Acceleration Measurement Optimization:

Mounting Considerations and Sensor Mass Effect

Marine Dumont, Acceleration Product Manager

Andy Cook, Mechanical Engineer
Norton Kinsley, Engineering Lab Supervisor
Kistler Instrument Corp. 75 John Glenn Drive, Amherst, NY 14228-2171 USA

ABSTRACT:

In the world of acceleration, a common topic is that of modal analysis. Within this topic, applications can range from the
study of bridges as vehicles roll across them, to the qualification of very delicate space and aviation equipment. One can see
that modal analysis contains a very large spectrum of applications. However, there is a small detail that is commonly
overlooked, and that is the proper mounting of very sensitive accelerometers. Mounting options include direct stud mounting,
wax mounting, magnetic mounting and a variety of options in-between. Although these options are diverse, they come with
varying stiffness and sometimes with the cost of addition mass, termed the mass loading effect. The first part of this paper
will take an in-depth look into some of the more common mistakes made during mounting, as well as a look into what can be
done to optimize the mounting to avoid unwanted results. In the second part of this paper, there will be an exploration into
the results of poorly mounted accelerometers; then a look in more detail at what the mass loading effect and stiffness are and
how these can drastically change the measurement results.

Key Words: Accelerometers, Resonance, Mounting, Sensor, Frequency

A- Sensor Mounting Considerations - Introduction

To obtain useful measurement information, an accelerometer must be coupled so that complete event information is
transferred. Mounting methods may vary, with some transferring event information more effectively than others. A high
performance accelerometer will behave like a low performance accelerometer if the mounting method is inadequate. The
transfer function behavior between the mechanical input properties and electrical output properties can be characterized by a
“Single Degree of Freedom” (SDOF) system with a mounted resonance frequency, which will decrease if the mounting
method becomes less stiff. This is illustrated in Figure 1 below.

Fig. 1 Oscillating Single Degree of Freedom system and its corresponding frequency spectrum

For more information, visit us on www.kistler.com/t&m/acceleration

If we simplify the sensor and its mounting method to be a similar oscillating system with a period T n in seconds and a natural
frequency fn = 1/Tn in Hz, where Hz = 1/seconds, the natural frequency of the system will be dependent on the mass m of the
system and the stiffness of the system through the spring constant k according to the equation (1) below:

1 k
fn  (1)
2 m

As a rule, the most rigid and lightest available mounting method option should be used at all times. The easiest mountings,
such as magnetic mounting and wax mounting, affect the high frequency event information reaching the accelerometer. The
reason for this is that it adds mass m to the system, as in the case of the magnet mounting base, and also reduces the spring
constant k as well. The application and the type of data desired should ultimately drive the mounting approach. If low
frequency events are being measured, a secure, easy mounting method is suitable.

Fig. 2 Sensor mounting configurations from lowest mounted resonance to highest mounted resonance

Figure 2 lists numerous accelerometer mounting options. To achieve the most accurate frequency response with the highest
stiffness k, the stud mounting method should be used. Unfortunately, this often leads to a more demanding preparation, such
as drilling and tapping a mounting hole, creating a high surface quality and incorporating the use of a torque wrench.
Although stud mounting an accelerometer produces the most accurate results, drilling and tapping a hole can cause changes
detrimental to the structure under study. Adhesive mounting using glue or wax is easier to handle but will restrict the
measurement temperature range and may also require solvent or heat to remove the sensor. Magnetic mounting bases allow
for a wider range of mounting positions, but with this flexibility, however, the magnetic mounting bases will restrict the
acceleration amplitude due to the higher mass m of the sensor and mounting base with respect to the magnetic force holding
the sensor to the structure. Magnetic mounting also requires a ferromagnetic surface as well as the additional weight of the
magnetic mounting base itself.

B- Sensor Mounting and Handling Rules

Once a mounting method is chosen, there are basic rules to follow in order to ensure the best measurement accuracy. First of
all, we are going to be looking into the stud mount method, which is the most commonly used mounting method. We can
start be looking at an application such as calibration.

Many accelerometers are specifically designed for stud mounting. Most mounting studs are machined from beryllium copper
which is known for high strength, low modulus of elasticity and high elastic limits. The studs on many types are removable
allowing for both stud and adhesive mounting. The following guidelines should be followed when stud mounting
accelerometers. Drill and tap an adequate hole to allow flush mounting of the accelerometer. Make sure the stud does not
bottom out and firmly secures the accelerometer. A chamfer should be machined at the top of the mounting hole to ensure
that the base of the accelerometer makes full contact with the mounting surface, as seen in figure 3. Completely clean the
surface prior to mounting and apply a thin coat of silicon grease to both the accelerometer and mounting surface. The
influence of this will be discussed later in this paper. Always use the proper sockets and a torque wrench when installing

For more information, visit us on www.kistler.com/t&m/acceleration

accelerometers. Tighten the accelerometer to the torque specified on the individually supplied calibration certificate. Do not
overtighten.

Fig. 3 Surface and mounting hole preparation recommendations - L1 and L2 would depend on the stud being used

Some accelerometers are specifically designed for adhesive mounting and require no special mounting adapters. Units
furnished with stud holes can also be used with adhesives. The mounting surface should be smooth and flat. A cyanoacrylate
type adhesive such as Eastman 910, Loctite 496, or super glue is recommended. See Table 1 for recommendations contingent
upon temperature and ‘Temporary/Removable’ or ‘Permanent’ configurations. While epoxies can also be used, cyanoacrylate
adhesives provide an extremely thin, stiff bond providing optimal frequency response.

Table 1 Adhesive recommendations according to temperature and Permanent/non-Permanent mounting

For more information, visit us on www.kistler.com/t&m/acceleration

When adhesive mounting an accelerometer with a tapped hole, make sure that no adhesive is allowed to enter the hole. This
could make stud mounting difficult or impossible at a future time. Remove the sensor with a manufacturer's recommended
adhesive solvent. Acetone is effective for the removal of cyanoacrylate adhesives. Once the adhesive is sufficiently softened,
you may use the proper size wrench to carefully twist the sensor loose from the mounting surface. Do not impact the sensor
with hard objects, such as a hammer, as this can damage the sensor physically and electrically. Do not attempt to remove
triaxial accelerometers solely by twisting with pliers, a wrench or impacting. Applying torque with a wrench or other tools
will damage the housing or the connector. Remove these accelerometers using a recommended cyanoacrylate solvent (e.g.,
Loctite 768), then twist with fingers.

If conditions permit, petro-wax is an ideal mounting material for these sensors. Bee's wax has also been used as a mounting
agent for many years but the recommended petro-wax (Kistler Type 8432 or P/N P102 from Katt and Associates) is a good
replacement for bee's wax since it has been formulated to provide improved frequency response. Wax is a good mounting
agent for lightweight sensors in temporary installations where near room temperatures are encountered. It is very often used
for calibration of sensors where stud mount is not possible, i.e. adhesive mount sensors only or a triaxial sensor with only one
mounting thread.

For all adhesives including wax, there is an optimal thickness to keep the mounting stiffness as high as possible. It is
recommended to apply 3 small amounts of wax onto the base of the sensor as illustrated in Figure 4. Press and turn the sensor
onto the mounting surface to spread the wax onto the entire surface. The influence of wax layer thickness will be shown later
on this paper.

Fig. 4 Apply three small amounts of wax on the sensor base

Magnetic mounting is a convenient way to take measurements on ferromagnetic surfaces. A magnetic mounting accessory
can add considerable mass m to the sensor and reduces high frequency response as mentioned previously. Care must be taken
to make sure the mounting surface is flat and clean. An oil or grease film greatly enhances the coupling characteristics, hence
improving the working frequency range. This mounting method is convenient, but it also can be very dangerous to the sensor
when mounting to a structure. When attaching to a structure where the magnetic mounting base is already threaded onto the
sensor, a sharp metal to metal impact can occur that can lead to sensor overload and damage. In order to prevent it, always
try to first mount the magnet onto the surface then thread the sensor onto the magnet mounting base or approach the
mounting surface with an angle such as illustrated below in Figure 5.

Fig. 5 Correct way of approaching mounting surface to installed sensor with magnetic base

In the next section, we are going to show the results of testing investigations and provide an easy methodology to determine
if a sensor and its mounting accessory mass have an influence on the measurement performed. This will also determine if the
proper mounting method has been selected to insure the needed frequency response.

For more information, visit us on www.kistler.com/t&m/acceleration

C- Hsu-Nielsen Source Test – Simple Method for Resonance Frequency Analysis

For a given accelerometer, the working frequency range with a sensitivity deviation of ±5% is typically around 20% of the
mounted resonance frequency as illustrated in figure 6. A higher input frequency spectrum excites the mechanical resonance
of the accelerometer and creates ‘amplified’ output signals. Linearity between the mechanical input and the electrical output
signal in this frequency range can be expected.

Fig. 6 Typical frequency response of an IEPE sensor when stud mounted

The Hsu-Nielson Source test method per EN ASTM Std. E976-1984 is an easy method using the breaking of a mechanical
pencil lead to produce wide frequency range excitation signal. Originally used for testing acoustic emission sensors, the test
is also excellent for testing the mounted resonance frequency of an accelerometer. The method is shown in figure 7 in which
an accelerometer is connected to an appropriate amplifier and the signal captured on a transient recorder or fast sampling rate
scope with trigger mode. Once the pencil lead breaks, it generates a wide frequency content pulse that makes the
accelerometer start ‘ringing’. This frequency is the resonance frequency in the mounted state for the accelerometer.

Fig 7 Sample setup of the Hsu-Nielson test to find the resonance frequency of an accelerometer (ASTM E976-10 [1])

Typically, a 0.5mm diameter, 2H pencil lead is used at an approximate length of 3mm. If a smaller intensity signal is needed,
a thinner diameter pencil lead can be used, e.g. 0.3mm or 0.35mm diameter. Conversely, if larger signal intensity is needed
then a thicker diameter lead, such as a 0.7mm diameter pencil lead, should be used. A softer pencil lead may also be used to
decrease the signal intensity as well. Pencil leads with a hardness of H, HB and B series will produce a lower frequency,
lower intensity input signal while harder leads such as the H series will produce a higher frequency, higher intensity input
signal. A plastic guide ring such as in Figure 7 is used to provide the right break angle to the surface of the test body. It also
prevents contact impact at the same time. Figure 8a and 8b show example of test results.

For more information, visit us on www.kistler.com/t&m/acceleration

 Fig. 8a & b Analysis of the transient response with FFT. (a - time domain, b - frequency domain)

D- Sensor and Mounting Accessory Mass Effect

The dynamic properties of a structure are dependent on its mass, rigidity and damping. If the mounting of an accelerometer
applies an additional mass to the structure, these dynamic properties change. The resonance frequency fn of the structure
reduces by approximately Δf where m is the mass of the structure and ma is the mass of the accelerometer (2). Vibration
amplitude a0 reduces approximately as shown by equation (3).

m a m
f  f n (1  ) (2)  (3)
ma  m a0 ma  m

If the effect of adding an accelerometer to a test structure is found to be negligible, then the mass of the accelerometer must
be negligible compared to the mass of the structure being measured. An example can be found in Figure 9, which shows the
same measurement being made using a low mass or a high mass accelerometer. One can see very quickly the influence of the
sensor mass on the resonance peak frequency and amplitude.

LOW MASS
ACCELEROMETER

HIGH MASS
ACCELEROMETER

HIGH MASS LOW MASS

ACCELEROMETER ACCELEROMETER

Fig. 9 Structural Analysis of the same structure with a high mass accelerometer (red) and a low mass accelerometer (purple).
Source: Experimental Techniques, Jan/Feb 2002

As a part of one’s daily work, this can be checked very easily using the Hsu-Nielsen test method from part C. A first Hsu-
Nielsen test should be performed with the sensor mounted on the structure and then a second test with double the
accelerometer mass using a second sensor used for mass only. If a second sensor is not available, then an equivalent mass
can be used. If there is no significant change to the results, this suggests that the sensor mass has little effect on the structure.

For more information, visit us on www.kistler.com/t&m/acceleration

E- Experimental Analysis of Mounting Method Influence on Frequency Response

A study has been conducted using the Hsu-Nielsen test method with a sensr mounted on a tungsten block to illustrate some of
the points mentioned in sections A and B. This will help provide an understanding as to why selecting the right accessory is
of high importance and why following the recommendation for usage of this accessory is important for measurement
accuracy. The first portion of the study uses a single axis PiezoStar sensor type 8703A50M5. This sensor is specified
frequency response up to 10 kHz at ±5% with a nominal resonance frequency of 40 kHz when stud mounted. The test data
will show how the different mounting types can influence the mounted resonance frequency for the very same sensor when
the rules for each recommended mounting option are carefully followed. The first resonance peak is the one to be used in this
investigations as it is the first one used to assume the +/-5% Frequency response (0.22 % of the first resonance peak).
Multiple resonance peaks can be seen that are a combination of the sensor element design, the tungsten block itself on which
the sensor is mounted on and the mounting method. Because the sensor element and the tungsten block are unchanged during
this investigations, any resonance peak change will be directly linked to the mounting method used.

The first measurement was conducted with the sensor being stud mounted using mounting grease and the appropriate
mounting torque. A resonance frequency of 39.1 kHz is measured as shown in figure 10. This resonance frequency value will
be used as a baseline for other measurement techniques as this method provides the best result. A second measurement was
performed with the same test conditions except the accelerometer was mounted using an isolated mounting stud. This should
reduce stiffness of the system so should reduce the resonance frequency. As seen in figure 11, the resonance frequency is
slightly lower at 37.9 kHz (-3% from baseline). The isolated mounting stud had very little influence on the resonance
frequency.

fn = 39.1 kHz
Fig. 10 8703A50M5 mounted using stud, recommended mounting torque and grease, fn = 39.1 kHz

fn = 37.9 kHz
Fig. 11 8703A50M5 mounted using isolated mounting stud, fn = 37.9 kHz

For more information, visit us on www.kistler.com/t&m/acceleration

The sensor was then wax mounted using the proper amount of wax and proper mounting technique. Figure 12 shows the
resonance frequency measured at 35.5 kHz (-9% from baseline). This provides an acceptable result allowing an accurate
method for sensor calibration. Of course, this does not allow for as high a frequency measurement as the stud mount.

fn = 35.5 kHz
Fig. 12 8703A50M5 mounted with ideal amount of wax, fn = 35.5 kHz

Last but not least, let’s have a look at the same test using a magnetic mounting base. The magnetic mounting base adds mass
to the accelerometer/mounting base system. The stiffness of the mounting system relies on the strength of the interaction
between the magnet mounting base and mounting surface. The test results show the resonance frequency is lowered down to
17.5 kHz (-55% from baseline) as shown in Figure 13. This confirms the statement from section A: the most rigid and
lightest available mounting method option should be used at all times.

fn = 17.5 kHz
Fig. 13 8703A50M5 mounted with magnetic mounting base, fn = 17.5 kHz

We advised in section B to use grease and the recommended mounting torque when stud mounting an accelerometer. Figure
14 shows measurement results using the 8703A50M5 sensor with different mounting configurations. It can be seen that the
worst case scenario would be low mounting torque and no grease. In this case, a reduction of the resonance frequency by
25% from baseline is observed. In the next case, where grease is used but the mounting torque is still too low, the resonance
frequency is reduced by 6% from baseline. Once the recommended mounting torque is applied along with the usage of
grease, the frequency response is optimized. In this test environment, the grease has very little influence on the resonance
frequency due the good quality of the mounting surface. In cases where mounting surfaces are poor and there are gaps
between the accelerometer and the mounting surface, the use of grease improves performance due to better adherence
between the mounting surfaces and the filling of gaps.

For more information, visit us on www.kistler.com/t&m/acceleration

 14a 14b

fn = 39.1 kHz fn = 38.0 kHz

14c 14d

fn = 36.6 kHz fn = 29.2 kHz

Fig. 14a, b, c & d 8703A50M5 stud mounted with vs. without grease and with vs. without recommended mounting torque –
a - stud mounted using grease and recommended torque: fn = 39.1 kHz, b - stud mounted using no grease and recommended
torque: fn = 38.0 kHz, c - stud mounted using grease and low torque: fn = 36.6 kHz, d - stud mounted using no grease and low
torque: fn = 29.2 kHz.

The next part of the study conducted looked into frequency responses using wax mounting methods. Figure 15 shows three
resonance plots of the same 8703A50M5 used in the other tests. The mounting conditions were defined by using a less than
ideal amount of wax, the recommended amount of wax and more than the ideal amount of wax. Figure 15b shows little
difference in resonance when using a less than ideal amount of wax, only a drop of 5.6%. But what cannot be seen on the
plot is that the strength of the wax mounting bond is much less which in turn leads to a lower useable acceleration level. It is
much more likely that the accelerometer will detach from the mounting surface when a less than ideal amount of wax was
applied. On the other hand, too much wax will lead to a more compliant mount leading to a decrease in resonance frequency
of about 60% from ideal.

For more information, visit us on www.kistler.com/t&m/acceleration

 ©2016, Kistler Group
920-688c-02.16
15a 15b

fn = 35.5 kHz fn = 35.5 kHz

15c fn = 14.2 kHz

Fig. 15a, b &c 8703A50M5 wax mounted – a - ideal amount of wax: fn = 35.5kHz, b – less than ideal amount of wax: fn =
33.5 kHz, c - more than ideal amount of wax: fn = 14.2 kHz

CONCLUSION:

The mounting method of an accelerometer should not only be considered by the ease of mounting criteria, but also by the
corresponding upper frequency response for the chosen mounting method. Each mounting solution acts as a spring; the more
flexible the coupling, the lower the frequencies that have to be produced in order to induce resonance phenomena. The most
rigid connection is possible with a stud mount, while the least rigid can be obtained with magnetic mounting. The wax
mounting and adhesive mounting methods fall somewhere in between, with layers as thin and hard as possible while still
attaining optimal performance. In this paper, a study of mounted resonance frequencies was performed using the most
popular mounting configurations in order to highlight how much sensor mass, mounting accessory mass and rigidity can
influence the mounted resonance frequency of the system and hence the frequency response of an accelerometer.

As an easy assessment, the Hsu-Nielsen test can be used to determine the resonance frequency of a mounted accelerometer.
From this result, the rule of thumb that the +5% frequency response deviation is somewhere around 0.2 times the resonance
frequency can be applied.

LITERATURE

[1] ASTM E976 - 10 Standard Guide for Determining the Reproducibility of Acoustic Emission Sensor Response

For more information, visit us on www.kistler.com/t&m/acceleration