Introduction
Pacific Instruments calibrates and aligns each I/O module it manufactures using NIST traceable test equipment prior to delivery. All
critical performance specifications for each channel on the I/O module are verified and recorded. When the module is certified by
Pacific it meets all of its performance specifications.
Users typically perform periodic calibrations during the life cycle of any I/O module. Depending on the application and facility requirements
this can range from annual calibration and certification to frequent verification of critical parameters such as gain accuracy and
stability. Pacific provides support for both periodic and frequent calibration. The ACS2000 calibration system may be used to automatically
perform all critical performance tests channel by channel. It is the same system used by Pacific during factory I/O module calibration
and certification. For frequent calibration, typically performed before a test, PI660 software provides gain and zero calibration
and linearity verification using voltage substitution and a NIST traceable voltage standard. Any gain or zero errors detected can be automatically
eliminated during data acquisition providing typical accuracies of +/-0.02%.
There are two methods of deriving engineering unit (EU) quantities from measurements data. One requires a calibrated signal source
whereas the other does not. EU calibration using transducer centric conversion data, established by the transducer manufacturer or calibration
laboratory requires that excitation, if used, and the channel’s transfer function be accurately known. This is accomplished by
voltage substitution calibration. Alternately, end-to-end EU calibration, for example resistive or DAC shunt, applies a precisely known
stimulus to the input, measures the output and determines the conversion coefficients. This method does not require that the channel’s
transfer function be precisely known but it must be linear over the measurement range. In many applications of end-to-end calibration,
voltage substitution is used to verify linearity of the channel.
This document describes the Pacific initial instrument calibration process, the PI660 software voltage calibration process and options for
engineering unit conversion and calibration.
TOP
Part I - Initial Calibration & Alignment
Pacific Series 6000 I/O modules are calibrated and fully tested prior to being delivered. The calibration process involves aligning excitation,
amplifier gain and zero and any other circuits requiring alignment using on board digital to analog converters (DACs). This completely
eliminates potentiometers and the requirement for time consuming mechanical adjustments. Calibration involves determining and
storing the proper DAC codes in the module’s EEPROM non-volatile memory. Matching “calibration” codes are then loaded in the DACs
when a specific channel parameter is programmed.
For example, excitation voltage is typically programmed in one-volt steps from 1 to 12 Volts. For an excitation regulator controlled by a
twelve-bit DAC there are 4,096 discrete output voltages. Calibration involves determining the DAC code for each of the 12 excitation
steps that produce outputs closest to the desired voltage step and storing the value in EEPROM. Accuracy of the calibration is determined
by the resolution of the ADC.
By directly programming the DAC codes the excitation voltage can be continuously adjusted over its full range. This feature provides
fine adjustment of excitation output voltage and can be used to compensate excitation line losses in long input cables. A 12-bit DAC
for example provides approximately 2.929 mV resolution. By starting with the DAC code for the calibrated step nearest the desired excitation
output and adding or subtracting a code for each 2.929 mV difference a very accurate excitation can be achieved.
Gain step calibration and continuously variable gain employ the same principals. The primary difference is that during gain calibration a
NIST traceable voltage standard is used to provide an 80% full scale input for calibrating each gain step. DAC codes are selected and
stored for each gain step that produce outputs closest to 80% of full scale.
The factory calibration and specification verification process is computer automated and provides an HTML document file identified with
the I/O module’s serial number. The HTML document contains results for each alignment or specification tested and indicates pass or
fail for the specific test with a summary of channel pass or fail in the header of the document. It contains hyperlinks to each individual
channel’s tests and results making it easy to navigate. All test results are archived at Pacific and customers may request copies for
their records.
TOP
Part II - Voltage Calibration
Accurate voltage measurement is essential to providing quality test data. At a minimum the user must verify that voltage measurements
are made in a linear fashion. For certain types of measurements the user must also ensure that the voltages measured are accurate,
and that they are traceable to the National Institute of Standards (NIST) primary standards.
The 6000 system has the ability to measure both channel accuracy and linearity using an external voltage standard. It currently supports
the Khron Hite EDC522 and EDC523 voltage programmable voltage references that are secondary standards. The process of performing
voltage calibrations involves connecting the voltage reference to the voltage input of a channel, programming the voltage reference
to provide voltages appropriate to the input sensitivity of the channel, measuring the channel’s output voltage and determining the
slope and offset for converting voltage measured to voltage traceable. The process involves multiple measurements made over the input
sensitivity range to assure that the channel’s transfer function is linear. The process can also determine that the channel is healthy and
its gain is properly set.
Using PI660, the turnkey software for the 6000 system, the operator first selects the channels for voltage calibration. The software will
then perform the calibrations automatically. When so instructed, PI660 programs the voltage reference to provide progressively increasing
voltages to the channels, typically -80% to +80% of the channel’s full scale in 9 steps including zero. The channel’s output voltages
are measured and a gain slope and offset are calculated. The process is repeated for each channel selected.
PI660 software is designed to make the voltage calibration process quick and to verify accurate results. In order to reduce processing
time it first divides the channels selected for calibration into groups that can be calibrated in parallel. Channels with the same gain can
be calibrated in parallel because they use the same voltages levels from the reference. The process collects 512 samples from each
channel at each discrete calibration level. It then calculates the mean and standard deviation for each level and the gain slope and offset
for each channel. The standard deviation byproduct is a measure of the channel’s noise level.
TOP
Quality Checks
PI660 next verifies each channel’s performance to user programmed tolerances. Performance checks include:
- Hysteresis error for calibrations that start at one value and return to that value
- Gain and offset error
- Noise level as determined by the standard deviations
- Linearity by comparing the mean of each measured calibration level to the line defined by gain slope and offset
- Overload by comparing each of the 512 data points collected for each calibration level to a predetermined limit of full scale
- Flat line error by assuring that each successive calibration level changed by a predetermined percentage of full scale
If any of the tests failed the operator is informed of the specific failure. Pass or failure is made visible on the channel selection screen,
which is the screen from which most system operations are launched. Pass or failure is also included in the logs and calibration
reports.
When a channel passes the rigorous nine-step voltage calibration process the user is assured of the channel’s linearity and operational
health. Assuming that the voltage reference has been properly calibrated and maintained the user is also assured of the channel’s ability
to accurately measure voltages that are traceable NIST.
TOP
Linearity as an Indicator of Channel Health
Data quality checking during voltage calibration provides a means to uncover hidden problems that might otherwise go undetected and
produce unacceptable errors in the measurement data. Consider the simulated voltage calibration data shown below. Based on the
slope and offset calculated from a least squares fit of the data the operator would be led to believe that the channel is making accurate
measurements. In this case the least squares fit of the data yields a “perfect” transfer function (shown by the solid line) with a slope
of 1.0 and an offset of 0.0. The linearity test (red dots) shows that some data points have gain errors of 20% full scale or 100% of
reading! PI660 indicates this type of error condition to the user. A channel that passes the linearity test has a much higher confidence
of making accurate measurements.
Simulated Non Linear Calibration Data
Zero Determination
Voltage calibration is used to characterize the slope of the transfer function. It is not used to characterize the offset. This is due to the
fact that the input path to the measurement channel or amplifier is electrically different when connected to the calibration voltage reference
than when connected to the measured input. All amplifiers being voltage calibrated have this characteristic.
All Series 6000 signal conditioning input modules have Autozero, a feature that shorts the amplifier inputs and adjusts the module’s
digital output to zero. This feature is also used by PI660 software to determine offset for the transfer function using the transducer’s
signal path. After the hardware has performed Autozero, PI660 again shorts the amplifier inputs and takes 1000 data points. It averages
the data points and uses the result as offset in the transfer function. By doing this extra step PI660 is able to provide offset
accuracies substantially better than specified for the signal conditioning hardware.
Calibrating for Line Effects
The user may go one step further and characterize offset back to an external transducer connector. This is a one time characterization
of the small differences between automatic zero shorted input and transducer connector short input generated by non-zero source currents
and Seebeck effect voltages in the transducer signal path.
Zero plug calibration as it is known consist of installing a signal shorting plug at the transducer connector. PI660 then compares the
average of 1000 “zero” data points from the external shorting plug to the average of 1000 “zero” data points using the autozero input
shorting mechanism. The difference is stored in the channel’s non-volatile calibration EEPROM. From that point forward the difference
between zero plug and the internal zero will be factored in whenever PI660 performs an autozero and calculates offset in the transfer
function.
Voltage Calibration Summary
In practice the techniques discussed here for voltage calibration have been shown to deliver total measurement accuracies of 0.02% for
gains up to 1,000 and 0.05% at the highest gain of 5000. The zero accuracy enhancement is highest at the highest gains since small
offset differences are more easily quantified.
TOP
Part III - Engineering Unit Calibration & Conversion
Engineering Units (EU) calibration and conversion takes many different forms. The crux of EU calibration and conversion is determining
a suitable means to convert a transducer’s voltage or current measurement to a real-world physical quantity. A linear equation is the
most often used method of performing this conversion. Non-linear equations and look up tables are other methods. It should be noted
that EU calibration is applied only to data displays and data exported in ASCII format. Data files recorded in binary format consist of
raw measurement data with the EU conversion information contained in a “stamp” appended to the data file.
PI660 coupled with System 6000’s built-in calibration services provide the means to perform most EU conversions and calibrations.
Series 6000 signal conditioning modules have EU calibration stimuli, and PI660 can programmatically invoke these stimuli. The stimuli
available depend on the module type. For instance, modules for strain gages or bridge transducers provide a known EU output by shunting
of the bridge circuit using one or more precision resistors. The process of EU calibration involves modifying the transducer’s output
by applying some known stimuli to the transducer, recording the output voltages of the transducer for each stimulus applied, and creating
a voltage to EU conversion equation based on the results.
In many cases transducers are provided to the user with calibration data sheets. The calibration data sheets contain information that
the transducer manufacturer or calibration laboratory determined about the transducer’s output characteristics when the transducer was
subjected to known stimuli. Some transducer data sheets provide the voltage measured to EU value conversion equations for specific
transducer excitation levels. Others provide lookup tables that describe transducer data conversion for nonlinear types of transducers or
prescribe shunt resistor values that can be applied to the transducer at specified excitation to provide specific EU value simulations.
PI660 handles all of these types.
Conversion Equation
With a known conversion equation the conversion equation coefficients can be directly entered into PI660 or downloaded from PI660’s
transducer database or a commercial database such as Microsoft Access. PI660 will then convert the voltages it measures for the channel
to the appropriate EU values by using the supplied conversion equation.
PI660 supports user provided lookup tables that specify EU verses transducer output voltage. Simply provide the lookup table entries to
PI660 and apply the lookup table to channels. PI660 converts the measured voltages to EU values using straight line fits to the data
where the equation for the straight line comes from interpolating or extrapolating the lookup table. NIST lookup tables for Types B, C,
E, J, K, N, R, S, and T thermocouples are included in PI660.
Shunt calibration is usually performed using shunt resistor values that have been determined to provide a specified EU output. In this
case the operator selects the shunt resistor to invoke and the corresponding EU levels for the shunt. Multiple shunt resistor values may
be used and PI660 will generate a conversion equation that best fits the corresponding EU levels.
Some transducers are calibrated by applying known EU stimuli, such as pressure from a stable, controllable plenum. In this case the
number of steps and their corresponding EU values are entered into PI660. During calibration PI660 pauses for the operator to apply
the stimulus and calculates a conversion equation that best fits the specified EU values for each step. This type of calibration is frequently
used for position measuring potentiometers and LVDTs also.
EU calibrations can be performed with or without operator interaction. Calibrations that require operator interaction are those that
require manual adjustment of the stimulus to the channels being calibrated. An example of this would be extending a string pot to a
known distance prior to collecting the data for the EU calibration step. Calibrations that do not require user interaction can proceed as
quickly as data can be sampled and the stimuli invoked. An example of this would be shunt calibration of a bridge type transducer.
For certain measurements it is desirable to remove an offset prior to acquiring test data. Such offsets may be the at-rest or quiescent
output of the transducer and their removal is generally known as tare removal. PI660 has a tool that for tare removal. The operator
specifies that the value measured for a channel or group of channels should be equated to a specific engineering unit value. This
causes PI660 to change the offset coefficient of the engineering unit conversion equations for the channels selected for tare calibration.
Quality Checks
PI660’s EU calibration software is very thorough. It mimics the data checking described in the previous Voltage Calibration section.
PI660 captures the EU calibration step data to files. The files contain the full 512 data points acquired for each step of the calibration
process. This enables the operator to graph for study and to assure proper channel operation. PI660 further allows substitution of a
calibration step or the full calibration with previously acquired calibration data from a different test.
EU Calibration Summary
Series 6000 signal conditioning modules provide the capability to EU calibrate many diverse types of transducers using conventional and
less conventional techniques. PI660 automates the EU calibration process and provides additional calibration techniques such as lookup
tables and external stimuli. It provides verification to assure the acquired calibration data is of highest quality.
TOP
Conclusion
The Series 6000 Data Acquisition System with PI660 software provide calibration techniques that result in the highest quality conversion
of voltage measurements to engineering unit quantities. The user can select which calibration techniques to employ. PI660 software
can enhance the basic system performance specifications by performing NIST traceable channel calibrations. It uses the NIST
traceable channel output, if available, as the input to the EU conversion equation or lookup table. The EU conversion equation can be
derived by performing an interactive or non-interactive EU calibration or it can be manually entered by the user. User specific EU conversion
lookup tables are supported as are the NIST thermocouple tables.
TOP
Contact
For additional information about calibration for highest accuracy measurements using PI660 Data Acquisition & Display Software, please feel free to contact us at:
Phone: 925-827-9010
Email: salesrequest@pacificinstruments.com
