SAND2017-2168C

TRACEABLE HIGH-FREQUENCY VIBRATION CALIBRATION OF ACCELEROMETERS IN THE

TEMPERATURE RANGE OF -60C TO 90C

Randy Mendoza
Primary Standards Laboratory
Sandia National Laboratories

Michael Mende
SPEKTRA Schwingungstechnik und Akustik GmbH Dresden

DISCLAIMER

Certain commercial equipment, instruments, or materials are identified in this paper in order to adequately describe
the experimental procedure. Such identification does not imply recommendation or endorsement by the authors or
Sandia National Laboratories, nor does it imply that the materials or equipment identified are the only or best
available for the purpose.

Sandia National Laboratories is a multiprogram laboratory managed and operated by Sandia Corporation, a wholly
owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy’s National Nuclear Security
Administration under Contract No. DE-AC04-94AL85000. This publication has been approved for unlimited public
release, SAND20XX-XXXX X.

ABSTRACT

The determination of accelerometer response to temperature variation is well understood in terms of accelerometer
design. However, traceable calibration of accelerometers under extreme temperature conditions, particularly at
higher frequencies, is a unique challenge. The Primary Standards Laboratory at Sandia National Laboratories, in
conjunction with SPEKTRA, has recently implemented a sinusoidal vibration calibration system for the traceable
calibration of accelerometers from 10 Hz to 10 kHz in the temperature range of -60C to 90C. This paper includes
the following: a brief overview of vibration calibration theory, a description of the calibration system and its
configuration, an approach for determining the temperature coefficient of a back-to-back reference standard
accelerometer, and a summary of improvements on existing capabilities and measurement uncertainties.

KEYWORDS

Accelerometer, Vibration, Calibration, Temperature, High Frequency, Shaker.

INTRODUCTION

The Flow, Acceleration, Shock and Humidity (FASH) project in the Primary Standards Laboratory (PSL) at Sandia
National Laboratories (SNL) calibrates accelerometers for vibration in the frequency range of 10 Hz to 10 kHz in
support of line organizations within SNL, as well as calibration facilities throughout the Nuclear Security Enterprise
(NSE). As program testing requirements have evolved, the FASH lab has been tasked with providing vibration
calibration data for accelerometers, not only under ambient conditions, but at extreme temperatures, as well. Initial
attempts at achieving such a capability were crude, and perhaps most importantly, not traceable to any meaningful
standard. The subject of this paper represents a step forward in terms of the quality of the vibration calibration data
at extreme temperature being produced by the FASH lab, as well a unique means to establishing traceability of these
measurements.

VIBRATION CALIBRATION METHODOLOGY

Secondary vibration calibration of accelerometers uses electrodynamic shakers to sinusoidally oscillate a device
under test (DUT) accelerometer mounted in a back-to-back configuration to a traceable reference (REF)
accelerometer (see Figure 1), in accordance with ISO 16063-21, “Methods for the calibration of vibration and shock
transducers – Part 21: Vibration calibration by comparison to a reference transducer”[1].

1
 Figure 1: DUT and REF in back-to-back configuration

The shaker ramps up to a desired frequency and acceleration level (i.e. 100 Hz, 10g) and the sensitivity of the DUT
accelerometer is measured. Accelerometer sensitivity is expressed as a ratio of the electrical output of the
accelerometer (charge or voltage) to the acceleration input to the system. Thus, piezoelectric accelerometer
sensitivity will be reported in units of pC/g, and piezoelectric accelerometers with integrated electronics to facilitate
on-board charge conversion will have sensitivities reported in mV/g. The calibration continues by sweeping through
a range of frequencies and determining the response curve of the DUT accelerometer, with the sensitivity at each
individual frequency plotted in terms of its deviation to the sensitivity at a reference frequency, usually 100 Hz. See
Figure 2.

Figure 2: Typical Frequency Response Curve

Per ISO 16063-21, this calibration must be conducted in accordance with the following environmental conditions:

Room Temperature 23±5°C

Relative Humidity 75% max

2
These ambient conditions play an important role in the calibration, as the traceable REF accelerometer itself will
have been calibrated at these conditions. Any deviations outside of the listed tolerances will affect the performance
of the REF accelerometer and could invalidate its traceability.

EXISTING VIBRATION CALIBRATION SYSTEM

The main components of the current vibration calibration system in the FASH lab are the following:

 Unholtz-Dickie electrodynamic vibration shaker

 MB Dynamics 405 signal conditioner with Win475 calibration control software
 National Instruments AT-MIO-16E ISA Data Acquisition board
 Delta Design temperature chamber
 Kistler 8002K piezoelectric REF accelerometer

The REF and DUT accelerometers are affixed to the shaker armature using a small, ceramic mounting fixture. The
REF and DUT devices mount to the bottom and top, respectively, of the mount via ¼-28 to 10-32 threaded adapters
and are torqued to 18 in-lb. The mount itself is secured to the armature using 4, 4-40 × 2” steel screws torqued to 8
in-lb. See Figures 3 and 4, below.

Figure 3: REF and DUT Attached to Mounting Fixture Figure 4: REF and DUT Assembly Mounted to Shaker

The shaker specifications are as follows:

 Frequency range: 10 Hz to 10 kHz

 Max load: 23 kilograms
 Max stroke: 25mm (pk-pk)
 Max acceleration: 50g (pk)
 Transverse motion specifications:
o 5% (10 Hz – 4 kHz)
o 25% (4 kHz – 10 kHz)

The measurement uncertainty for this system is ±4% @ k=2 for calibrations at ambient temperature.

VIBRATION CALIBRATION AT EXTREME TEMPERATURES

When accelerometers are used in applications where large thermal variations are a component of the testing
requirements, it is necessary to determine the accelerometer response to those shifting environmental conditions.
Performing a back-to-back vibration calibration at temperatures outside the ambient range has proven to be a
challenge, as it is necessary to effectively limit the exposure of the REF accelerometer to any meaningful thermal
variation outside of the ISO 16063-21 environmental specifications listed in the previous section, while still
exposing the DUT accelerometer to the specified extreme hot or cold test temperature of interest. Until now, the
only viable methodology of maintaining the ambient environmental conditions surrounding the REF accelerometer

3
has been to physically remove it from the chamber space by mounting it under an extension mount. This mount then
projects into the temperature chamber, where the DUT accelerometer would be mounted. See Figures 5 and 6,
below.

Figure 5: REF and DUT Attached to Extension Fixture Figure 6: Extension Assembly Attached to Shaker

The system was retrofitted with a Delta Design temperature chamber to provide the extreme hot and cold conditions
necessary for the calibration. A six-inch diameter access port was cut in the bottom of the chamber floor, and the
entire chamber sat on a platform that allowed it to be positioned with the access port directly over the shaker
armature. See Figure 7.

Figure 7: Temperature Chamber Positioned Over Shaker Armature

This methodology, however, comes with considerable drawbacks. First, the degree to which the REF accelerometer
is isolated from any thermal variation was found lacking. A series of tests was conducted to determine the operating
temperature inside the armature where the REF accelerometer was mounted during testing. During these tests, a

4
thermocouple was mounted as near as possible to the REF accelerometer inside the shaker armature (while the
shaker was stationary) and the chamber was cycled through its full range of temperatures. The results are
summarized in Table 1 below.

Chamber Temp (C) REF Accel Actual Temp (C)  from Ambient (C)
22.5 23.9 1.4
71.0 42.6 20.1
-54.0 -14.2 -36.7

Table 1: REF Accelerometer Temperature

At the extreme end of the test temperature range, the environment in which the REF accelerometer operates deviates
from the ambient by as much as -36.7C, altering its charge output when vibrated and thus invalidating its
traceability to the SI. In addition to insufficient temperature isolation, use of the extension mount had dynamic
consequences. Using a single axis accelerometer as a DUT (attached to the mounting fixture via a ¼-28 to 10-32
threaded adapter and torqued down to 18 in-lb.), a series of tests was conducted to determine the degree to which the
extension mount affected the frequency response of the DUT. First, the accelerometer was tested at ambient
temperature from 10 Hz to 10 kHz while mounted to the standard fixture (Figure 4 above) and its frequency
response recorded. This test was then repeated using the extension mount (as shown in Figure 6). See results in
Figure 8, below.

DUT Frequency Response--Standand vs. Extension Mount

+22.5 °C Standard Mount +22.5 °C Extension Mount
17
% Dev from Ref Freq. (100 Hz)

12

-3

-8

-13
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
Frequency (Hz)

Figure 8: DUT Frequency Response Mounting Fixture Comparison

The frequency response of the DUT accelerometer when mounted to the extension fixture tracks that of its response
when mounted to the standard fixture up to roughly 2 kHz. Beyond that, the deviation in the responses suggests that
the extension fixture leaves the DUT more susceptible to transverse motion errors in the higher frequency range.
The mounting positions of the REF and DUT with the extension fixture are rather far apart, calling into question the
degree to which the extension fixture adheres to the requirements for a back-to-back calibration methodology set
forth in ISO 16063-21. Finally, the way the mounting fixtures attach to the shaker armature make it impossible to
position a temperature sensor close to either the REF or DUT accelerometers while the shaker is in motion. The
thermocouple is mounted to the chamber floor, in the recessed area just above the DUT accelerometer. As a result,
the best temperature measurement possible is actually of the air circulating around the DUT accelerometer, as
opposed to the actual case temperature of the accelerometer itself. These issues are significant enough to call into
question the suitability of the system to perform reliable, traceable vibration calibrations over the desired

5
temperature and frequency ranges. With this in mind, it was decided that a new system would be designed to
conduct these types of calibrations. Such a system would have to include the following features:

 Traceable frequency range: 10 Hz to 10 kHz

 Temperature range: -60C to 90C
 Isolation of REF accelerometer from thermal variation
 True back-to-back calibration capability
 Ability to mount temperature sensor directly to accelerometer in the test chamber

TEMPERATURE VIBRATION CALIBRATION SYSTEM DESIGN OVERVIEW

The following subsections detail the key characteristics of the Temperature Vibration Calibration System. The
system is based on a SPEKTRA CS18 MF calibration system with the following components and specifications:
 SPEKTRA SE-10 electrodynamic shaker
o Frequency range: 10 Hz to 10 kHz
o Maximum load: 200 grams
o Maximum displacement: 8mm (pk-pk)
o Maximum acceleration: 35g (pk)
 SPEKTRA Type PA14-180 power amplifier to drive the shaker
 SPEKTRA SRS 35 Vibration Control Unit (VCU) for signal synthesis, acquisition, conditioning and
processing
 Votsch VT 7004 temperature test chamber
 Votsch VT7004-S Type K Thermocouple
 PCB 353B17 Internal REF Accelerometer
 Endevco 2270 Back-to-Back REF Accelerometer
 Dell PC running Windows 7 Professional
The system is configured as follows: Instead of positioning a temperature chamber over a large shaker armature, the
much smaller and more efficient SE-10 shaker is positioned so that it projects into an access port in the bottom of
the temperature chamber. See Figure 9.

Figure 9: Rendering of SE-10 Shaker Within Temperature Chamber

6
The 353B17 internal REF accelerometer is mounted within the shaker armature and torqued to 18 in-lb. Attached to
the mounting surface of the armature is a ceramic thermal isolator that is tapped with a ¼-28 through hole, which
projects into the temperature chamber. A rubber thermal membrane attached to the floor of the chamber is inserted
into a groove on the ceramic isolator. Both the ceramic isolator and the rubber membrane aid in preventing any
thermal transfer from inside the chamber into the armature where the 353B17 is mounted. To this end, there is also a
fan mounted on the underside of the shaker to help regulate the temperature of the internal REF accelerometer. The
Endevco 2270 is mounted to the ceramic isolator and torqued to 18 in-lb. See Figure 10.

Figure 10: Cross-Section of SE-10 Shaker

7
 PRINCIPLE OF OPERATION

The key insight to achieving truly traceable secondary vibration calibrations at extreme temperatures is that the REF
accelerometer must be placed inside the temperature chamber with the DUT, necessitating the addition of the
Endevco 2270 back-to-back REF accelerometer to the calibration scheme. This amounts, effectively, to conducting
two transfer calibrations. The unbroken chain of traceability consists of the following steps:
1. Primary calibration of the PCB 353B17 internal REF accelerometer (ISO 16063-11)
a. Determination of temperature coefficient
2. Endevco 2270 back-to-back REF accelerometer primary calibration (ISO 16063-11)
3. Endevco 2270 back-to-back REF accelerometer mounted to shaker (via thermal isolation mount) inside
temperature chamber
a. Endevco 2270 back-to-back REF accelerometer undergoes secondary calibration using PCB 353B17
accelerometer as REF in temperature range of -60C to 90C
b. Temperature coefficient determined
4. DUT accelerometers mounted directly back-to-back with Endevco 2270 REF and calibrated in the range of -
60C to 90C
These steps will be discussed in more detail below.

PCB INTERNAL REFERENCE ACCELEROMETER CALIBRATION

The PCB 353B17 internal REF that is mounted on the underside of the shaker armature receives a primary vibration
calibration using a laser doppler vibrometer as a reference measurement, in accordance with ISO 16063-11,
“Methods for the calibration of vibration and shock transducers – Part 11: Primary vibration calibration by laser
interferometry”[2]. This primary calibration is directly traceable to the German National Metrological Institute,
Physikalisch-Technische Bundesanstalt (PTB). See primary calibration data in Figure 11.

PCB 353B17 Primary Calibration Data

13
% Dev from Ref Freq. (100 Hz)
8
3
-2
-7
-12

0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
Frequency (Hz)

Figure 11: PCB 353B17 Primary Vibration Calibration Data

The degree to which this internal REF accelerometer is exposed to the thermal variation in the chamber above must
be limited. This is achieved, for the most part, through the ceramic thermal isolator, the rubber thermal membrane,
and the fan mounted to the bottom side of the shaker. When a temperature sensor was placed in the armature next to
the internal REF and the chamber was cycled through the temperature range of -60C to 90C, it was determined
that the temperature variation inside the armature was 10C to 35C. A test was conducted to determine the
temperature response of the PCB 353B17 internal REF accelerometer. The accelerometer was removed from the

8
shaker and placed into a Peltier chamber. The Peltier chamber (with the accelerometer mounted within it) was then
mounted to a shaker and, using a laser doppler vibrometer as a reference, the temperature response of the
accelerometer was measured in the limited frequency range of 3Hz to 400 Hz and temperature range of 0C to 60C
[4]. See Figure 12.

Figure 12: Peltier Chamber with PCB 353B17 REF Accelerometer

Figure 13: Temperature Response of PCB 353B17 REF Accelerometer

The solid line in Figure 13[4] plots the temperature response of the PCB 353B17 REF accelerometer compared to its
calculated response according to the manufacturer’s specification (dashed line). Using this data, a temperature
coefficient was determined and applied to the primary calibration data. The PCB 353B17 was then mounted back in
the armature of the shaker, where it was used as a reference to perform a secondary calibration on the Endevco 2270
back-to-back REF accelerometer in the chamber.

9
 ENDEVCO 2270 BACK-TO-BACK REF ACCELEROMETER CALIBRATION

The Endevco 2270 also receives a primary calibration at room temperature in accordance with ISO 16063-11. This
primary calibration forms the basis for establishing its traceability. See Figure 14. (Note: The Endevco 2270 is
calibrated with no mass attached, as well as with a 20g mass mounted to it in order to determine mass loading
effects on its frequency response.)

Endevco 2270 Primary Calibration Data

Endevco 2270 Primary Calibration Data Endevco 2270 Primary Cal Data - 20g Mass
% Dev from Ref Freq. (100 Hz)
8
-2
-12

0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
Frequency (Hz)

Figure 14: Endevco 2270 Primary Vibration Calibration Data

The Endevco 2270 is then mounted within the thermal chamber (Figure 15) and receives a secondary calibration
using the 353B17 accelerometer mounted on the underside of the armature as reference (ISO 16063-21) throughout
the temperature range of -60C to 90C.

Figure 15: Endevco 2270 Mounted Within Thermal Chamber

This temperature response data allows for the determination of a temperature coefficient that can then be applied to
the primary data. See Figures 16[4] and 17[4], below.

10
 Figure 16: Endevco 2270 Frequency Response at Extreme Temperatures

Figure 17: Combined Endevco 2270 Temperature Response

The solid line in Figure 17 plots the temperature response of the Endevco 2270 accelerometer compared to its
calculated response according to the manufacturer’s specification (dashed line). Using this data, a temperature
coefficient was determined and applied to the primary calibration data. The reference transducer’s traceability is
established and its temperature and frequency response fully characterized. DUT accelerometers are mounted

11
directly, back-to-back, with the 2270 REF accelerometer (Figure 18) and receive a secondary calibration (ISO
16063-21).

Figure 18: DUT Accelerometer Mounted Back-to-Back With Endevco 2270 REF

TEMPERATURE CHAMBER CONTROL THERMOCOUPLE

The temperature chamber is controlled with a Type-K thermocouple that is attached directly to the 2270 REF
accelerometer. This allows for direct measurement of the accelerometer case temperature, as opposed to the air that
is circulating throughout the larger chamber space. See Figure 19.

Figure 19: Type-K Control Thermocouple Mounted Directly to Endevco 2270 REF

MEASUREMENT UNCERTAINTY

Each uncertainty component was quantified by either of the following methods: statistical methods, which are
termed type A uncertainties, or any other method (including calibration data), which are termed type B uncertainties.
All uncertainty components were converted to standard uncertainties, representing 1 standard deviation (k=1),
before combining. The uncertainty contributors can be divided into the following classifications: uncertainty
contributors from the instrumentation used as reference standards and uncertainty contributors unique to the Units
Under Test (UUTs).

12
1. Uncertainty contributors that arise from the primary calibration and temperature response of the PCB
353B17 REF accelerometer (Table 2).

Total Combined Uncertainties for PCB 353B17 REF Accelerometer

Standard Uncertainty %
Source of Uncertainty Distribution
(k=1)
Primary Vibration Calibration normal 0.250
Long-Term Sensitivity Stability (Drift) rectangular 0.577
Amplitude Nonlinearity rectangular 0.577
Temperature Effects rectangular 0.289
Transverse Motion Effects rectangular 0.577

Combined standard uncertainty (%) normal 1.070

Combined expanded uncertainty (%) normal (k=2) 2.140
Table 2: Total Combined Uncertainty for 353B17 Internal Reference Accelerometer

2. Uncertainty contributors that arise from the transfer calibration and temperature response of the Endevco
2270 REF accelerometer (Table 3).

Total Combined Uncertainties for Endevco 2270 Back-to-Back REF Accelerometer

Standard Uncertainty %
Source of Uncertainty Distribution (k=1)
PCB 353B17 Uncertainty normal 1.070
Long-Term Charge Sensitivity Stability (Drift) rectangular 0.115
Temperature Effects rectangular 1.732
Transverse Motion Effects rectangular 0.346

Combined standard uncertainty (%) normal 2.048

Combined expanded uncertainty (%) normal (k=2) 4.096

Table 3: Total Combined Uncertainty for 2270 Back-to-Back Reference Accelerometer

3. Uncertainty contributors that arise from the calibration and performance characteristics of the SPEKTRA
SRS 35 VCU, including the data acquisition unit (DAQ), signal conditioners, A/D conversion and other
associated electrical hardware (Table 4).
Total Combined Uncertainties SRS 35 VCU
Standard Uncertainty %,
Source of Uncertainty Distribution (k=1)
Absolute Measurement of Voltage normal 0.105
Amplifier Gain Coefficients normal 0.020
Influence of High-Pass Filter normal 0.105
Signal Conditioner Frequency Response normal 0.200

Combined standard uncertainty (%) normal 0.250

Combined expanded uncertainty (%) normal (k=2) 0.500

Table 4: Total Combined Uncertainty for SRS 35 VCU

13
 4. Uncertainty contributors that arise from the calibration of the Votsch VT7004-S control thermocouple
inside the temperature chamber (Table 5).

Total Temperature Chamber Control Thermocouple Uncertainty

Temperature Range Standard Uncertainty %, (k=1)
-60°C to 90°C 0.667
Table 5: Total Combined Uncertainty for Chamber Control Thermocouple

The total system uncertainty, including contributors unique to Piezoelectric (PE) and Piezoelectric with Integrated
Electronics (IEPE) type DUTs at extreme temperatures and at ambient, respectively, are summarized in tables 6 and
7 below.

Temperature Shaker Calibration System Combined Uncertainty (Extreme Temperatures)

PE UUT IEPE UUT
Standard
Standard
Source of Uncertainty Distribution Distribution Uncertainty
Uncertainty (%)
(%)
Combined REF Calibration Uncertainty normal 2.048 normal 2.048
Calibration of SRS 35 VCU normal 0.250 normal 0.250
Transverse Motion rectangular 0.231 rectangular 0.577
Temperature Chamber Control normal 0.667 normal 0.667
ThermocoupleEffects on UUT
Temperature rectangular 2.598 rectangular 1.115
Accelerometers
Amplitude Nonlinearity rectangular 0.289 rectangular 0.577
Repeatability of UUT normal 0.324 normal 0.362

Combined standard uncertainty (%) normal 3.419 normal 2.596

Combined expanded uncertainty (%) normal (k=2) 6.839 normal (k=2) 5.193

Table 6: Total Combined System Uncertainty at Extreme Temperatures

Temperature Shaker Calibration System Combined Uncertainty (Ambient Temperature)

PE UUT IEPE UUT
Standard Standard
Source of Uncertainty Distribution Uncertainty Distribution Uncertainty
(%) (%)
Combined REF Calibration Uncertainty normal 1.093 normal 1.093
Calibration of SRS 35 VCU normal 0.250 normal 0.250
Transverse Motion rectangular 0.231 rectangular 0.577
Temperature Chamber Control normal 0.333 normal 0.333
Thermocouple
Amplitude Nonlinearity rectangular 0.289 rectangular 0.577
Repeatability of UUT normal 0.324 normal 0.362

Combined standard uncertainty (%) normal 1.397 normal 1.581

Combined expanded uncertainty (%) normal (k=2) 2.793 normal (k=2) 3.161

Table 7: Total Combined System Uncertainty at Ambient Temperature

14
 SUMMARY OF IMPROVEMENTS

Vibration Calibration System Comparison

UD Shaker TempShaker
 Retrofitted for temperature testing  Designed specifically for vibration calibrations at
 Industrial shaker extreme temperatures
 No ability to monitor accelerometer case  Modern calibration shaker
temperature directly  Temperature chamber controlled via thermocouple
 REF accelerometer not isolated from thermal mounted directly to REF accelerometer case
variation  Implementation of back-to-back REF whose
 Use of extension mount not back-to-back temperature coefficient is directly measured
 Temperature calibrations untraceable  Unbroken chain of traceability at extreme
 Ambient system uncertainty: ±4% @ k=2 temperatures
 Ambient system uncertainty: ±3% @ k=2
 Extreme temperature system uncertainty: ±7% @
k=2

15
 REFERENCES

1. ISO 16063-21:2003. Methods for the calibration of vibration and shock transducers – Part 21: Vibration
calibration by comparison to a reference transducer.
2. Sill, Robert D., Peres, Marco. “A New Solution for Shock and Vibration Calibration of Accelerometers.” Cal
Lab, October 2007, pp. 41-45.
3. ISO 16063-11:1999. Methods for the calibration of vibration and shock transducers – Part 11: Primary vibration
calibration by laser interferometry.
4. Mende, Michael. “Determination of the Back-to-Back Reference Accelerometer Temperature Coefficient.” Rev.
1.1 (2016-01-30)

16