance.
We recommend a much more frequent calibration certification than is usual
for spectrophotometers used in color matching applications. To make this feasible,
the procedure must be very simple and quick to perform; our procedure
fulfills both requirements.
In support of ISO 9001 procedures we publish all data relevant to our instruments.
We make recommendations for a working methodology and recordkeeping
that significantly facilitate the introduction of ISO 9001 compliant
procedures.
1.3 Previous work
Todays spectrophotometers can be easily calibrated making a single quick
measurement of a white reference tile that is traceable to an absolute standard.
This procedure is sufficient for routine work, but periodically the instruments
calibration must be verified by measuring a number of different tiles.
Unfortunately this verification is often ignored.
In todays world of contract manufacturing, where parts are progressively
manipulated by a number of suppliers and then assembled, quality control
can be a nightmare if it is not designed very meticulously. This is the main
goal of the ISO 9000 standards series mentioned earlier. The prescribed procedures
for a spectrophotometer can consist for example in a verification with a
single color tile each three months and a larger set each year. This larger set
usually consists in a number of gray samples to determine linearity and various
vivid samples.
For process control we need a more frequent verification. If possible, we would
like to use the verification data to increase the precision of the measurement
data.
Most research in this direction has been to improve the calibration of instruments.
The techniques (see e.g., [9]) consist in modeling the various errors,
such as white reference, black reference, wavelength errors. A set of reference
samples characterized with an absolute instrument by a standards body is
measured with the instrument under consideration. The obtained data is used
to estimate the parameters for the error model. The analyst can then measure
a sample and apply the model to the measurement data to obtain a characterization
that is traceable to a primary standard.
Such a procedure is quite difficult and requires a very high commitment. Fortunately
it is necessary only when calibration standards are manufactured,
because in calibration a scale is zeroed and accuracy is therefore very impor-
tant.

An example is the standard for a corporate color that is used in all goods
manufactured or distributed by a company. In most applications of spectrophotometry
accuracy is less important than precision, the ability to repeat a
certain measurement.
The focus of the work in this report is to design a validation procedure that is
as simple as a calibration. Such a procedure does not need a high commitment.
5 Procedures
5.1 Overview
Each of the sub-sections in this section is a self-contained step-by-step
description of a qualification procedure. This sub-section is an overview outlining
how the various procedures fit together. It shall also clarify why the procedures
have been defined and what benefits they provide. We generally follow
the recommendations of Christopher Burgess [3] and of Gretag-Macbeth [10].
It was not so many years ago, that the standard instrumentation for color
measurements consisted of a spectroradiometer, a set lamps, and a BaSO4
(barium sulfate) cake. The instruments sensor was cooled with a Peltier element,
which caused condensation, requiring a silicagel cake in the instrument
that had to be baked periodically in an oven. A careful log had to be maintained
on the lamps on time so they could be renewed when a certain usage
time had past. Finally, the barium sulfate cake had to be inspected carefully
and sanded when necessary.
Setting up the instruments was also a challenge, because the geometry had to
be calibrated for each measurement session. The instruments were connected
to a minicomputer running simple software that could compute the colorimetric
quantities and sometimes could produce simple diagrams or histograms.
Todays spectrophotometers are simpler to use than the remote controller for a
television set, are completely self-contained, ruggerized, and have a sophisticated
embedded software system that can compute a plethora of quantities,
evaluate them statistically, and plot them.
Unfortunately this can mislead users in thinking that measuring color is as
easy as flipping a light switch. The fundamental physics on which color measurements
are based have not changed; color has not become easier to measure,
and modern instruments merely provide more comfort and convenience.
The main area of concern in practical colorimetry is the validation of color
measurements. As outlined by Burgess, this consists of the following points:
the fitness for purpose of the spectrophotometer for the task
demonstrable compliance with the manufacturers performance criteria
demonstrable compliance with established standards and practices
documented evidence for continued operability and data integrity
40 Spectrophotometer Calibration and Certification
For HP Internal Use Only
These requirements for validation are met with the qualifications for design,
installation, operation, and performance. These qualification steps are summarized
in the following sections.
5.1.1 Design qualification
The design qualification is a specification for the instrument:
photometric requirements
environmental conditions
sample presentation
data acquisition and presentation needs
operability factors; health & safety issues
cost/benefit analysis
The photometric requirements include such parameters as precision, accuracy,
uniformity with respect to color, aperture, reading frequency, and geometry.
Color metrology is still a field of active study in the basic sciences; the
colorimetric calculations are still being improved considerably and standardization
bodies such as the ASTM, CIE, and ISO/IEC regularly update the recommended
procedures. The instrument should be current with the latest
progress in the technology, such as color matching function interpolation and
color difference evaluation.
The environmental conditions cover parameters such as operational limits for
temperature and humidity, ruggedness, orientation in space, electrical supply
stability, and the generation of heat and noise by the instrument.
The sample presentation refers to the instruments ability to reliably measure
the sample. This is not limited to the physical size of the sample, but also covers
considerations such as the ability to measure liquids, wet samples, cockled
samples, and to select the appropriate backing. If the sample is ink on a substrate,
the substrate can be opaque, translucent, or transparent; it can be
glossy or opaque.
The data acquisition needs usually are a serial RS-232 or USB connection and
a full specification of the protocol. In the past years there has been increased
demand for support of the http data transfer protocol. Depending on the users
skills and time, there are varying demands on a client implementation of the
protocol, which can range from a simple driver to a class library or a specialized
application.
The operability factors include documentation and training. Color is an illusion
and colorimetry is the art to predict an illusion from a physical measurement.
Each analyst using the instrument must be thoroughly trained in the
basics of colorimetry and in the operation of the equipment. Refresher courses
must be given periodically, and a supervisor must be responsible for the analysts
training.
5.1.2 Installation qualification
The second qualification is the compliance with the above specification, i.e.,
does the instrument work the way the manufacturer claims it should? This
reports provides methods and guidance for the following qualification steps:
written test protocol with acceptance criteria
certification from the installation engineer
full written records
adequate documentation and manuals
health and safety information from manufacturer
Regardless of their exterior ruggedness and environmental specifications,
internally spectroradiometers are very delicate instruments that should never
be opened or disassembled.
The installation engineer has also to assure that the analysts using the
instrument fully understand the operation of the instrument and the procedures.
Especially in the collection of historical data for process control, meticulous
adherence to procedures and accurate records are essential for the ability
to perform time series analyses.
When several sites use the same process, then the procedures must be the
same, so it is possible to compare the data.
5.1.3 Operational qualification
The operability items ensure that the instrument works for the specific applications:
training for staff

5.1.3 Operational qualification
The operability items ensure that the instrument works for the specific applications:
training for staff
testing protocols for standards
testing protocols for subsystems
calibration procedure and defined test methods and procedures
challenging the system with known standards
record what is done and why it is done
ensure data transfer integrity to external software

In essence, the operational qualification requires that the procedures in this
report be followed and that the described tools be used. If the application
changes, then this report must be revised as necessary, historical data must be
safeguarded as necessary, and the analysts must be retrained.
The major immediate change is to port the database aspect from the simple
demonstration application described in this report to a database that is fully
integrated with the existing data storage and management system.
Another major change will be to upgrade the procedures to full ISO 9001 compliance.
Such a compliance does not only facilitate the business transactions
with suppliers and customers. It also entails a critical review of all procedures
and eventually leads to improved product quality, higher manufacturing
yields, and improved profits.
Accumulating data by itself does not serve any useful purpose. The linchpin of
the process is the design of data analysis algorithms that allow the datas use
to provide a feedback mechanism in the manufacturing line to keep the process
within specified tolerances. The quality of the collected data directly
determines the statistical confidence intervals for the correlation between colorimetric
data and manufacturing drifts.
Once sufficient historical data is available and data analysis algorithms are
designed, the tolerances described in this report will have to be reconsidered.
This reconsideration will produce a different set of qualifications.
5.1.4 Performance qualification
The performance qualification regulates the ongoing compliance, ensuring
that the instrument continues to work in the intended manner. The items in
this qualification include:
regular service program
performance monitoring with warning and action limits
individual log books and databases for spectrophotometers
regular peer review
simple change procedure
The essence of this report is to define a good set of procedures for the performance
qualification, and the following sections will describe each procedure in
detail. These procedures ensure that each instrument performs within certain
limits and assures the validity of the data generated.
Procedures 43
While the database designed for this demonstration deals only with measurements
per se, there should also be a database table to log all other events that
impact the data, such as
lamp changes
problems and service calls
instrument substitutions
firmware upgrades
optical upgrades
procedure changes
manufacturing line changes impacting the colorimetry
manufactured good changes (e.g., ink formulation)
environmental changes
personnel changes
peer reviews
With the last item we imply that each site is visited on a regular basis and the
procedures and data be reviewed by a peer or a consultant. This ensures that
the data of the various sites is fully exchangeable and the sites do not drift
away from each other.
Changes in ink formulation are important because they can lead to metameric
matches. Metameric matches require accuracy, i.e., absolute certification,
while the procedures described in this report are based on relative certification,
i.e., only precision is certified. This aspect may have to be reconsidered.
The log should not only exist digitally in a database, but should also exist in
printed form, because it is more convenient and robust to use during peer
reviews. For this reason, when we designed our demonstration database we
took care to design a printable form layout of each record, allowing physical
collection in a log book.
5.2 White calibration
The white calibration must be performed when one of the following events has
occurred:
begin a new measurement session
end of a pause
10 consecutive measurements have been performed
the ambient temperature has changed by more than 5C

Frequent white calibration is important in the case of process control, because
a drift in the instrument might be interpreted as a drift in the process. When
an automatic stage such as the SpectroScan is used, the controller software
must be programmed to perform the periodic calibration automatically. With
the SpectroScan, the reference tiles should be lodged permanently in one of
the apposite receptacles in the stage and the entire set-up should be shielded
from direct sun light.
When the instrument is used manually, the white calibration must be performed
in the following steps:
1. set the white reference parameter to absolute
2. set the filter wheel to the desired position (SPM-100) or apply the required
filter (Spectrolino)
3. check the white reference for cleanliness and place it on an even surface;
for the Spectrolino, always insert the reference tile in the special calibration
base (see Figure 2 on page 18)
4. adjust the spectrophotometer to the white reference and initiate the white
calibration
Using the calibration plate (9) and ensuring the instrument fits perfectly in
the positioning pins (6) is important. Ceramic tiles are never uniform and the
purpose of the base is to ensure you always measure the same spot of the tile.
For this reason it is necessary to remove the positioning foot (see Fig. 3) from
the Spectrolino and measure directly on the calibration plate holding the clip
(7) with the white reference. For details, see Construction on page 18.
If you have several instruments, make sure that the serial number on the
instruments identification plate is the same as that on the white reference (5).
5.3 Quarterly certification
Instrument error is not the only source of incorrect measurements. More often
the error is in the measurement procedure. This can have many causes, such
as documentation error, personnel changes, etc. The best insurance against
this type of problems is to participate in the CTS Collaborative Reference Program
for Color and Appearance.
5.3.1 Background
Collaborative Testing Services Inc. (CTS) is an organization that conducts
interlaboratory tests. In such a test, the performance of a laboratory is not
compared to an absolute reference. Instead, multivariate comparative statistical
methods developed in conjunction with the National Institute of Standards
and Technology (NIST) are used to gauge performance.

With these techniques a consensus value among a large number of participating
laboratories using a variety of instruments and methods is determined.
The values of each laboratory are then compared with this consensus value.
Such an analysis is also known as a real world analysis.
The concept of formalized, large-scale interlaboratory testing was developed
in the late 1960s by NIST (then known as the National Bureau of Standards,
NBS). In the mid 1970s, full responsibility for the laboratory proficiency testing
program was transferred to CTS. CTS maintains compliance with ISO
Guide 431996: Development and Operation of Laboratory Proficiency Testing.
It is important to understand that proficiency testing does not gauge the performance
of an instrument. Rather, as the name implies, it assesses the profi-
ciency of a laboratory.

5.3.2 Color and color difference
Analysts measure a set of three opaque color paint chips made by Munsell
Color, Macbeth Division. The colors are selected from throughout the full color
spectrum. The chips consist of a metameric and nonmetameric match with
small color differences.
The measurements are performed according to ASTM E 1349 immediately
after opening the sealed moisture-free barrier bag containing the specimens.
Each specimen is backed with the other specimens when making measurements.*
Each specimen is measured twice, rotating the specimen 90 for the
second measurement. This latter step is very important for the Gretag instruments,
because they are based on 45/0 geometry, which is susceptible to
polarization.
When you receive the report with the results, use your lab code to find the
data you have contributed. You will find one of two data flags: 0 or X. If the
data flag is 0, your results are consistent with those of the other participants
and you may proceed. If the data flag is X, stop work immediately and consult
an expert to review your data and your testing procedure. This review will
consist of four steps, as delineated in the decision tree in Fig. 16.
In the first step repeat the performance test and verify that you can approximately
reproduce the results in the report. With this you check for gross
errors, such as using the paint chips from an old test or measuring the chips
from the back side. We strongly recommend a peer review if this happened.
In the second step perform the certification procedure with your Green reference
tile. If this test fails, you probably have a defective instrument and

should repeat the CTS test with the backup instrument. If you would have
passed the test with the backup instrument, follow the procedure to replace a
defective instrument (not described in this document).
If the certification with your reference tile passes, proceed as follows with step
three. Request the panic reference tile (see Section 2.2.1 on page 15) and use
the certification database to perform an absolute certification. If this certification
passes, you made a procedural error in performing the CTS test. You
should obtain training in ASTM procedure E 134990 and the CTS test in particular.
All procedures in your laboratory should be reviewed by an expert.
Step four. If the absolute certification fails but the relative certification failed,
your laboratory is set up incorrectly and an expert should review all operations.
Returning to the CTS report, after reviewing the data flag, examine the column
labelled AE*. If the result is followed by an * your AE is greater than 2
but less than 3 standard deviations from the grand mean. You should review
your procedure and perform the relative certification procedure to establish if
you can continue working.
If the result is followed by an X, your AE is greater than 3 standard deviations
from the grand mean. In this case you should follow the decision tree in
Fig. 16.
5.3.3 Spectrophotometric analysis
In this test analysts report the reflectance data at 16 selected wavelengths in
the 400700 nanometer range (i.e., at 20 nm intervals) on one of the opaque
color chips. The same backing rule as in the previous test is followed. Also in
this test it is important to repeat the measurement after rotating the specimen
by 90 because of polarization in the instrument.
The results are evaluated as in the color & color difference test by examining
the data flag and the procedure explained in the previous section can be followed.
The reflectance values at single wavelengths can also be flagged. Since
our intended use of the spectrophotometer is just colorimetry, a few errors in
the reflectance values are not a cause for alarm.
5.3.4 Useful aids
One of the advantages of the CTS program is that it takes only a few minutes
each quarter to complete. There are a few tools to make this work even easier.
Copying the results from the screen into the CTS form is tedious and can
introduce errors. If you have the KeyWizard, download a convenient Acrobat
form from http://chroma.hpl.hp.com/Acrobat/Notebook/Colorimetry/cts.pdf (if
you are viewing this document on-line, the link is active and you can click on
it). JavaScript is used to calculate the data required in the CTS form from the
data provided by the SMP-100 or Spectrolino.
If you have an older version of the Acrobat Reader, you may have to download
the free Acrobat Forms Fill-in Plug-in 3.5 Update from the Adobe Web site.
This two-page form will automatically fill out the CTS form, which can be
printed and sent to CTS.
If you have Acrobat Exchange, you can save the blank form after you have
edited the laboratory number and analyst name. You can also save the completed
form for future retrieval or to submit it to a document repository. In any
case, keep a hard copy in the log book.
If you want to further examine the data, there is an Excel spreadsheet to tabulate
the data, plot the spectra, and compute the resultant AE of your laboratory.
This spreadsheet is available from http://chroma.hpl.hp.com/color/
CTS1098D65.xls.
Finally, the historical CTS data for Spectrolino serial number 10255 in Palo
Alto can be found at http://chroma.hpl.hp.com/color/cts.html.
5.4 Yearly calibration
Once a year the instrument should be sent to the manufacturer for the following
operations:
install firmware and hardware updates (this is to ensure that all instruments
are comparable and interchangeable)
measure the full set of color ceramic tiles, each mounted in the calibration
base
compute the error compared to an absolute instrument traceable according
to ISO 9001
compute the error compared to the average of 10 instruments of the same
construction
measure 10 times the white tile to assess the short term repeatability
The manufacturer will return the instrument with a calibration certificate
that must be stored in the document repository. An example of the yearly calibration
certificate can be found on the intranet at http://chroma.hpl.hp.com/
color/absCal98.html. This is the same data as in Section 2.4, Spectrophotometer,
on page 20.
5.5 Initial characterization of the Green tile
We have discussed earlier (Section 2.1.2 on page 13) the distinction between
absolute and relative certification. For the relative certification, when you
receive a new spectrophotometer you must first determine the aim values of
the green tile for this instrument. We refer to this preparation as the initial
characterization.
Since the color reference tile has a life time of 5 years, we recommend to collect
the data for a number of illuminants and filters, because it is possible that
during such a long period of time the viewing conditions chosen as a Corporate
standard can change. We suggest to characterize at least for two illuminants,
D50 and D65, and two filter settings, no filter (U for unfiltered) and D65 filter.
The observer angle is 2 and the color space is CIELAB.
Although it is sufficient to record the L*, a*, and b* coordinate values, for convenience
we require to record also the correlates for chroma and hue. We suggest
to also keep a record of the reflectance spectrum of the reference tile
measured with the various filters. The benefit of recording the entire spectrum
is with the CTS test (see Section 5.3.3 on page 47). When there are several
flagged values, it is a good idea to compare the values at those wavelength
for the color reference with the initial values. Finally, the spectral data can be
useful for diagnostic purposes when an instrument fails.
The steps are similar to those discussed earlier for the white calibration. It the
demonstration database, select Set up relative certification for an on-line
guide.
1. set the white reference parameter to absolute
2. set the first illuminant
3. set the observer angle
4. carry out the white calibration as described in Section 5.2 on page 43
5. create a new measurement set in the certification database
6. measure the color reference tile 10 times at a 10 second interval
7. record the color coordinates in the colorreference table of the database
8. repeat for each filter and illuminant
9. back up all database files
In step 6. we recommend to use the Color Measurement Set Acquisition view
in the measureset table of the database. If you cannot make 10 measurements,
we encourage to make at least 3 measurements at 10 second intervals and to
enter the average values in the color reference table. The average values are
displayed in the evaluation view of the measurement set table.
There is no command to transfer by program the color coordinates from the
measurement set table to the color reference table. This is intentional, so the
color reference table is a database by itself that can be easily redistributed
within the Company any time it is necessary. For example, this table could be
maintained in a central location and made available on the intranet through a
servelet. Depending on the platform, the color coordinates can be transferred
from one table to another using copy & paste or drag & drop.
In the color reference database, the part number for the clip with the green
tile that fits in the Spectrolino calibration plate is 36.65.25. It is important to
record the serial number because even if the ceramic tiles are from the same
batch, the uniformity is not perfect.
Always record the current ambient temperature. If the tile is stored in a closed
container, the container should be opened half an hour prior to use. The temperature
tolerance depends on the effect of thermochromism (see Section 6.4
on page 55) for the particular tile and the planned tolerance for the certification.
For example, the Green tile in the BCRA CCS II standard exhibits a change of
0.68 AE*ab units when the temperature changes by 10C, the measurement
geometry is spinc 0/t, and the illuminant is D65 with the CIE 10 observer [8].
A tolerance of 2C in this case translates to a measurement uncertainty of
0.68 * 0.4 = 0.27 CIELAB units. This is a large error in the case of process con-
trol, but it seems to be realistic in the real world. If the temperature excursion
can be reduced, then a smaller interval should be specified for the temperature
tolerance. The raw data in Table 16 on page 57 should be used to compute
the colorimetric differences for the particular instrument geometry and viewing
conditions prescribed.
Conceptionally, we treat a reference tile as a different color reference for each
viewing condition. In the database we store a separate record for each combination
of illuminant and observer angle. Since the serial number must be a
unique identifier, we have followed the convention of creating a serial number
by combining the serial number on the clip label (see Figure 2 on page 18,
item 5) with the illuminant and the observer angle.
For example, the colorimetric data for reference tile 221 measured for illuminant
A and the 2 CIE observer is stored in the record with the serial number
field 221-A-2. This convention is only a pro memoria and does not have to be
followed. (See Table 2 on page 16 for the colorimetric data.)
In Appendix VI on page 77 we reproduce the colorimetric values from the initial
characterization of five Green tiles with instrument 10255. These values
are just examples and cannot be used with other instruments.
5.6 Accuracy check
The accuracy check is a simple verification that must be repeated at the beginning
at every working shift for instruments that are used in process control.
For instruments used in less critical applications, such as color reproduction
or ICC profile creation, the verification must be repeated at least every three
months.
The first preparatory step in this procedure is to start the certification database
and browse the references until the relevant reference record is found.
Check the current room temperature and verify that it lies within the limits
indicated in the record, taking into account the tolerance of your thermometer.*
Complete all preparatory steps:
1. check room temperature
2. set the white reference parameter to absolute white
3. set the illuminant
4. set the observer angle

* The tiles should be stored at ambient temperature and if they are in a closed container the
container should be opened 30 minutes before the measurement.

Return to the certification window and select the view Perform certification. In
this view follow the instructions on the screen:
1. click on the New button to create a new record. The system generates a
new sequential certificate number and fills in the fields with the current
date and time
2. enter the filter on the spectrophotometer. When you proceed to the next
field, on the right of the filter designation you will see the warning
WRONG. Ignore this warning for the moment; at the end of the certification
it will indicate if the unit has detected the same filter as you indicated.
This is a consistency check
3. enter the serial number of the color reference tile. The pop-up menu shows
the entries in your color reference database
4. carry out the white calibration as described in Section 5.2 on page 43.
Enter the current room temperature. The system will instantly verify that
it lies within the limits; the instrument check will fail if the temperature is
outside the limits
5. check the color reference for cleanliness and place it in the receptacle
6. enter the instruments serial number and a description of the instrument
and the reference
7. click the Go button to jump to the measurement set window and make sure
it is set to the data acquisition view. On the bottom on the screen are the
instructions for the KeyWizard; you have already performed the white calibration
and a new record has already been created, so these two steps must
be omitted
proceed to step 5 in the instructions on the screen and perform the prescribed
number of measurements. This number is usually the same as the
number of repetitions you normally perform in the instruments operation
at the end go to the evaluation view and verify that the standard deviation
is small. If necessary, you can add a comment for future reference
8. return to the certification window, where you will notice that the system
has verified the settings on the instrument against the values in the color
reference database
click in the next two fields to refresh them
the first field shows the CIE94 color difference and the second field shows
the pass/fail message. Currently the tolerance for absolute measurements
is 0.8 and for relative measurements it is 0.3
9. Click on the Print button to print a certificate and file it in the log book
If the test fails, check for the following conditions:
is the room temperature correct?
are the white reference and the color reference both clean?
do the reference tiles have the correct serial numbers?
if you are using the SpectroScan, are the two clips in the correct bays?
if you are using the Spectrolino in hand-held mode, did you remove the
positioning foot?
is the filter clean?
After checking all conditions, repeat the accuracy check. If it fails again, follow
the rules for a failed quarterly certification summarized in Figure 16 on
page 46.

6 Thermal issues
In a manufacturing situation it is not always possible to control the room temperature.
For example, in a factory that has been built in four copies, the following
temperature ranges have been measured:
At first glance, these ranges seem to quite benign, especially when compared
to the typical operating temperature range for spectrophotometers, which is
from 10C to 40C. However, there are four important issues, especially when
ceramic tiles are measured: noise in the sensor, dimensional stability of the
monochromator, heating the sample, and thermochromism in the sample.
6.1 Sensor noise
CCDs have a temperature-dependent leakage current that flows even when no
light impinges upon the element. This signal is called dark current. Modern
instruments no longer require cooling with a Peltier element and instead dissipate
heat produced by the light source with passive systems.
Because customers want small instruments and short measurement intervals,
instrument designers tend to dimension the cooling elements small and the
measurement intervals short. In a well-balanced system, the noise will be just
under the precision stated in the instrument specifications.
The instrument specifications are tailored to the market in which the instrument
is sold. For example, an instrument sold in the graphic arts market will
not require a precision larger than 0.5 units. Since this is a relatively large
market, such instruments are inexpensive, and since the target users are not
technicians, the user interfaces tend to be very user-friendly.
If an isolated measurement is taken with such an instrument, the precision is
typically much better than stated in the specifications. This leads us use
graphic arts instruments in process control applications. The important point
is not to perform measurements and the maximum rate allowed by the instru-
Location Low temperature High temperature
Dublin 19C 23C
Puerto Rico 21C 22C
Corvallis 22C 23C
Singapore 24C 26C
TABLE 13. Room temperature in various factories
ment, but to delay measurements until the heat generated by the lamp has
completely dissipated. In the case of the Spectrolino, this interval is 10 seconds.
This is why in the validation procedures we prescribe to wait for 10 seconds
between measurements and why we log the time of each measurement.
6.2 Optics stability
Heat also causes the optics in the instrument to expand. A common device
used as a monochromator is a holographic diffraction grating. When the grating
expands, the spectrum interval detected by the sensor moves and the
instruments calibration is no longer valid. The instrument designer can minimize
this effect by choosing a material that has a very good dimensional stability,
but there is a cost trade-off.
The user has two choices to avoid this problem. The first choice is to recalibrate
the instrument frequently, say every 10 measurements. The second
choice is to wait between measurements until the instrument has completely
cooled, which in the case of the Spectrolino is 10 seconds.
To err on the safe side, in our procedures we have recommended to do both,
wait for 10 seconds and recalibrate every 10 measurements. It is now clear to
the reader that the calibration can be much less frequent. We have not performed
experiments to determine the exact required calibration frequency, but
it might be an order of magnitude larger.
In the specific case of the Spectrolino, in the models designated as Rev. A or
newer the diffraction grating is on a new glass type that is dimensionally very
stable under temperature changes.* With this improvement it will be worthwhile
to relax the calibration requirements.
6.3 Sample heating
A typical light source for spectrophotometers is a gas filled tungsten lamp.
Such a lamp has a strong radiation in the infrared part of the spectrum, which
can heat up the sample. In the Spectrolino, for example, the filament heats to
3000K on a surface of 1.5 mm 0.5 mm.
If the measurement time is kept very short to minimize this effect, then only a
small number of photons will be available at the sensor and the dark current
will cause a very poor signal to noise ratio. The classical solution to this problem
is to keep the light on until the sensor saturates, and then to determine
the spectral power from the integration time of the radiant flux. With dark
samples this time can be quite long and considerably heat the sample.
* In the Spectrolino S/N 10255 used to obtain the data for this report we have upgraded the
grating to the new type.
Thermal issues 55
We do not know how the Spectrolino and the SPM-100 work, so we have
assessed the problem with an experiment. We set the Spectrolino with serial
number 10255 to the following conditions:
illuminant D65
2 observer
no filter
no data transfer
We measure the temperature of a green reference tile (serial number GBB-1),
then we perform 30 measurements at the maximal speed allowed by the
instrument and immediately measure the tiles temperature again. The result
of three repetitions is shown in the table below.
The error due to thermochromism would be approximately 0.01 CIELAB
units. Given that in the certification procedures the tile is measured at most
10 times and there is a 10 second wait between measurements, heating of the
ceramic reference tile by the instrument during a measure is not a problem.
6.4 Thermochromism
Thermochromism refers to a color change with temperature change. Among
others, it occurs with some tiles that contain selenium. Anne Compton
reported [6] that when a tile is heated from 25 to 35C a typical temperature
change in the instruments using prolonged polychromatic irradiaton
common at that time the effect is a small reduction in reflectance with an
increase in temperature. She observed that in general thermochromism
occurs on steep spectral profiles. The largest value of AE*ab occurred for the
dark green tile and was 0.9, a difference that is larger that the tolerances in
our procedures. The instrument geometry was diffuse/0.
Mark Fairchild and Frank Grum [7] conducted a similar experiment* with an
instrument of 0/45 geometry. For the same temperature change they
observed a AE*ab value of 0.7 for a green tile. However, they observed much
larger changes for the Orange (1.6) and Red (1.5) BCRA Series II (CCS II)
tiles.
More recently the National Physical Laboratory (NPL) has released a comprehensive
collection of spectral reflectance and thermochromism data [8] on the
NPL Master Set of the BCRA CCS II tiles. The colorimetric data for the green
tile is shown in Table 15 for spex and spinc geometry; unfortunately no data is
provided for 0/45 geometry, but the interested reader can calculate it from
the data provided in the article. Some data for the other tiles is in Table 17.
In the last column is the deviation of Spectrolino S/N 10255 from the average
values obtained from 10 Spectrolinos and reported in Table 6 on page 21.
Thermochromism is linear, so if we take from Table 13 on page 53 the maximum
temperature span of 7C, for the spinc measurement we would obtain a
difference of 0.48 AE*ab. This value is almost twice the inter-instrument
agreement, i.e., if ambient temperature is ignored during measurements, a
significant error is introduced and the data is less valuable.
Remember also that the tolerance for the relative accuracy check in
Section 5.6 on page 50 is 0.3. Hence, a temperature difference of 4C is by
itself the limit to pass the validation, assuming the instrument is perfect.
With the 2C temperature tolerance prescribed in that procedure, a 0.12 unit
error is already introduced by the temperature uncertainty, leaving 0.18 units
for drifts in the instrument. The Dublin factory, with a temperature excursion
from 19C to 23C, will have a challenging time to pass the instruments quali-
fication. In reality, the procedure described in this report is more lenient,
because it is based on the CIE94 color difference, which is closer to CMC.

In time series analysis we examine a long series of measurements and try to
find trends. For highly significant assertions the analyst will always use the
same instrument. If the operation is very meticulous, then the repeatability
will be close to the short-term repeatability, which for Spectrolino S/N 10255 is
0.023 AE*ab units. This number is an order of magnitude smaller than the
effect of thermochromism with the 2C tolerance, hence temperature variations
will limit the usability of the calibration data.
This last qualification is important. Ink on paper most likely does not suffer
the thermochromism problem, so the measurements by themselves do not
limit the precision of time series analysis. The limitation is when the measurements
of the green tile are used to calibrate the data in the time series.
What we have seen is that this calibration is limited by the uncertainty of the
temperature.
If the temperature is stable for a period of 30 minutes (the time for a ceramic
tile to adapt to a small temperature change) or longer, and an accurate thermometer
is available, for example a mercury in glass thermometer graduated
in 0.1C intervals, then highly accurate measurements can be made by correcting
the spectral data for the temperature difference before the colorimetric
quantities are computed.
The correction data is listed in the following table listing the differences in
reflectance caused by a 10C (25 to 35C) change in temperature for the Green
tile in the Master Set of the BCRA CCS II tiles:
Wavelength [nm] difference
380 0.0003
390 0.0002
400 0.0002
410 0.0001
420 0.0001
430 0.0002
440 0.0003
450 0.0005
460 0.0009
470 0.0017
480 0.0030
490 0.0049
500 0.0065
510 0.0064
520 0.0048
530 0.0033
540 0.0020
550 0.0009
560 0.0006
570 0.0002
580 0.0001
590 0.0004
600 0.0005
610 0.0006
620 0.0006
630 0.0007
640 0.0007
650 0.0007
660 0.0007
670 0.0007
680 0.0006
690 0.0005
700 0.0003
710 0.0001
720 0.0002
730 0.0004
Wavelength [nm] difference

The gray curve plots the NPL data for the difference in reflectance caused by a
10C (25 to 35C) change in temperature (units on right axis) tabulated in
Table 16 above.
Table 17 from reference [8] shows the colorimetric differences caused by a
10C temperature change. From this data one might assume that the cyan or
even the deep blue tile would be a better choice for the problem at hand. However,
as we noted in the discussion of Figure 1, Reflectance of the green and
cyan master tiles, on page 15, the green tile with is sloped spectral distribution
curve is better suited for detecting the critical shift of the instruments
scale.
The deep blue tile exhibits the lowest thermochromism, but with its flat spectral
distribution curve, by itself it is not useful at all to certify an instruments
calibration. In summary, the green tile is the most suitable tile for certifying a
spectrophotometers calibration; due to its thermochromic properties, ambient
temperature must be controlled and recorded.
Spectral reflectance of green tile GBB-1 measured with Spectrolino 10255 at 24.41C
(left ordinate) and difference in reflectance caused by a 10 change in temperature
from 35C (right ordinate).
Red 1.01 0.56
Orange 0.97 0.85
Bright yellow 0.76 0.44
Green 0.68 0.37
Difference green 0.72 0.39
Cyan 0.32 0.22
Deep blue 0.06 0.04
TIle AE*ab AE CMC (1:1)
Introduction
There are many practices for making spectral measurements and performing colorimetric
computations. The methods used
may result in different numerical values for the same parameter for the same sample. Thus, one
may not be able to make
valid comparisons using data derived using different methodologies. The purpose of this standard
is to specify a
methodology for the measurement of graphic arts images that result in valid and comparable
data. While this standard
references the standard established for graphic arts viewing conditions, it is not intended to
provide an absolute correlation
with visual color appearance.

1
Graphic technology Spectral measurement and
colorimetric
computation for graphic arts images
1 Scope and field of application
This standard establishes a methodology for reflection and transmission spectral measurement,
and computation of
colorimetric parameters for graphic arts images. Graphic arts includes, but is not limited to, the
preparation of material for,
and volume production by, production printing processes that include offset lithography,
letterpress, flexography, gravure
and screen printing.
This standard also applies to images to be manufactured in limited quantities such as those
produced with photographic, ink
jet, thermal transfer, diffusion, electrophotography, mechanical transfer or toner technologies
(e.g., off-press proofs) when
used for graphic arts applications.
This standard does not address any requirements on the measurement of emission spectral data
from video monitors nor
does it supersede the specification of other measurement geometries appropriate to specific
application needs, such as the
evaluation of materials (e.g., ink and paper) used in the graphic arts.
NOTE Procedures for color measurement of spectral data from video monitors are included in ASTM E
1336: "Standard test method
for obtaining colorimetric data from a video display unit by spectroradiometry."[1] The use of integrating
sphere geometry for paper
evaluation is covered in ISO 2469, Paper, board and pulps Measurement of diffuse reflectance factor.[2]
2 Normative references
The following standards contain provisions that, through reference in this text, constitute
provisions of this standard. At the
time of publication, the editions indicated were valid. All standards are subject to revision, and
parties to agreements based
on this standard are encouraged to investigate the possibility of applying the most recent editions
of the standards indicated
below.
ASTM E 308-99 "Standard Practice for Computing the Colors of Objects by Using the CIE
System," American Society for
Testing and Materials
CIE Publication No. 15, Colorimetry
ISO 5-2, PhotographyDensity measurements Part 2: Geometric conditions for transmission
density
ISO 5/3, PhotographyDensity measurements Part 3: Spectral conditions
ISO 5-4, PhotographyDensity measurements Part 4: Geometric conditions for reflection
density
ISO 3664, Viewing conditions Graphic technology and photography
CGATS.17, Graphic technology Exchange format for spectral measurement, colorimetric, and
densitometric data in
electronic form
NOTE See the Bibliography for further information listing other documents that are not necessary for
compliance with this standard
but may be of interest to the reader. It also contains contact information for sources of relevant documents.
Referenced documents in this
text will be shown as [number] for reference.
3 Definitions, abbreviations, and symbols
For the purposes of this standard, the following definitions, symbols and abbreviations apply:
3.1
calibration
set of operations that establish, under specified conditions, the relationship between values of
quantities indicated by a
measuring instrument or measuring system, or values represented by a material measure or a
reference material, and the
corresponding values realized by standards
[ISO International Vocabulary of Basic and General Terms in Metrology]
NOTE Contrary to a common usage, calibration is not the process of adjusting a measurement system such
that it produces values
that are believed to be correct. Calibration permits either the assignment of values of measurands to the
indications (creating a reference
table) or the decision to reset or adjust the device. Following the resetting or adjusting of the device, a
calibration should be repeated to
ensure that the new device setting(s) provide indications within the accepted values.
3.2
CIE standard illuminants
illuminants defined by the CIE in terms of relative spectral power distributions; e.g., A, C, various
D and F illuminants
3.3
illuminant
radiation with a relative spectral power distribution defined over the wavelength range that
influences object color perception
[CIE Publication 17]
3.4
measurement source
characteristic of the radiant flux (light) incident on the specimen surface
3.5
opacity of substrate (0/45, D50, 2 degree, CIE Y)
Os
measure of the property that describes the ability of a sample to hide a surface behind and in
contact with it
3.6
reflectance factor (at a surface element, for the part of the reflected radiation contained in
a given cone with
apex at the surface element, and for incident radiation of given spectral composition,
polarization and
geometric distribution)
R
ratio of the radiant or luminous flux reflected in the directions delimited by the given cone to that
reflected in the same
directions by a perfect reflecting diffuser identically irradiated or illuminated
NOTE The term commonly used in the industry and in this document for reflectance factor is reflectance.
3.7
sample backing
material placed behind the sample during measurement
3.8
self-backing
sample backing composed of multiple sheets of substrate material identical to the substrate on
which the image has been
placed
3.9
transmittance/ (for incident radiation of a given spectral composition, polarization, and
geometrical
distribution)
ratio of the transmitted radiant or luminous flux to the incident flux in the given conditions
NOTE It is important to specify the geometry that establishes the given conditions of measurement.
3.10
transmittance factor
ratio of flux transmitted by a specimen in a given optical system to the flux transmitted when the
specimen is removed from
the sampling aperture. This is the case when e.g. radiation penetrating a slide situated in a
projector and reaching a screen
is compared to the radiation when the slide is removed from a projector and only an empty slide
mount is in the projector.
4 Spectral measurement requirements
4.1 Instrument calibration
The measurement device or system shall be verified (calibrated and possibly adjusted) in
accordance with its manufacturer's
instructions. See Annex A, Instrument calibration.
NOTE CGATS.11/PIMA IT2.11 [5], defines the use of a certified reference material (CRM) to check
calibration of a measurement
system and provides additional information relating to the use of CRMs, the determination of combined
standard uncertainty and data
reporting requirements.
Where multiple instruments are used for measurement there will be differences in the resulting
data due to the individual
characteristics of the instruments and variations in measurement conditions. Annex B provides
information on what could be
done to help multiple laboratories or areas obtain consistent measurement data.
4.2 Measurement source for fluorescing materials
To minimize the variations in measurement results between instruments due to fluorescence, the
measurement source for
determination of spectral reflectance data for colorimetric calculations shall match the specified
CIE illuminant for the
intended application over the wavelength range of potential energy absorption. Subclause 5.1
specifies the use of CIE
illuminant D50 for graphic arts colorimetric computations. Annex C provides further information on
fluorescence and
techniques to test for its presence.
NOTE It is recognized that many instruments presently do not have a measurement source that matches
this illuminant. These
instruments are technically not in compliance with CGATS.5 and the user is cautioned when reporting data
for samples that fluoresce.
4.3 Wavelength range and interval for measured values
The data should be measured from 340 nm to 780 nm at 10 nm intervals and shall be measured
from 400 nm to 700 nm,
inclusive, at intervals of no more than 20 nm. The reference for spectral data shall be based on
computed data at 10 nm
intervals where the spectral response function is triangular with a 10 nm bandwidth at the half
power point.

4
NOTE Instrumentation with different intervals and response functions may produce different results. These
differences may be
reduced by proper selection of bandpass characteristics (shape) for a given interval and applying the proper
method of calculation for the
bandpass characteristic and interval selected.
4.4 Reflectance factor measurement
4.4.1 Sample backing material
This standard makes provision for a black backing, or where appropriate a white backing or a
self-backing (see 3.8)
condition for use during measurement.
Substrates with opacity of substrate as defined in 4.4.1.1, equal to or greater than 99, are
considered opaque for the
purposes of this standard and the backing used for measurement is not relevant.
The black backing in 4.4.1.2 shall be used where back printing is present (text or images
printed on both sides of the
substrate).
The black backing in 4.4.1.2 should be used where the substrates opacity is 95 or above.
The white backing in 4.4.1.3 or the self-backing in 4.4.1.4 should be used where the
substrates opacity is below 95.
NOTE 1 ISO 5-4 requires that black backing be used for spectral measurements from which ISO standard
status densities are to be
computed. Therefore, densitometric data computed from measurements made using white backing does not
comply with the ISO 5-4
standard. Further, colorimetry data computed with white backing may not be applicable for some process
control applications.
NOTE 2 Users are cautioned that where self-backing is used issues of diffuse blurring need to be addressed
and the accuracy of
typical instruments and targets used in graphic arts may need to be addressed for this issue.
The sample backing chosen, shall be reported as in 6.2 i) and 6.3.3, Data reporting.
NOTE 3 The choice of sample backing material appropriate to the application has become important
especially in color management
due to the differences in opacity of proofing system substrates and production printing paper materials.
There may be no correct answer,
only application specific compromises. When visual comparisons are to be related with measured
colorimetric data, the backing used
during viewing may strongly influence color and tone appearance. Similarly, the backing used during
measurement may have an impact
on the data obtained depending on the opacity of the substrate materials. Therefore, the backing used
during visual evaluation should be
similar to the backing present during measurement. See the Committee for Graphic Arts Technologies
Standards Recommended Industry
Practice: Developing a color characterization data set Analysis and reporting (currently under
development) for further information.
NOTE 4 At the time of issue CGATS has not identified commercial sources of the black or white backing
materials.
See Annex D for more detailed discussion on the issues related to sample backing during
measurement.
4.4.1.1 Determination of opacity of substrate, Os (0/45, D50, 2 degree, CIE Y)
The following procedure shall be used to determine the opacity of substrate (0/45, D50, 2 degree,
CIE Y).
Two measurements shall be made of the material, one backed with a black backing material
(4.4.1.2) and the second
backed with a white backing material (4.4.1.3).
Compute the opacity of a substrate (0/45, D50, 2 degree, Y), using the following:
Os = (Yb/Yw) x 100 (1)
where
Os is opacity (0/45, D50, 2 degree, Y)
Yb is the Y tristimulus value computed from measurements made using black backing

5
Yw is the Y tristimulus value computed from measurements made using white backing
NOTE Although the symbol 0/45 is used to designate the measurement geometry, 0/45 and 45/0 geometry
are assumed to be
equivalent and either may be used.
4.4.1.2 Black backing
A black sample backing material, as defined in ISO 5-4 shall be used.
NOTE ISO 5-4 defines the following requirements:
spectrally non-selective, i.e., the total range of spectral diffuse reflection density throughout the
wavelength interval from 400 nm to
700 nm shall not exceed 5% of the average density obtained over the same interval;
diffuse-reflecting (i.e., no perceptible specular reflection when viewed at any angle under typical office
room illumination conditions);
and
an ISO visual reflection density of 1.50 0.20. (This is approximately equivalent to a CIE L* value range
of 15 - 27 or a reflectance
range of 2% - 5%).
4.4.1.3 White backing
A white sample backing material shall have the following characteristics:
shall be known to be opaque (e.g., ceramic or other coated materials);
diffuse-reflecting (i.e., no perceptible specular reflection when viewed at any angle under
typical office room illumination
conditions);
CIELAB C* less than or equal to 3.0;
non-fluorescing (no emission in the response band of interest when excited by the instrument
source); and
spectral reflectance values greater than those shown in Table 1 in the range 400 700 nm.
Table 1 Spectral reflectance values
Wavelength Spectral reflectance
400 0.30
410 0.30
420 0.75
450 0.75
460 0.80
670 0.80
680 0.75
700 0.75

6
NOTE This corresponds to a CIE L* greater than 92.
4.4.1.4 Self-backing
When measurements are made using self-backing, multiple layers of material, identical to the
imaged substrate, shall be
used such that the opacity of the total pad is greater than 99 when determined as defined in
4.4.1.1. To be considered
identical the backing material should be from the same lot/batch/roll as the imaged substrate
and not exhibit visible color
appearance differences. The CIE color difference values between the individual sheets making
up the stack of backing
material, and between the individual sheets and an unprinted area of the imaged substrate should
vary no more than 1.0
*
ab E . .
4.4.2 Measurement geometry
Measurement geometry shall be 45/0 or 0/45 and conform to the geometric conditions defined in
ISO 5-4. The sample shall
lie on a flat surface while being measured. The instrument base and the sample shall lie in the
same plane.
NOTE 1 The use of 45/0 or 0/45 geometry will not adequately address variations in all surface
characteristics. Other instrumentation
can be used to detect specific characteristics such as "bronzing." See Annex E.
NOTE 2 It is recognized that many instruments do not conform to the ISO 5-4 requirement for a 2 mm
boundary beyond the sampling
aperture due to the physical size of the control elements which are normally measured. Annex F provides
further information on aperture
size.
4.4.3 Resolution
Reflectance factor shall be measured to the nearest 0.0001 relative to a perfect reflecting diffuser
having 1.0 reflectance
factor at all wavelengths. This data may be reported as either reflectance factor or percent
reflectance (i.e., reflectance
factor multiplied by 100).
4.5 Transmittance factor measurement
4.5.1 Measurement geometry
Measurement geometry shall be normal/diffuse (0/d) or diffuse/normal (d/0) and conform either to
the geometric conditions
defined in ISO 5-2 or CIE 15. The diffuse geometry used, sphere or opal, shall be reported. (See
Annex E.3.)
4.5.2 Resolution
Transmittance factor shall be measured to the nearest 0.0001 relative to a perfect transmitting
diffuser having a
transmittance factor of 1.0 at all wavelengths. This data may be reported as either transmittance
factor or percent
transmittance (i.e., transmittance factor multiplied by 100). See Annex E.3.
5 Colorimetric computation
5.1 Calculation of tristimulus values
To provide consistency with graphic arts viewing conditions, defined in IS0 3664, calculated
tristimulus values shall be based
on CIE illuminant D50 and the CIE 1931 standard colorimetric observer (often referred to as the
2 standard observer) as
defined in CIE 15. Computation shall be either at 10 nm or 20 nm intervals. Factors representing
the product of CIE
illuminant D50 and the 2 standard observer data, to be used for weighting spectral reflectance
and transmittance data shall

7
be those given in Table 2 for 10 nm intervals and Table 3 for 20 nm intervals, as taken from
ASTM E 308. The user is
strongly encouraged to use data at 10 nm intervals to improve the accuracy of the results.
NOTE 1 The 2 standard observer was selected rather than the 10 standard observer, because it more
closely matches the size of
image detail found in printed material.
NOTE 2 ASTM has introduced an additional set of weighting functions for the computation of tristimulus
values from spectral data in
the most recent versions of ASTM E 308. These new tables are known as Tables 6 (of ASTM E 308) and
are intended for use where
measurement data have not been corrected for spectral bandpass dependence. The weighting functions
used here and in the 1995
edition of CGATS.5 are based on Tables 5 of ASTM E 308, which are intended for use where measurement
data have been corrected
within the measuring instrument for spectral bandpass dependence.
Since currently available instruments do not provide information regarding the inclusion of corrections for
spectral bandpass dependence,
CGATS recommends that no change be made at this time. If instruments include bandpass correction in
their reported spectral data then
Tables 5 are appropriate for use. Where they do not, Tables 6 are more appropriate for use. The small
differences in results between
using Tables 5 versus Tables 6 do not justify the confusion that would be introduced into the graphic arts
industry by a change at this
time. For typical graphic arts images (IT8.7/1 [6] scanner targets or IT8.7/3 press characterization targets
[8]) the CIE *
ab E . color
difference between the use of these tables is an average of approximately 0.2 and a maximum of 0.7.
If the measured spectral data begin at a wavelength greater than 360 nm, then all the weighting
values in Table 1 or Table 2
for wavelengths less than the first measured wavelength shall be summed and added to the
weighting value for the first
wavelength measured. If the last measured spectral data point is at a wavelength less than 780
nm, then all the weighting
values in Table 1 or Table 2 for wavelengths greater than the last measured wavelength shall be
summed and added to the
weighting value for the last measured wavelength.
NOTE 3 This procedure is consistent with ASTM E 308-99.
The general form of these computations is:
=780 =780
X = (R x W(X)) X = (T x W(X)) (2)
=360 =360
=780 =780
Y = (R x W(Y)) Y = (T x W(Y)) (3)
=360 =360
=780 =780
Z = (R x W(Z)) Z = (T x W(Z)) (4)
=360 =360
where
R is the reflectance value at wavelength ;
T is the transmittance value at wavelength ;
W(X) is the weighting factor at wavelength for tristimulus value X;
W(Y) is the weighting factor at wavelength for tristimulus value Y;
W(Z) is the weighting factor at wavelength for tristimulus value Z.
If measured data is at intervals and bandpass smaller than 10 nm, the method of Annex G shall
be used to widen the
bandpass of the data.
NOTE 4 The weights given in Table 1 or Table 2 are based on triangular bandpass characteristics as
referred to in 4.3.
The values of Xn=96.422, Yn=100.000 and Zn=82.521 shall be used to do colorimetric
calculations.

8
NOTE 5 Adding the values of the weights from 360 nm to 780 nm in Table 2 or in Table 3 does not give a
sum equal to the values for
Xn, Yn and Zn. This is because the writers of ASTM E 308 computed Xn, Yn and Zn to greater precision than
given by the summation of
the table values. The sums for X, Y and Z in the tables are of value as a data entry check of the tables.

9
Table 2 Spectral weights (W) for illuminant D50 and 2 observer
for calculating tristimulus values at 10 nm intervals
Wavelength (nm)
W(X)1
W(Y)1
W(Z)1
360 0.000 0.000 0.001
370 0.001 0.000 0.005
380 0.003 0.000 0.013
390 0.012 0.000 0.057
400 0.060 0.002 0.285
410 0.234 0.006 1.113
420 0.775 0.023 3.723
430 1.610 0.066 7.862
440 2.453 0.162 12.309
450 2.777 0.313 14.647
460 2.500 0.514 14.346
470 1.717 0.798 11.299
480 0.861 1.239 7.309
490 0.283 1.839 4.128
500 0.040 2.948 2.466
510 0.088 4.632 1.447
520 0.593 6.587 0.736
530 1.590 8.308 0.401
540 2.799 9.197 0.196
550 4.207 9.650 0.085
560 5.657 9.471 0.037
570 7.132 8.902 0.020
580 8.540 8.112 0.015
590 9.255 6.829 0.010
600 9.835 5.838 0.007
610 9.469 4.753 0.004
620 8.009 3.573 0.002
630 5.926 2.443 0.001
640 4.171 1.629 0.000
650 2.609 0.984 0.000
660 1.541 0.570 0.000
670 0.855 0.313 0.000
680 0.434 0.158 0.000
690 0.194 0.070 0.000
700 0.097 0.035 0.000
710 0.050 0.018 0.000
720 0.022 0.008 0.000
730 0.012 0.004 0.000
740 0.006 0.002 0.000
750 0.002 0.001 0.000
760 0.001 0.000 0.000
770 0.001 0.000 0.000
780 0.000 0.000 0.000
Check Sums
White point
X = 96.421
Xn= 96.422
Y = 99.997
Yn= 100.000
Z = 82.524
Zn= 82.521
NOTE These weighting functions are extracted with permission from Table 5.9 of
ASTM E 308-99.

10
Table 3 Spectral weights (W) for illuminant D50 and 2 observer
for calculating tristimulus values at 20 nm intervals
Wavelength (nm) W(X)1 W(Y)1 W(Z)1
360 -0.001 0.000 -0.003
380 -0.007 0.000 -0.034
400 0.100 0.001 0.459
420 1.651 0.044 7.914
440 4.787 0.325 24.153
460 4.897 1.018 28.125
480 1.815 2.413 15.027
500 0.044 6.037 4.887
520 1.263 13.141 1.507
540 5.608 18.442 0.375
560 11.361 18.960 0.069
580 16.904 16.060 0.026
600 19.537 11.646 0.014
620 15.917 7.132 0.003
640 8.342 3.245 0.000
660 3.112 1.143 0.000
680 0.857 0.310 0.000
700 0.178 0.064 0.000
720 0.044 0.016 0.000
740 0.011 0.004 0.000
760 0.002 0.001 0.000
780 0.001 0.000 0.000
Check Sums
White point
X = 96.423
Xn= 96.422
Y = 100.002
Yn= 100.000
Z = 82.522
Zn= 82.521
NOTE These weighting functions are extracted with permission from Table 5.10 of
ASTM E 308-99.

11
5.2 Calculation of other colorimetric values
Colorimetric values and color differences shall be calculated using the equations of CIE
Publication 15. These equations, as
well as equations used to compute color differences, are included in Annex H for information.
5.3 Calculation of densitometric values
The computation of densitometric values from spectral data is defined in ISO 5/3.
NOTE The ISO 5 series of documents is currently being revised and is expected to be available at the end
of 2002. ISO 5/3 will
contain the agreed upon procedures to compute the various status densities from spectral data. For further
information the reader is
directed to contact TC42 Secretariat at http://www.pima.net/standards/iso/tc42/WG21/WG21_pow.htm.
6 Data reporting
6.1 General
There is no requirement that data be reported electronically. Where data is reported
electronically, the provisions of 6.3
shall be followed.
6.2 Required information
When data generated in accordance with this standard are reported, they shall be accompanied
by the following information:
a) a statement that the measurements and computations are in conformance with CGATS.5;
b) the originator of the data;
c) the creation date of the data;
d) a description of the purpose or contents of the data being exchanged;
e) a description of the instrumentation used, including, but not limited to, the brand, model
number and sampling aperture
information;
f) the measurement source (light source and filter conditions used);
NOTE Examples of filters to report include uv, polarisation, light balancing, etc.
g) the wavelength interval (10 or 20 nm) used for calculation;
h) the spectral products weighting function (Status or Type response) when density data is
reported; and
i) the sample backing (black, white, self-backing or na). The spectral reflectance of the white
backing or the stack of selfbacking
material shall be reported.
6.3 Electronic data reporting
6.3.1 General
When data is reported in electronic form it shall be in compliance with one of the data formats
(ASCII and XML) of
CGATS.17.

12
The first 7 characters in the file shall be CGATS.5
The primary definition of keyword is contained in CGATS.17 and repeated here for information.
Where a conflict exists
between the definitions of keywords in this standard and those contained in CGATS.17,
CGATS.17 shall take precedence.
6.3.2 Required keywords or corresponding XML tags
The following keywords (or corresponding XML tags) as defined in CGATS.17 shall be used:
ORIGINATOR Identifies the specific system, organization or individual that created the data file.
DESCRIPTOR Describes the purpose or contents of the data file.
CREATED Indicates date of creation of the data file.
INSTRUMENTATION Used to report the specific instrumentation used (manufacturer and model
number) to
generate the data reported. This data will often provide more information about the
particular data collected than an extensive list of specific details. This is particularly
important for spectral data or data derived from spectrophotometry.
MEASUREMENT_SOURCE Illumination source used for spectral measurements. This data helps
provide a guide to
the potential for issues of paper fluorescence, etc.
NUMBER_OF_FIELDS Number of fields (data format identifiers) that are included in the data
format definition that
follows. The value associated with this entry is an integer.
BEGIN_DATA_FORMAT Marks the beginning of a definition of a data set. Data format identifiers
indicate
the relationship between fields, and their value. Data format may have an arbitrary
number of entries, and is closed by END_DATA_FORMAT. The values of the data format
must be either data format identifiers or keywords.
END_DATA_FORMAT Marks the end of a data format definition. END_DATA_FORMAT must be
preceded by
BEGIN_DATA_FORMAT, and must be followed by white space.
NUMBER_OF_SETS Number of repetitions or sets of data corresponding to the data format
fields that are
included in the data to follow. The associated value is an integer.
BEGIN_DATA Marks the beginning of a stream of data.
END_DATA Marks the end of a stream of data.
6.3.3 Data type dependant keywords
Where data is based on reflectance measurements the following keywords shall be used:
SAMPLE_BACKING Identifies the backing material used behind a hard copy sample during
reflectance
measurement. Allowed values are black, white, "self-backing", or na.
NOTE When sample substrates are opaque (i.e., Os is greater than 99% according to 4.4.1.1) the sample
backing is not an issue
and should be reported as na.
Where data is based on transmittance measurements the following keywords shall be used:
DIFFUSE_GEOMETRY The diffuse geometry used. Allowed values are sphere or opal

13
6.3.4 Optional keywords
Optional Keywords for use in reporting data are defined in CGATS.17.

14
Annex A
(informative)
Instrument calibration
A.1 Introduction
Webster's Dictionary defines calibration as, "to standardize by determining the deviation from a
standard so as to ascertain
the proper correction factors." It also defines adjustment as, "a correction or modification to reflect
actual conditions." Both
are used with measuring instrumentation. The definitions indicate a means of normalization and
both, if used correctly, will
yield accurate values of a measured sample.
According to the ISO International Vocabulary of Basic and General Terms in Metrology,
calibration is a set of operations
that establish, under specified conditions, the relationship between values of quantities indicated
by a measuring instrument
or measuring system, or values represented by a material measure or a reference material, and
the corresponding values
realized by standards.
Contrary to a common usage, calibration is not the process of adjusting a measurement system
such that it produces values
that are believed to be correct. Calibration permits either the assignment of values of measurands
to the indications
(creating a reference table) or the decision to reset or adjust the device. Following the resetting or
adjusting of the device, a
calibration should be repeated to ensure that the new device setting(s) provide indications within
the accepted values.
The user should be aware that even though they have calibrated their instrument using the
manufacturers calibration
process, the instrument may have only been verified relative to the calibration reference.
A.2 Calibration, adjustment and standardization of measuring systems
When trying to determine the procedure for calibrating a measuring device, one can sometimes
be confused as to what the
procedure actually does. If one picks up a meter stick, he would assume that the meter stick had
been properly
manufactured and is therefore calibrated. Yet, with very precise measurements, technically aware
people know that the
meter stick will only be accurate within a certain temperature range. If it is too cold or too hot, the
material may have
changed its physical dimensions thereby throwing it out of "calibration".
Even so, the meter stick can be compared to another meter stick "calibrated" at the higher
temperature. Once the bias for
the original meter stick is known, the original meter stick can then be used to make very accurate
measurements if a
correction factor for calibration adjustment is used. This concept is not often used but is actually
technically acceptable in
most circumstances. Once the correction factor has been established, the user can say the meter
stick has been calibrated
for the higher temperature, in which case a description of the conversion factor must include its
intended target temperature
and temperature range.
Many see the act of adjusting the instrument to a predetermined value or values as calibration. As
described above, this is
not necessary to calibrate a device. Additionally, it may do nothing to calibrate a device. A case in
point is an analog voltage
meter. With this type of device, the meter's indicator needle can be adjusted to properly align with
zero. Is this calibrating the
device? Certainly the "Zero" value has been set, but the up-scale values have not been
normalized. In this instance, we
have adjusted the device, but have not calibrated it.
Finally, there is "standardization". This is what many people are most familiar with and may
simply call calibration. In this
instance, a physical reference is used to normalize the device. The reference should be for a
specific instrument or
instrument type and must have values traceable to a standards laboratory, such as NIST.
Calibration adjustment requires an
instrument to be set to known values as indicated on the physical reference. In the case of a
reflection instrument the

15
instrument is used to measure a low density or white reference. The instrument is then adjusted
so the device displays the
value for that reference, as listed on the reference itself or on its accompanying documentation.
Likewise, the high density or
slope of the instrument is checked with a physical reference and if the instrument does not
display the appropriate value or
values, then the instrument is adjusted to display the proper value(s).
Older instruments actually required calibration-adjustment along the entire measurement range
as the electronics were not
completely linear. With the advent of higher quality components, instrument linearity is now
ensured. Those instruments that
may not be inherently linear now provide linearity adjustments to help compensate for any
differences.

16
Annex B
(informative)
Improving inter-instrument agreement
There are a few items to consider to obtain agreement between different devices of like
configuration (e.g., two 0/45
spectrophotometers).
B.1 Instrument geometry
Optical measurement instrumentation has been designed for various applications. Sphere
geometry spectrophotometers
were developed for the measurement of colorants (dyes and pigments). The use of 0/45
instrumentation is preferred when
the purpose of the measurement is to simulate how the human eye perceives light and color. This
standard specifies the use
of 0/45 instrumentation because of its applications in the graphic arts, the industrys use of
reflection densitometers (also
having 0/45 instrument geometry) and the need to have data that at least in some way can
correlate to color as perceived by
the human eye. Whichever is used, when comparing data, geometries should be the equivalent.
Refer to Annex E for further
discussion on instrument geometries.
NOTE It has been found that instrument geometries can be reversed with differences found to be negligible
or non-existent. As an
example, although the 0/45 geometry is indicated here, 45/0 geometry instruments would be deemed
equivalent in geometry.
B.2 Black backing
ISO and ANSI presently require, for most situations, that materials to be measured be backed
with a flat, black material with
a density of 1.50 0.20D (see Annex D for more detailed information on this topic). This helps
reduce the variation of the
measurement when images are printed on the reverse of the area to be measured. It also
provides a consistent backing
situation. If a sample is backed with a white backing and measured and then backed with a black
material and measured,
given the same area was measured, the difference in the backing may affect density on thin
substrates by 0.02 or more and
similar unwanted variations on color measurements will occur. Unless the samples substrate is
opaque, the sample backing
must be at the very least consistent.
B.3 Calibration
Measurement devices should each be calibrated using its own, up-to-date, calibration reference.
The reference should be
the type and kind supplied by the manufacturer of the instrument or a substitute as specified by
the manufacturer. It is good
practice to include verifying the reference certification date, ensuring the white reference
reflection values in the instrument
are correct, and then making both a white reflection reference, and a zero reflectance
measurement (made in a black trap
apparatus, or in a dark, non-reflecting location) as part of the full calibration procedure. In all
cases follow the procedures
specified by the manufacturer of the instrumentation and calibration reference. In most modern
instruments, many of the
calibration procedures have been automated to require little human intervention. It is always a
good idea to verify a
calibration once completed by measuring the calibration reference and comparing the
measurements to the target values.
B.4 Screen ruling
Areas with tone value should not be measured with an aperture that is too small to effectively
evaluate the sample. ANSI
and ISO standards have recommended minimum aperture sizes for use when measuring tone
values with various screen
rulings. See Annex F, Table F-1 for more detailed information

17
B.5 Common aperture configuration
Different apertures measure different areas of a sample. Therefore, one should always use a
common aperture size and
shape for comparison measurements or measurements from multiple locations. If you are
measuring tone values (dot area,
for instance), refer to the chart in Annex F to ensure that the aperture is large enough to assure
valid measurements at the
selected screen ruling.
B.6 Assure common measurement conditions
Some cases of measurement disagreement can be traced to the comparison of data measured
under different conditions.
For density values, different Status responses are used throughout the graphic arts and
associated industries. Status T,
Status E and Status I responses are all used within the graphic arts, and Status A is used
primarily in photo processing
applications. In addition instruments may have specialized, non-standard responses that were
developed prior to the
adoption of national standards or for customer specific applications. In the case of these
spectrophotometers, an operator
usually has a choice of options that may be set for the series of measurements that are to be
made. Such choices include
the Illuminant, Standard Observer, UV source component included or excluded, polarizing filter
selected or not, and specular
component included or excluded (this condition would be used with sphere spectrophotometers
only). Make certain the
measurement conditions of each instrument are identical or equivalent.
B.7 Optional filters
For some applications, supplemental filters are used. Polarization filters or UV cut-off filters are
not uncommon on
instruments manufactured within the past decade. Measurements from models with such filters
cannot be compared to
measurements from models without these filters. If you are unsure whether your instrument has
these filters, please contact
the manufacturer of the instrument for clarification.
B.8 Common channel
Most densitometers measure samples using all four channels (cyan, magenta, yellow and visual)
simultaneously. Likewise,
spectrophotometric data can be weighted to give densitometric data for all four filter channels for
each Status response.
These devices may be configured to display one or more of the density channels. It is important
to ensure that comparisons
are made between common channels of density measurements. Without care, one might
incorrectly compare different color
channels with one another.
B.9 Common model and manufacturer
The best agreement is achieved when devices from a single manufacturer, and better still, from
the same product family are
employed. If instruments have not already been purchased, serious consideration should be
given to the acquisition of
instruments from a single manufacturer and of a common model type.
B.10 Substrate considerations
Many instruments feature the option to measure the absolute value of a sample, or alternatively,
to measure the substrate in
addition to the sample, and then calculate the measured value relative to the substrate, also
known as the relative value.
This alternative method can help draw conclusions from measurements from two different types
of media such as proofs and
printed materials. To ensure faithful comparisons, always consider and report whether measured
values are absolute or
relative to the substrate when comparisons are made.

18
B.11 Reference considerations
Many instruments employ a feature allowing the user to enter reference data into the memory of
the instrument manually or
through a reference sample measurement process or downloaded from an interfaced computer
system. Measurements are
then made relative to the reference. This can be helpful when comparisons between different
samples are required or should
pass/fail parameters be used. Care should be taken to ensure this reference function is not
accidentally activated. Such
errors could result in measurement differences between two or more instruments.

19
Annex C
(informative)
Fluorescence in measurement
The problems of fluorescence measurement in colorimetry are not well characterized.
Unfortunately, without complex
measuring and calculation methods it is impossible to predict the sensation that will be perceived
under any real source. For
most applications this is not practical and a technique is needed to define the likelihood of
fluorescence of the sample being
measured and, preferably, some method to estimate its likely affect on the measurement results.
Florescence occurs when quanta are emitted from matter as a result of return to a steady atomic
state from an excited state
following absorption of quanta at a shorter wavelength. Thus, when certain wavelengths of
electromagnetic radiation affect
the energy state of the absorbing medium, they are then re-emitted by that medium at different
wavelengths, thereby
creating problems of measurement if the emitted wavelengths fall within the visible spectrum. It is
impossible to calculate
the radiance data (and hence tristimulus values) that will be produced by the sample, under a
particular light source, without
a detailed knowledge of the characteristics of excitation and emission of the sample (as well as
the spectral power
distribution of the viewing source). The only satisfactory method to achieve this, apart from
measuring under the viewing
source itself, is to use an instrument with two monochrometers in order to define the reflectance
spectra for each incident
wavelength band. The radiance under any real source can then be computed, providing the
spectral power distribution of the
source is also known.
Unfortunately, the complexity of this technique makes it impractical to implement. In general, the
instrument user does not
know the spectral power distribution of the real source to be used; therefore, even knowing the
details of the fluorescent
activity does not really help. The best we can do in practice is provide a method which specifies
the likelihood of the
problem, with only a limited idea of its magnitude.
Various methods are feasible for this; three are recommended here.
a) Dual source measurement: The sample should be measured with two sources one of which
approximates the spectral
power distribution of illuminant A, the other approximates that of D65. (The latter has a
significantly higher ultraviolet
emission than the former.) If the resultant spectral data is then used to compute the tristimulus
values, relative to D50, for
both sources any difference between them provides an indication of fluorescence. By quoting *
ab E . the magnitude of that
difference may be estimated. It is acceptable if the two sources are obtained by filtration of a
single source. Ideally, the
instrument manufacturer should provide spectral power distribution data that should be quoted
when communicating the
results to ensure the method may be approximated elsewhere, if required.
b) Single ultraviolet cut-off filter: A filter may be inserted between the source and sample in the
instrument that has little or no
ultraviolet transmission. By measuring the sample with and without this filter the degree of
fluorescence may be indicated.
c) An extension of method b), which provides greater information about the excitation
wavelengths (and hence the
significance of fluorescence under any real source) is to use two ultraviolet cut-off filters, one at a
time, and compute the
tristimulus values for each. Each ultraviolet filter should be inserted between the source and
sample. The two filters should
provide minimum transmittance in the wavelength bands 320-360 nm and 360-400 nm
respectively and have maximum
transmittance outside of these ranges. The manufacturer should specify these values and they
should be quoted in the
communication. The manufacturer should also specify the relative energy falling on the sample,
compared to that in the
visible region, for the source utilized, at the same wavelength bands. These may best be
presented graphically and
numerically and should also be communicated with the measurement data.
Again, the color difference may be quoted as an estimate of the magnitude of fluorescence; this
time there will be two color
differences to report, one for each filter measurement.
Both methods b) and c) presuppose that ultraviolet excitation is the problem; for most colorants
this is the case.

Any of the above methods may be used and should be referred to as CGATS.5 (Fluorescence
specification) method A, B
and C. The color difference(s) obtained should be quoted (together with the tristimulus values
obtained), and any
information pertaining to sources and filters should also be reported where available, or failing
that, the instrument type
should be reported.
NOTE It is because of fluorescence that instruments that contain lamps that provide approximately
continuous spectral
characteristics are preferred. Fluorescent lamps, with their associated line spectra, can provide
unpredictable results in such situations.
Fluorescence is an issue involving the interaction of spectral power distribution of the light source
and the material being
illuminated. It is important to note that changes in the properties of any of these items will change
the measured or viewed
results. For this reason, it is suggested that users of this standard consult with various
manufacturers and suppliers of
lighting equipment, materials, and measuring instruments about the implications of fluorescence.
Changes in the uv
component of any light sources may affect data measurements or image viewing. Changes in
material characteristics may
also yield changes in fluorescence. Because of this potential for fluorescence to impact viewing
and measurement, it is
important to evaluate both lamps in viewing booths and sources in instruments to maintain the
specified color temperature,
spectral energy distribution and overall light level as specified in ISO 3664.

21
Annex D
(informative)
Sample backing during measurement
D.1 Introduction
Many substrates used in photography, printing, and publishing are not totally opaque. Therefore,
the characteristics of the
material used behind a sample during measurement may affect the reflectance factor measured.
This in turn will change the
densities or colorimetric values computed from the spectral reflectance factor data.
If the substrate is opaque, or nearly opaque, the backing has little or no impact on the values
measured. If the substrate has
a very low opacity, the reflectance factor of the backing, particularly in areas of low ink coverage,
may dominate the
measured results. Between these extremes is an area of great uncertainty and confusion that
impacts data used for process
control, image analysis, and data exchange.
The common options for backing materials are:
- A standard black backing as defined for densitometry in ISO 5-4.
- A standard white backing as defined in 4.4.1.3 of this standard, and
- Self-backing where multiple sheets of the substrate being measured are used.
There are pluses and minuses associated with each of backing material choices.
The use of a standard backing behind the sample minimizes measurement variation, and
standard backing is especially
important in the following cases:
Job specification (e.g., printing a job to specified aims)
Industry specification (e.g., printing types of products, such as magazine advertisements, to
specified aims)
Communication of measurement values between organizations or physical locations
The choice of sample backing during measurement of the spectral reflectance of printed samples
requires a trade off
between several conflicting requirements. These are:
1. The desire to maintain compatibility with the requirements of the ISO 5-4 so that both density
and colorimetric
parameters can be obtained from a single set of spectral measurements data.
2. The need to compare colorimetric data from proofs and printed sheets which may be on
substrates having different
values for opacity of substrate (see 4.4.1.1)
3. The need to measure printed areas on substrates that are very thin, translucent, or
transparent.
D.2 Issues related to the choice of backing material
There are a number of issues relating to these requirements of which one needs to be aware.

22
Black backing has been a provision of ISO 5-4 "Photography - density measurements - Part 4:
Geometric conditions for
reflection density" at least since the 1970s. This type of backing was chosen as a result of tests
performed in the early
1960s. It was found that if a sample containing an image was placed on a white backing with the
image side down and
measurements made on the back of the paper substrate, variations in measurement, sometimes
substantial variations, could
occur. If a black backing was used, such variations of the paper substrate material were not
observed. In the printing and
graphic arts publication industry, it is very common to have a printed image on the opposite side
from the one being
evaluated.
Additionally, later tests also showed that a black backing gave less lateral diffusion error than a
white backing.
The concern in using black backing, seen by many, are that the colorimetric values obtained with
measurements made over
black backing depend on the opacity of the print substrate. As the opacity of the substrate is
reduced the measured values
of chroma and lightness are also reduced, most prominently those of the yellow solid. This is
probably why the ink makers
and the paper industry use white backing or self-backing,
While all of this is true, it needs to be said that publication products are overwhelmingly printed on
both sides and that the
back side print, or the facing page, may disturb the visual impression of the front side to a
considerable degree. Lighter
colors are reduced in chroma and lightness in some places, untouched in others. Text blocks,
heavy lettering, or fully
colored backgrounds may be visible. If the same page is viewed over a black backing, the
disturbing structures are
suppressed, the chroma and lightness reduction remains but becomes uniform over the page.
The provision of a black backing is particularly important in view of customer expectations. If the
print buyer is shown a
proof that simulates the colors of a one-sided press print on a white backing, they may be very
disappointed with the finished
product if the substrate is, say, ultra low weight paper.
There are also views that color management works better if proofer profiles are based on proofer
characterization data
measured with a white backing. The use of a black backing causes the measurements to indicate that
colors are generally darker than
they would be if measured under conditions that were more typical when images are judged visually.
However, other users prefer selfbacking
and create an opaque backing by using multiple sheets of unprinted substrate under the sample to be
measured.
While theory would indicate that a white or self-backing material should only be applicable to
those situations where the
proofing substrate and the printing substrate differ in opacity, there is practical evidence indicating
that the use of data based
on black backing may be associated with problems in profiles where the substrate is common
between proof and print. The
cause of these differences is unknown and the problem may lie with the profile building software
or color management
architecture, not the measurement data.
It has been observed that using a stack of material backing (i.e., paper) had the disadvantage of
not keeping the sample
surface flat, specifically it would assume a convex shape in the area where it was not against the
instrument aperture plate.
This problem would be most prevalent with bench-top instruments, which generally have a larger
sized aperture (8mm or
greater), but not much of a problem with smaller aperture instruments (6mm or less). In addition,
the use of self-backing
becomes an issue when exchanging either colorimetric or densitometric data between
sites/organizations because of the
difficulty in ensuring consistency of self-backing materials.
D.3 Recommendations
Where the substrate is of very low opacity, translucent, or transparent, the only logical alternative
is to use a white backing.
Using the opacity of substrate measure defined in 4.4.1.1 allows the decisions concerning
backing material to be divided into
four categories.
For opacity of substrate values above 99 the substrate is considered opaque and the backing is
irrelevant.
For opacity of substrate values between 95 and 99 a black backing is suggested.

23
For opacity of substrate values below 95 a white backing is recommended.
For opacity of substrate values between 90 and 95, and occasionally above 95, a trade-off must
be made among the
conflicting requirements described in the introduction to this annex. There may be no correct
choice of backing (black,
white, or self-backing), only application specific compromises.
One of the problems is that, as the value of the opacity of substrate becomes less, the effect of
the backing on measured
color increases. However, as ink coverage increases the opacity also increases - more for dark
colors than light colors (cyan
vs. yellow) and the effect of the backing used during measurement is lessened. In addition, the
effect of the backing used is
different when working with halftone images (reflectance is a combination of paper and solid ink)
than it is with continuous
tone images or those that vary between halftone and continuous tone image structure (gravure,
error diffusion, stochastic,
etc.).
As more is learned about the ability to use and "correct" substrate relative data, and the
correlation of the effect of backing
on the measurement of various types of images, additional options may become available.

24
Annex E
(informative)
Measurement geometry
E.1 Reflection measurement geometry
The appearance of any opaque object is largely a function of the spectral reflectance of that
object in combination with a
wide range of surface effects such as gloss, shape, texture, etc. Unfortunately, this combination
of spectral reflectance and
surface effects is difficult to characterize and measure. The basic instrumentation available to
make reflectance
measurements is limited to three configurations or geometries. These are 0/45 (or 45/0),
integrating sphere or integrating
sphere with the specular component excluded. Unfortunately, data from any one of these
configurations cannot be modified
to match data measured with another of the configurations. For many applications the
configuration chosen is not critical.
For others, specific configurations offer unique advantages and for some applications the
configuration used has its basis in
historical practice and databases rather than application requirements.
In developing a colorimetry standard for graphic arts applications one of the major goals is to
develop a measurement profile
that allows the widest interchange and applicability of data. This means that one instrument
configuration must be chosen
and some applications either will not be able to use the standard or will not be in an optimum
application situation.
In developing CGATS.5, one of the principal applications considered was the appearance of
printed images on paper.
Information available from the companies representing the participants in the working group
developing the standard
strongly indicated that the 0/45 or 45/0 geometry provides the best correlation to the reflectance
seen by a human observer
using the standard viewing conditions defined for the graphic arts. In addition, in the graphic arts,
colorimetric data is often
used in conjunction with densitometric data. The current reflection densitometry standards specify
the use of 0/45 or 45/0
geometry. The CGATS subcommittee on metrology (SC3) has therefore chosen 0/45 or 45/0
geometry as the preferred
reflectance measurement geometry for graphic arts applications and colorimetric data exchange.
One example of an important graphic arts application that may need to supplement this standard
with additional
instrumentation is the determination of "bronzing" of inks. Annex E.2 explains the phenomenon of
bronzing and the
associated measurement issues that complicate this problem.
E.2 Bronzing
One example of the use of an integrated sphere reflection spectrophotometer to evaluate surface
effects in printing is the
measurement of bronzing.
Bronzing is a phenomenon in which the appearance of a highly pigmented ink film changes
dramatically as the direction of
illumination or viewing is changed toward the specular or mirror direction. When illuminated with a
white light, the specular
reflection from a blue ink film that exhibits bronzing will have a reddish hue. With red inks, the
specular reflection will have a
yellowish hue. Bronzing is both a problem and, in some cases, a desired objective.
In printing, the source of bronzing has been commonly attributed to pigment particles in a highly
pigmented ink being pushed
up through the surface of the vehicle. The orientation of pigment particles on the printed sheet
and changes in pigment
particle size have been suggested in other explanations for bronzing. Recent work by the NPIRI
Color Measurement Task
Force [22] proposes the theory that bronzing is the result of broad-band interference in the light
reflected near the specular
angle. Not every pigment will exhibit bronzing but those that are susceptible to bronzing will show
the phenomenon both in
single pigment inks and in mixtures of inks.
In order to detect a bronze when making color measurements, the color must be observed at the
angle of specular
reflectance. Therefore, 0/45 or 45/0 geometry instruments will not detect the color of a bronzed
surface. Integrating sphere
instruments can be utilized to measure bronzed samples. Two measurements can be made: one
of the total reflectance

25
(SCI) and one with the specular reflectance excluded. The specular reflectance is then the total
reflectance (SCI) minus the
diffuse reflectance (SCE) at each wavelength. If bronzing is present, the specular component will
show a difference in the
color of the specularly reflected surface light from the color of the illuminant.
If there is no bronzing, there will be no change in chromaticity coordinates and any difference in
the magnitude of the
specular reflectance will indicate a change in gloss.
A direct measurement of the specular reflectance can be made using a goniospectrophotometer.
Light reflected at specific
wavelengths is measured as a function of the angle of illumination and viewing. Such
instrumentation is rare in the graphic
arts industry because it is expensive and normally found only in research laboratories.
What can be done if a bronze is observed and/or measured? One solution is to reformulate the
ink with a lower pigment
concentration. Another solution is to reduce the time it takes for the ink to set, which will reduce
the absorption of the ink
vehicle into the substrate. Also, the printed sheets can be coated with an over-print varnish to
minimize bronzing.
Sr = SCI - SCE (E.1)
where
SCI = specular component included;
SCE = specular component excluded;
SR = specular reflectance.
E.3 Transmission measurement geometry
Measurement geometry is defined in this standard to be 0/d or d/0. For many years such a
geometry was only defined by
reference to an integrating sphere as is currently the case in CIE 15 and older versions of ISO 5-
2. However, to more
accurately reflect the measurement geometry commonly employed for densitometry, the most
recent edition of ISO 5-2
defined ISO diffuse density. This geometry uses a diffusing medium such as opal glass or a
plastic to diffuse the incident or
emitted light. The characteristics of the diffuser are clearly specified.
For the purposes of this standard both methods of diffusing the light are considered to be
acceptable. Sphere
measurements must be of total transmittance, as defined in CIE 15. (This is the transmission
equivalent of specular
included geometry for reflectance measurements.) Measurements made using ISO diffuse
density measurement geometry
must use a diffusing material which meets the characteristics defined in ISO 5-2.
Total transmittance means, for example, that an instrument in which the sample is placed at the
entrance port of an
integrating sphere must have a diffusing material, similar to the interior of the sphere, at the point
of impingement of the
incident beam. Alternatively, the entrance port must be at a small angle to the point of
impingement and a white reflector
placed there. If neither of the above are followed, much of the light at the point of impingement
may be reflected back
through the entrance port. Thus, the entrance port acts as a light trap. Such measurements,
which are the equivalent of
specular excluded geometry for reflectance measurements, known as diffuse transmittance, are
not acceptable for this
standard.
As stressed in ISO 5-2, measurements made using the CIE and ISO methods do not produce
exactly the same results. This
is primarily due to inter-reflections occurring between the sample and the diffuser in the ISO
method (which reflects many
practical situations). However, since transmission spectrocolorimetry is not widely undertaken at
present, no de facto
standard method has evolved. Measurements are made both with sphere instruments and using
procedures in which
samples placed on diffuse illuminators are measured with spectroradiometers. Neither technique
has evolved as the
preferred method (unlike reflection spectrocolorimetry, in which 0/45 or 45/0 clearly has).
It is not the intent of this standard to impose one method or the other. However, the procedure
suggested as desirable in
ISO 5-2, in which the measurements are designated as either sphere or opal, is strongly
recommended. Experimental
evidence [13] indicates that the differences obtained by the two procedures are most significant at
high transmittances and
the error introduced is generally linear with transmittance (and CIE XYZ values). CIE XYZ values
that are measured using
an opal reference will be higher than those obtained using an integrating sphere reference. The
degree of difference is a
function of the opal material as well as the surface characteristics of the material being measured
(smoothness, matte level,
etc).

26
As a reference, measurement of Kodak Ektachrome sheet film made with an opal reference
results in CIE XYZ values that
are 1.0416 times those obtained using an integrating sphere as reference. Translation to CIELAB
values cannot be
predicted directly because of the manner in which X, Y, and Z individually contribute.

27
Annex F
(informative)
Aperture size in reflectance measurements
Target elements on print control strips are seldom larger than 5 mm square and small spot 45/0
and 0/45 geometry
spectrophotometers are available to read them. The small size of the targets and the small
aperture required to read them
requires special consideration of variations due to translucent blurring (lateral scattering) error.
When a sample is translucent, some of the illuminating light penetrates the sample and scatters
laterally to points outside of
the area viewed by the instrument detector, causing the reported reflectance values to be lower
than they would be if all the
reflected light were collected. The interaction between the translucency of the sample and optical
configuration of the
instrument is called translucent blurring, and the difference in the reflectance factor measured on
the translucent sample
compared to the corrected reflectance factor is called translucent blurring error.[14,15]
White glass reflectance standards and pressed powder pellets often used to calibrate large
aperture spectrophotometers are
generally translucent. Graphic arts proofing and printing substrates are translucent to some
degree.
To minimize translucent blurring error, large uniform samples are measured by illuminating a spot
larger than the
measurement aperture (over-illumination). ISO 5-4 requires that the "irradiated area of the
specimen shall be greater than
the sampling aperture, and its boundary shall lie at least 2 mm beyond the boundary of the
sampling aperture." Based on
the principle of optical reciprocity, equivalent measurements can be made with the viewing area
larger than the irradiated
area (over-collection).
For targets 5 mm and smaller it is not practical to over-illuminate (or over-collect) with a 2 mm
annular ring since the
measurement aperture would be 1 mm or less. For graphic arts applications it is recommended
that the boundary of the
irradiated area should be at least 0.5 mm outside the boundary of the sampling aperture.
To minimize errors due to translucent blurring (light laterally spreading within the specimen
beyond the limits of the
aperture), it is important to use opaque calibration reference materials with low internal light
scattering properties. This
eliminates one source of translucent blurring error.
Compared to white glass reference materials, targets on paper are relatively opaque and the use
of black backing further
reduces the scattering that causes translucent blurring error. Where the material being measured
is translucent, it is
desirable to have the illuminating beam significantly smaller or larger than the viewing beam.
ASTM E 1164 [16] specifies
an annular ring approximately equal to the depth of penetration of the light into the specimen.
This spreading within the
specimen may occur as the result of coarse texture, translucency, or creped construction.
Another factor that must be considered in the selection of instrument aperture when measuring
halftone images is the
relationship to screen ruling. Table F.1 shows the minimum recommended aperture size as a
function of common screen
rulings.

28
Table F.1 Minimum and recommended aperture size and sampling area
Nominal Screen Frequency Round Sampling Aperture
(mm)
Non-round Sampling Area
(mm) 1)
lines/inch
lines/cm
Minimum
Recommend
ed
Minimum
Recommend
ed
65
26
3.8
5.7
11.3
25.5
85
33
3.0
4.5
7.1
15.9
100
39
2.6
3.9
5.3
11.9
120
47
2.1
3.2
3.5
7.8
133
52
1.9
2.9
2.8
6.4
150
59
1.7
2.6
2.3
5.1
175
69
1.4
2.1
1.5
3.5
200
79
1.3
2.0
1.3
3.0
1) Area can be achieved using either a single measurement, or by averaging multiple measurements in
random or adjacent locations on the sample. This includes the use of scanning type instruments that
make continuous measurements from a sample.
NOTE These sampling aperture sizes are valid for instruments that are not moving relative to the target and
have a round aperture.
For scanning instruments or instruments with apertures that are not round, the equivalent measurement
area should be covered.

29
Annex G
(informative)
Procedures for widening the bandwidth of narrow bandpass
instruments
This standard describes procedures for tristimulus integration of spectral measurements taken
with either 10 nm or 20 nm
bandwidth instruments. The ASTM method for tristimulus integration assumes that the instrument
bandwidth and sampling
interval are approximately equal (20 nm bandwidth implies a 20 nm sampling interval). A
triangular response function with
the half-power point equal to the bandwidth is also assumed. Some laboratory instruments make
measurements with
bandwidths that are smaller than this. How does one integrate reflectance measurements
acquired at 5 nm or 2 nm
bandwidths? First, the data must be collected at intervals at least equal to the bandwidth and
preferably at intervals that are
less than the bandwidth. Then a triangular weighting function can be applied to readings to obtain
10 nm intervals and
pseudo 10 nm bandwidth data. For example, assume that the laboratory instrument can make
measurements with a 2 nm
bandwidth and 2 nm sampling interval over the spectral range of 360 nm to 780 nm (a typical
analytical laboratory
instrument). The spectral data will be at 360, 362, 364, ... 776, 778, 780 nm. A proper triangular
function for these
conditions would then be 11 points, 5 on either side of the wavelength of interest. The sum of the
weights should normalize
to unity (1.0). Since the slope of the triangle is linear, the numerators will be the integer steps
from 1 to 5, 6, 5 to 1 (1 2 3 4 5
6 5 4 3 2 1). This sums to 36 so that the weights will be in increments of (1/36).
1/36 x Rf(440) + 2/36 x Rf(442) + 3/36 x Rf(444) + 4/36 x Rf(446) + 5/36 x Rf(448) + 6/36 x
Rf(450) + 5/36 x Rf(452) + 4/36 x
Rf(454) + 3/36 x Rf(456) + 2/36 x Rf(458) + 1/36 x Rf(460)
This process would be repeated at wavelengths within the range of 360 nm to 780 nm at 10 nm
intervals. Similar
procedures would be used for other bandwidth and sampling intervals.

30
Annex H
(informative)
Calculation of colorimetric values
It is important to note that the issue of uniform visual color spaces and color difference equations
is the subject of significant
activity within CIE and related organizations. Much of this work applies to textile evaluation under
D65. However, by analogy
this work may be extrapolated to apply to printed images viewed under D50. The reader is urged
to follow the work of CIE
for current developments in both uniform visual color spaces and color difference computations.
Currently, for graphic arts
applications, CIE CIELAB uniform color space and its associated color difference computations
suffice.
H.1 CIE 1976 (L*a*b*) color space: CIELAB color space (see CIE 15)
The CIE standards for uniform color spaces and color difference estimation are currently being
revised. The equations
shown below represent the CIE agreed-upon definitions and relationships at the time of
publication of this standard. The
reader is advised to consult the CIE website (http://www.cie.co.at) for the latest publications in
these areas.
Specifically, a new color difference equation, CIE2000, is in review. CIE is also studying new
uniform color spaces that may
provide improved relationships between large and small color differences.
The following equations are therefore provided for the information and convenience of the reader.
For detailed work, the
reader is urged to consult the references shown.
L* = 116[f(Y/Yn)] - 16 (H.1)
a* = 500[f(X/Xn) - f(Y/Yn)] (H.2)
b* = 200[f(Y/Yn) - f(Z/Zn)] (H.3)
for: X/Xn > 0.008856, f(X/Xn) = (X/Xn)1/3
Y/Yn > 0.008856, f(Y/Yn) = (Y/Yn)1/3
Z/Zn > 0.008856, f(Z/Zn) = (Z/Zn)1/3
for: X/Xn o 0.008856, f(X/Xn) = 7.787(X/Xn) + (16/116)
Y/Yn o 0.008856, f(Y/Yn) = 7.787(Y/Yn) + (16/116)
Z/Zn o 0.008856, f(Z/Zn) = 7.787(Z/Zn) + (16/116)
where
Xn = 96.422, Yn = 100.000 and Zn = 82.521 for the conditions of 5.1.
( )2
1 2 2 *
ab * b * a C + = (H.4)
hab = tan-1(b*/a*) (H.5)
where
0 o hab < 90 if a* > 0
b* c 0
90 o hab < 180 if a* o 0
b* > 0
180 o hab < 270 if a* < 0
b* o 0
270 o hab < 360 if a* c 0
b* < 0

31
Bibliography
[1] ASTM E 1336: Standard test method for obtaining colorimetric data from a video display unit
by spectroradiometry
[2] ISO 2469, Paper, board and pulps Measurement of diffuse reflectance factor
[3] CIE 17.4, International Lighting Vocabulary
[4] ISO 5-1, Photography (sensitometry) Density measurements Terms, symbols and
notation
[5] CGATS.11/PIMA IT2.11, Graphic technology and photography Reflection and transmission
metrology Certified
reference materials Documentation and procedures for use, including determination of
combined standard uncertainty
[6] IT8.7/1, Graphic technology Color transmission target for input scanner calibration
[7] IT8.7/2, Graphic technology Color reflection target for input scanner calibration
[8] IT8.7/3, Graphic technology Input data for characterization of 4-color process printing
[9] ISO International Vocabulary of Basic and General Terms in Metrology
[10] D. C. Rich and D. Martin, Improved model for improving the inter-instrument agreement of
spectrocolorimeters,
Analytica Chimica Acta, 380, 263-276, (1999)
[11] Larry D. Irvine, Samuel P. Marin, and Philip W. Smith, Constrained interpolation and
smoothing, Constructive
Approximation, 2, 129-151 (1986).
[12] J. Hyman, "Accurate Monotonicity Preserving Cubic Interpolation", SIAM, 4, (4), 645-654,
(1983)
[13] David Q. McDowell. Integrating sphere vs. opal-glass as the reference for colorimetry of
transmission targets for
scanner characterization, Color Imaging: Device-Independent Color, Color Hard Copy, and
Graphic Arts II, SPIE
Proceedings, Volume 3018, 1997, pp. 181-187.
[14] Hsia, J.J., "Optical Radiation Measurements: The Translucent Blurring Effect - Method of
Evaluation and Estimation",
NBS Technical Note 594-12, Oct. 1976.
[15] Spooner, D.L., "Translucent Blurring Errors in Small Area Reflectance Spectrophotometer
and Densitometer
Measurements", TAGA Proceedings - 1991, p. 130-143.
[16] "ASTM E 1164, Practice for Obtaining Spectrophotometric Data for Object-color Evaluation,
American Society for
Testing and Materials, 100 Barr Harbor Drive, West Conshohocken, PA 19428-2959.
[17] CIE Technical Report 116, Industrial Colour-difference Evaluation, USNC/CIE c/o TLA-
Lighting Consultants, Inc., 7
Pond St., Salem, MA 01970; or CIE Publication 116 - Industrial Colour-Difference Evaluation
Central Bureau of the CIE, P.
O. Kegelgasse 27, A-1033 Vienna, Austria
[18] British Standard BS 6923:1988, Method for calculation of small colour differences
[19] ISO Standard 105-J03:1995, Textiles Tests for colour fastness Part J03: Calculation of
colour differences
[20] AATCC Test Method 173-1998.
[21] Sigg, Franz, "Errors in Measuring Halftone Dot Areas," Journal of Applied Photographic
Engineering, Vol. 9, No. 1, Feb.
1983, pp. 27-32.

35
[22] NPIRI Color Measurement Task Force NPIRI Color Measurement Task Force, A Method for
the Identification and
Assessment of the Presence of Bronzing in Printing Ink Films, TAGA Proceedings 2001, p.
444-463
[23] CIE Technical Report 132, CIE Pub 142-2001, Improvement to Industrial Colour-Difference
Evaluation, ISBN 3 901 906
088

Calibration Certification of Spectrophotometers

Hochgeladen von

Dokumentinformationen

Originalbeschreibung:

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Calibration Certification of Spectrophotometers

Hochgeladen von

Copyright:

Verfügbare Formate

CALIBRATION & CERTIFICATION OF SPECTRO PHOTOMETERS

Das könnte Ihnen auch gefallen