Understanding the

uncertainties associated with

the use of Metrology Wells
Application Note

Introduction
trolled temperature along a portion
Dry-well calibrators are stable heat
of its length. The vertical gradient
sources used in process and labora-
in the insert is termed “axial unifor-
tory environments for calibration of
mity.” Due to dissimilarities in con-
temperature sensors. All heat sources
struction and length of temperature
introduce measurement errors as
sensing elements, one must consider
a result of their mechanical design
the axial uniformity of Metrology
and thermodynamic properties.
Wells. According to EA (European
These effects can be quantified in an
Accreditation) guideline 10/13, dry-
effort to determine the heat sources’
wells should have a “zone of suffi-
contribution to the measurement
cient temperature homogeneity of 40
uncertainty. Fluke Calibration has
mm” from the bottom of the well. We
developed Metrology Wells to reduce
recommend a zone of at least 60 mm
the errors typically seen in the usage
to cover sensor lengths of the units
of dry-wells. Metrology Wells come
under test (UUTs) and the reference
with a calibrated control sensor and
standard, which often needs at least
have an option for a built-in ther-
50 mm.
mometer readout. The Metrology
Well’s uncertainty will significantly
vary depending on mode of use. The
Application Tip: Axial uniformity
uncertainties associated with each errors can often be improved
method of use are discussed. The beyond the specification by align-
purpose of this application note is ing the centers of the sensing
to help technicians and metrologists element of the reference probe
understand and quantify the mea- and the UUT (see Figure 4).
surement uncertainty when using
Metrology Wells.
Radial uniformity
Uncertainties associated with The thermal gradient from one well
the use of Metrology Well to another is referred to as radial
and its built-in reference uniformity. Measurement errors
thermometer input caused from imperfect radial unifor-
Best performance is usually real- mity are attributed to the distance
ized when Metrology Wells are between wells and heaters, the ther-
used as stable heat sources and an mal properties of the insert material,
external reference thermometer or and effects from uneven heat distri-
the optional built-in reference ther- bution caused by non symmetrical
mometer input is used as the refer- loading.
ence standard (see examples 1 and
2 on page 4). Typically, the major
Application Tip: Best results
sources of uncertainty are caused by
are found when using a compari-
imperfect axial and radial uniformity, son insert (Insert B, D, or E) with
loading, instability, stem conduction, a reference probe of the same
reference thermometer accuracy, and diameter as the UUT and mea-
unit under test characteristics. suring directly across from the Figure 1. The Metrology Well display can help determine when
UUT. stability is reached, but a better practice is to rely on display of the
Axial uniformity reference that offers more resolution.
Each Metrology Well insert (remov-
able sleeve with several drilled Loading effect
wells) is exposed to ambient envi- The number of probes inserted into a
ronment at the top end and to a con-
Metrology Well impacts the amount

From the Fluke Calibration Digital Library @ www.flukecal.com/library

 Application Tip: Stem conduc-
HOT tion error can be determined by
immersing the probe into a bath
AMBIENTT YPICAL and noting its change in temper-
ature when raised at set incre-
HEAT
ments. This is the best practice
in that each probe has unique
characteristics that contribute to
stem conduction and should be
evaluated individually.

reference readout, UUTs, and their

readouts. These errors include refer-
ence probe and thermometer readout
Figure 2. Stem conduction calibration uncertainty or accuracy,
of heat drawn away from or into a reference probe drift and hysteresis,
Metrology Well. This is referred to as reference probe self-heating, UUT
“loading effect,” which can be mea- short-term drift and hysteresis, and
sured by inserting probes into wells UUT readout accuracy. This paper
and noting the change in the refer- does not go into detail on these
ence reading. Metrology Well design contributors, but more information
characteristics, such as 203 mm (8 can be obtained by contacting Fluke
in) of immersion and dual-zone con- Calibration at www.flukecal.com.
trol, help minimize the loading error.
Uncertainties associated with
d
Stability the use of a Metrology Well
Stability over time affects calibra- and its calibrated control
tions. EA 10/13 defines stability sensor
as a temperature variation “over a
Metrology Well control sensors come
30-minute period.” The Metrology
calibrated with a traceable certificate
Well display can help determine
of calibration and an accuracy state-
when stability is reached, but there
ment which allows the display to be
can be a significant difference
used as a reference standard. Many
between the actual and indicated
measurement errors contribute to the
20d temperature during stabilization.
uncertainty when using the Metrol-
The better practice is to rely on the
ogy Well in this manner.
display of the thermometer readout,
which offers more resolution (see
Axial uniformity
Figure 1).
The vertical gradients in Metrol-
ogy Wells are more apparent when
Stem conduction error
using the calibrated control sensor as
Stem conduction is heat flux along
the reference than when using the
the length of the thermometer stem.
external reference thermometer. This
This affects both the reference ther-
is because it’s not always practical
mometer and the UUT (see Figure
to align the sensing elements of the
2). Recommended depth of immer-
control sensor (which is fixed in the
sion for a probe is calculated as fol-
block) to the UUTs (see Figures 4
lows: [20 x probe OD] + [length of
and 5).
the sensor] (see Figure 3). Because
Metrology Wells have deep immer-
sion (160 mm [6.3 in] to 203 mm [8
Radial uniformity
The calibrated control sensor is fixed
in]), the error from stem conduction
Figure 3. Recommended depth for probe in the block and is often not equi-
immersion is usually a very small contributor to
distant to the UUTs from the heaters.
the over all uncertainty.
Thus, radial uniformity is a contribut-
ing factor and is considered.
Reference probe, thermometer
readout, and UUT considerations
Loading effect
There are many other sources of
Errors from loading typically are
uncertainty to consider that result
from the use of a reference probe,
much larger when using the con-
trol sensor as the reference stan-

2 Fluke Calibration Understanding the uncertainties asscociated with the use of Metrology Wells
dard as opposed to an external
reference probe, because the con-
trol sensor is isolated in the block
and does not compensate for the
loading of the wells in the insert.

Short-term and long-term drift

Each Metrology Well control sensor
has a short- and long-term drift
associated with its use. Drift will
vary depending on use and care
of the Metrology Well. Sensor drift
can be determined by regular cali-
bration or intermediate checks at a
fixed temperature.

Hysteresis
Hysteresis is the difference in a
Metrology Well’s actual tempera-
ture resulting from the direction
from which that temperature was CENTER LINE OF SENSORS

approached. It is greatest at the

mid-point of a Metrology Well’s
range.

Control senor calibration

Figure 4. Aligning the centers of the sensing element of the reference
Metrology Well control sensors are
probe and the UUT
calibrated and come with NIST-
traceable reports of calibration.
The accuracy of the control sensor
may vary from ±0.1 °C to ±0.25
°C. The calibrated display becomes
most useful when it is used in the
same way it was calibrated— when
the 6.35 mm (0.25 in) hole is
loaded with the UUT and readings
are compared against the display
temperature.

Other considerations
The errors from short-term drift,
hysteresis, and readout accuracy
of the UUT apply in the same way
as when a Metrology Well is used
with an external reference.
CENTER

CONTROL
SENSOR CENTER

Figure 5. The vertical gradients in Metrology Wells are more apparent

when using the calibrated control sensor as the reference.

3 Fluke Calibration Understanding the uncertainties asscociated with the use of Metrology Wells
Example 1 - Uncertainty Table 1.
calculation associated with Specification (mK) Probability Uncertainty (mK)
Metrology Wells using a Distribution
reference probe and a ther- Axial uniformity 20 Rectangular 11.55
mometer readout as the
standard. Radial uniformity 10 Rectangular 5.77

The example in Table 1 illustrates Loading effect 5 Rectangular 2.89

the calibration measurement Stability 5 Rectangular 2.89
uncertainty contributed by Metrol- Reference probe 4 Rectangular 2.31
ogy Wells. The actual values will calibration
vary with application and tem-
Reference probe drift, 3 Rectangular 1.73
perature. This example does not including hysteresis
consider sources of uncertainty
Reference probe stem 2 Rectangular 1.15
originating from characteristics
conduction
of the UUT. This example is at 0
°C. Equipment considered for this Thermometer readout 6 Rectangular 3.46
example consists of the following: accuracy
•• Model 9171-B-R, Metrology Well Combined Standard Uncertainty (RSS) 14.32
with built-in reference input
•• Model 5626-15-D, High-tempera- Expanded Uncertainty (k=2) 28.64
ture, Secondary PRT, 100 ohm
Table 2.
Example 2 - Uncertainty Specification (mK) Probability Uncertainty (mK)
calculation associated with Distribution
Metrology Wells using the Axial uniformity 20 Rectangular 11.55
calibrated display as the ref-
erence standard Radial uniformity 10 Rectangular 5.77

The example in Table 2 illustrates Loading effect 5 Rectangular 2.89

the calibration measurement Stability 5 Rectangular 2.89
uncertainty contributed by the Hysteresis 25 Rectangular 14.43
Metrology Well. The actual values
Control sensor accuracy 100 Rectangular 57.74
will vary with application and
temperature. This example does Long-term drift 100 Rectangular 57.74
not consider sources of uncertainty Combined Standard Uncertainty (RSS) 84.06
originating from characteristics
of the UUT. This example is at 0 Expanded Uncertainty (k=2) 168.13
°C. Equipment considered for this
example consists of the following:
•• Model 9171-B, Metrology Well
with calibrated control sensor
used as the reference standard.

Note pertaining to Examples 1 and 2:

Standard Uncertainty calculated by
dividing the stated specification by
the 3 to obtain a “standard deviation
like” quantity based on rectangu-
lar distribution. Combined standard
uncertainty uses a root sum square
method assuming that individual
standard uncertainties behave inde-
pendently.

Figure 6. Total uncertainty is significantly reduced when using an

external reference thermometer as the reference standard as shown
above.

4 Fluke Calibration Understanding the uncertainties asscociated with the use of Metrology Wells
Fluke Calibration. Precision, performance, confidence.™

Fluke Calibration Fluke Europe B.V.

PO Box 9090, PO Box 1186, 5602 BD
Everett, WA 98206 U.S.A. Eindhoven, The Netherlands
For more information call:
In the U.S.A. (877) 355-3225 or Fax (425) 446-5116
In Europe/M-East/Africa +31 (0) 40 2675 200 or Fax +31 (0) 40 2675 222
In Canada (800)-36-FLUKE or Fax (905) 890-6866
From other countries +1 (425) 446-5500 or Fax +1 (425) 446-5116
Web access: http://www.flukecal.com
©2006-2012 Fluke Calibration. Specifications subject to change without notice.
Printed in U.S.A. 9/2012 2510068B_EN

Modification of this document is not permitted without written permission

from Fluke Calibration.