ACCELEROMETER SETS Type 4316

ACCELEROMETER PACKAGES Type 4356

The Miniature Accelerometer Type 4316 is de-

s igned especially with a view to vibration meas-
urement on lightweight structures, where a heavy
tra nsducer would change the mode of vibration
and thereby invalidate the results obtained. It
will be found valuable in research projects in-
volving thin plates and shells, such as aircraft
skins, car bodies, component testing etc.

B UEL&K.JJER
Naerum. Denmark . ~ 80 05 00 . ~ BRUKJA, Copenhagen . Telex: 5316

BB 4316 / 4356
 Accelerometer Sets
Type 4316

Accelerometer Packages
Type 4356

NOVEMBER 1966
 Contents
0. Introduction 3
1. Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2. Factory Calibration .......... : . . . . . . . . . . . . . . . . . . . . . . . 15
3. Mounting Methods .................................. 24
4. Measuring Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
5. Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
6. Instrument Combinations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
7. What to Order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
8. Conversion Charts, Tables etc. . . . . . . . . . . . . . . . . . . . . . . 47
9. Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
 0. Introduction
The accelerometer is an electromechanical transducer which produces at
its output terminals an e.m.f. proportional to the acceleration to which the
transducer is subjected. The output signal can be electronically processed and
read on a meter or some other suitable indicating device.
The Bri.iel & Kjrer accelerometers are small, light and of rugged construction,
suitable for most vibration measurements, both in the laboratory and in field
environments. Versatility in use has been a prime goal and special care has
been taken to ensure a wide frequency range, high sensitivity and good
temperature characteristics. Furthermore, these accelerometers have negligible
mounting error and low sensitivity to severe environmental conditions, such as
humidity, high temperature, corrosive atmospheres and magnetic fields.
The Miniature Accelerometer Type 4336 is designed especially with a view to
vibration measurement on lightweight structures, where a heavy transducer
would change the mode of vibration and thereby invalidate the results . ob-
tained. It will be found valuable in research projects involving thin plates and
shells, such as aircraft skins, car bodies, component testing etc.
Measurements can be carried out as measurements of acceleration using one
of the B & K preamplifiers or cathode followers, and as measurements of
velocity or disptacement employing integration networks.
For absolute measurements it is necessary to know the sensitivity of the vibra-
tion transducer employed. Each B & K accelerometer is supplied with individual
calibration data and a frequency response curve, all taken as part of the
production test procedure. The aim has been to supply the maximum amount
of information on the calibration sheet to ensure a completely predictable
performance of the accelerometers.
The Miniature Accelerometer has type number 4336 when on its own in the
production and calibration stages. When it is sold it is accompanied by
certain accessories such as connecting cable, mounting studs etc., and is
available only as Accelerometer Set Type 4316 or as Accelerometer Package
Type 4356. See section "What to Order" at the end of this book.

3
 1. Description
Construction.
The Miniature Accelerometer Type 4336 is of the piezoelectric compression
type with a construction indicated schematically in Fig. 1.1.

?"""'-=:<'--tt--Spring
Isolator
Mass
Output
Terminals
Piezoelectric Disc
Base
266118

Fig. 1.1. Schematic drawing of a piezoelectric accelerometer.

The transducing element consists of a piezoelectric*) disc on which is resting

a heavy mass. The mass is preloaded by a stiff spring and the whole assembly
is mounted in a metal housing with a thick base. When the accelerometer is
subjected to vibration, the mass will exert a variable force on the piezo-
electric disc. This force is exactly proportional to the acceleration of the
mass. Due to the piezoelectric effect a variable potential will be developed
across the disc, which is proportional to the force and therefore to the
acceleration of the mass. For frequencies much lower than the resonance
frequency of the mass and the stiffness of the whole accelerometer system
the acceleration of the mass will be virtually the same as the acceleration of
the whole transducer, and the potential produced will therefore be pro-
portional to the acceleration to which the transducer is subjected. This
potential can be picked up from the output terminals of the accelerometer and
used for determination of the vibration amplitude, waveform and frequency.
It has been the aim with the miniature accelerometer to make available a
transducer of extremely low mass, which at the same time has the versatility
of a general purpose accelerometer. High precision and stability coupled with
a rugged construction make the accelerometer suitable for the most demand-
ing applications both in the laboratory and in the field.

*) Lead Zirconium Titanate.

4
 r
~
(D +
~
M
c::) ~
('I)
I ci
E f

E
N
a;

Fig. 1.2. Physical dimensions of the Miniature Accelerometer.

Physical Dimensions.
It is generally true that an accelerometer should be as light as possible in
order not to influence the vibration of the specimen on which it is mounted.
For measurements on heavy machinery etc. this causes no problem but for
lighter structures, such as for example a thin metal plate, the transducer
weight is important.
The resonance frequency of a vibrating single degree of freedom system is

fa= _1_ 1/ k
2 .n V rna
where rna is the effective lumped mass and k is the lumped stiffness restrain-
ing the mass.
It is easily seen that adding another mass, e.g. an accelerometer, will result

v
in a change in resonance frequency according to the formula

f1 rna
T = m1 + ffio
where f1 is the new resonance frequency and m1 is the added mass.
Thus any mass added to a vibrating structure will influence the vibration to
some extent, but provided the added mass is small the influence is often
negligible. A 10% increase in mass will reduce the resonance frequency by
about 5%.
The sensitivity of a piezoelectric accelerometer is directly proportional to the
mass acting on the piezoelectric disc. This limits the extent to which the
accelerometer weight may be reduced without loosing too much of the
sensitivity. For many applications it is also important that any strain developed
in the material on which the accelerometer is fixed is not transmitted to the
piezoelectric element as this would show up in the output signal, indicating
an acceleration which actually is not present. It is therefore necessary to make
the accelerometer base thick, in order to isolate the piezoelectric disc from
such strain. The requirements to high sensitivity, low strain sensitivity and good

5
high frequency response are conflicting with requirements to small physical
dimensions and weight, so that a miniature accelerometer Is bound to have
a relatively low sensitivity. The housing of the miniature accelerometer has
been made of titanium, and a high ratio of sensitivity to weight has been
obtained without sacrificing any of the more important qualities of the
accelerometer.
The total weight of the Miniature Accelerometer Type 4336 is 2 grams (o.07 oz.).
Its main dimensions are given in Fig. 1.2. The weight is given without cable as
it is difficult to predict how much the cable will add to the weight of the
accelerometer. However the plug has been designed to give a very small
extra mass and very reliable contact, and is situated very close to the mount-
ing surface in order to allow the cable to be fixed to the mounting surface as
close to the accelerometer as possible. This will ensure the most reliable
operation. Furthermore two types of cable are included. The soft type should
be used at temperatures up to 100°C (212° F) and the stiffer teflon cable at
higher temperatures. For very critical work where the cable cannot be
mounted directly onto the vibrating surface close to the accelerometer we
recommend to remove the outer coating of the cable from about 10 mm from
the plug and over as long a part as necessary. See Fig. 1.3. The sleeve is
easily removed by a sharp razor-blade'. Avoid however to cut down into
underlying braided shield.

~ smww
-J1om~100mm _J z66!20
0.4'' 4"
Fig. 1.3. Removal of outer cable insulation to reduce weight and stiffness.

Environmental Sensitivity.
Accelerometers are often used to measure vibration in the field or on speci-
mens subjected to severe environmental tests. It is therefore important that
their sensitivity to environmental changes is as small as possible. The factors
that may influence accelerometer performance are primarily temperature,
humidity and rapidly varying ambient pressure (Sound). The temperature
effect is to reduce the voltage sensitivity of the accelerometer at higher
temperature, but if the accelerometer has undergone a suitable temperature
cycling process in the production stages, the sensitivity will revert to its
normal value when the temperature is brought back to normal again. Beyond
a certain temperature (the Curie point) the piezoelectric element is permanent-
ly damaged.

6
The Bruel & Kjrer accelerometer Type 4336 is designed to be used for
temperatures up to 260°C (500° F) without cooling. The sensitivity will be
slightly reduced at the higher temperatures, but the necessary heat cycling
process has been carried out, so that no permanent change will take place.
It should be noted that great care has been taken to use materials which will
withstand high temperatures. The thermal coefficient of expansion of insulating
material and metal parts are carefully matched in order to maintain humidity
sealing.
It is, however, very difficult to guarantee a perfect seal for immersion in
water when the accelerometers have been exposed to the high temperature
limit. Therefore accelerometers which are to be used in liquids should not
be exposed to more than 150°C for long periods of time.
The accelerometers are brought up to 250° C and left to cool down to about
50° C several times during production and calibration until performance Is
stable.

dB
6
5
4

Charge

Joo •c
-2
-3
-4
-5
-6
z66121

Fig. 1.4. Typical performance characteristics for the accelerometers in the

temperature range -100 to + 260°C.

Tests have also been conducted in order to find the influence of low
temperatures on accelerometer performance. The voltage sensitivity increases
steadily down to some -100°C and then levels out, while the capacity undergoes
a gradual decrease with decreasing temperature. The charge sensitivity is
practically constant.
Typical performance characteristics are given in Fig. 1.4 for the temperature
range -100 to +260°C (-150 to +500°F).

7
Cooling.
When the accelerometers are fixed on surfaces with temperatures exceeding
250°C it is possible to reduce the temperature of the accelerometer base by
inserting a cooling plate between the base and the measuring surface.
When a mica washer such as the one included in the Accelerometer 4316 is
used between the mounting surface and the cooling plate, experience has
shown that temperatures up to 350-400°C may be measured on the mounting
surface with less than 250° C in the accelerometer base.
A stream of air passing over the cooling plate considerably aids the cooling
process. The set-up shown in Fig. 1.5 has been tested in the laboratory.

Aluminium
Screen

Air~- · ~
77777777777777~ 5r//?M7/
266f22.

Fig. 1.5. Forced air cooling of an accelerometer.

In this set-up the mounting surface mas heated to 450°C while less than 200°C
were measured in the accelerometer base.
For high temperature measurements one should always employ Teflon cable,
B & K part No. AO 0052. Cable No. AO 0051 is based on Polyethylene and
PVC which will not withstand more than 100°C. Also for low temperature
measurements, from --40°C downwards, the Teflon cable should be employed .

Vacuum Test.
The B & K accelerometer Type 4336 is sealed and tested under water in an
evacuated jar for leaks. This eliminates the risk of inferior performance in
moist atmospheres or environmental test chambers where heavy condensation
may take place.
A photograph of the test set-up is shown in Fig. 1.6. The smallest leaks show
up immediately as bubbles rising to the surface.
When the accelerometers are used for measurements in liquids or in very
moist environments it is necessary to seal the cable entry, as shown in Fig. 1.7.
A good sealant for the cable entry is for example Dow Corning Silastic RTV 731
(room temperature vulcanizing silicon rubber) or General Electric equivalent.
These sealants show excellent performance for a wide temperature range
(-1 00 to + 500 o F).

8
 Fig. 1.6. Vacuum test set-up.

Sealant

m»»»mmEIW.:.m, 26572rl

Fig. 1.7. Sealing of cable entry.

Acoustical Sensitivity.
Effects due to acoustical excitation are unavoidable with piezoelectric accelro-
meters, but may be reduced by careful design. The B & K accelerometers are
of a rigid, mechanically isolated construction and pressure variations in the air
will have little effect on the force exerted on the piezoelectric element. One
may generally assume that the vibrations induced in the vibrating specimen
will give rise to a much higher acceleration signal than the direct acoustical
excitation of the accelerometer itself. In cases where very low accelerations
are to be measured in an intense acoustic field, however, care must be exer-
cised in order to obtain correct results. Typical acoustic sensitivity for the
accelerometers has been measured and found to be less than o.5 t-tVI t-tbar,
i.e. less than o.5 t-tV for 74 dB sound pressure level. At 140 dB sound pressure
level the output is less than 1 mV.

9
Magnetic Sensitivity.
The magnetic sensitivity has been found to be less than 1 p,V/Gauss for the
least favorable orientation of the accelerometer in the magnetic field.

Polarity.
The polarity of the accelerometers is such that an acceleration directed from
the mounting surface into the body of the accelerometer results in a positive
e.m.f. on the centre conductor of the output terminals.

1/1
~~A
Volts
60 f-
~ 2615
.----.
A-r' 'l
-~
I os1 1ve Pu se
I I I
50 - I~ / ~egative Pulse

I
1-

/
I Shock
Machine
I/
40 f- Memo-
scope

1/
v
30
/
20 /
/
10 /
/
v
0 1000 2 3 4 5000 6 7 ~ 9 10000 11 12 13 14 15 OOOg
266fZ'f

Fig. 1.8. Part of shock test machine and typical output versus shock
amplitude plot.
Shock Performance.
In general the accelerometers will withstand shocks higher than 16ooo g in
the positive direction, however, the mass is pressed against the ceramic with
a certain force. This force determines the upper limit to which the accelero-
meter output will increase linearly with increased shock amplitude in the
negative direction. When this limit is reached the output will be clipped as
seen from the typical curve shown in Fig. 1.8.

Effect of Mounting Torque.

The effect of mounting torque is less than 1 % change in sensitivity up to
Sooo Hz for 2-20 kgcm (2 to 16 lb-in) mounting torque.

10
Long Term Stability.
A prime goal during design and manufacture has been to achieve maximum
stability under severe environmental conditions. However, when an accelero-
meter is used for measurement under severe conditions of shock and heat,
one should not rely on this as a primary standard. It would be advisable to
recalibrate such an accelerometer at relatively short intervals of time.
A calibration vs. time history is closely followed for a number of representative
units taken from production lots and they show less than 2 % change per year.

Piezoelectric Materials.
The quality of piezoelectric accelerometers depends largely upon the per-
formance of the material used for the sensing elements. Monocrystalline
materials such as quartz and Rochelle salt, have been used in the past, but
are now superseded by polycrystalline, artificially polarized ceramics like
barium titanate, lead zirconium titanate, lead metaniobate or similar materials.
Large variations in electrical and mechanical properties are obtained by small
changes in composition, and most manufacturers are continuously investigating
and improving their compositions in order to make them more suitable for their
particular purposes.
The most important factors for general purpose accelerometers are Curie
point, sensitivity, temperature stability, capacity, resistance and time stability.
Unfortunately some of these are conflicting, high sensitivity for example, has
often to be sacrified for good time or temperature stability.
In the table below are given some properties of commonly used ceramics as
given by their manufacturers. The Curie point is the temperature at which the

Piezoelectric Dielectric
Manufacturer and Curie point
Constant Constant
Trade Name oc 10- 12 Coui./Newt. (Rei. Perm.)

Clevite
Ceramic "B" 115 149 1300

Clevite
PZT4 325 285 1475

Quartz & Silice

p 1-60 351 400 1500

Plessev
Casonic Grade 3 120 83 1000
~

_ ,

Quartz 300 2 4

B & K (1966) 350 300 1500

11
ceramic changes its crystal structure and looses its polarization, the piezo-
electric constant indicates the sensitivity, and the dielectric constant indicates
the capacity for a given shape and size. Trade names are used as the true
composition of the ceramic is usually not disclosed by the manufacturer.

Natural Resonance Frequencies.

The natural resonance frequency of an accelerometer is not fixed. It depends
not only upon the mass and stiffness of the accelerometer but also upon the
mass (and stiffness) of the object on which it is mounted.
Some confusion persists as to what is the natural frequency and different
definitions are seen.

Seismic mass

Contact stiff ness

Piezoelectric disc

Contact stiffness
M
Piezoelectric disc

Contact stiffness K= Equivalent

stiffness

Mass of base

a) b)
266o20

Fig. 1.9 a. Schematic drawing of accelerometer as a dynamic system.

b. Simplified system.

The situation is best Illustrated with a drawing. Fig. 1.9a shows a schematic
drawing of an accelerometer which consists of several masses and springs
connected in series. These springs are representations of the contact stiffness
between the various parts of the accelerometer.
This mechanical system can be further simplified into that of Fig. 1.9b. Here
M is the seismic mass resting on the piezoelectric element, B is the mass of

12
the accelerometer base and housing. K is the equivalent stiffness of the
system between M and B. The natural frequency of such a system is equal to

fo=fM F ~
where fM is the natural resonance frequency of the mass M upon the spring of
stiffness K.
Now two resonance frequencies are easily thought of:
1. The free hanging resonance frequency, i.e. the resonance frequency obtain-
ed with the system freely suspended in air.
This resonance frequency is entirely dependent upon the ratio of M and 8,
and it is seen that making the base 8 very light the resonance frequency
may be very high. This resonance frequency is therefore of no practical
value, in fact the higher it is, compared with the mounted resonance fre-
quency defined below, the poorer is the mechanical construction of the
transducer. (A thin base may cause bending of the piezoelectric due to
strain from the mechanical part on which the vibration is measured).
2. The mounted resonance frequency with the accelerometer firmly fixed to
a structure of infinite stiffness and mass.
This results in a value for 8 of infinity and the resonance frequency is
equal to fM which is the natural resonance frequency of the mass M on the
spring of stiffness K.
This resonance frequency is of great practical value as it is approximately
the one obtained when the accelerometer is mounted on a structure which
is heavy compared with the accelerometer mass.
Unfortunately it is difficult to measure the mounted resonance frequency as
defined above, since an infinitely heavy mass is hard to bring into motion.

~//~//#/~~
B=base+180 grams

266ol9
Fig. 1.10. Possible configurations for definition of natural resonance frequency.

13
A compromise has therefore to be made and there are standards which define
the mounted resonance as that obtained with the accelerometer mounted on
a one inch cube of steel (Bruel & Kjrer use a steel block of weight 180 gram).
Such a block is easily set into controlled motion at frequencies up to some
50 kHz.
This mounted resonance may be used directly as a basis for vibration measure-
ments as long as the accelerometer seismic mass is much smaller than the
mass of the calibration mounting block. The natural frequency obtained is
practically identical with the mounted resonance frequency on an infinite mass.
However when the accelerometer mass is large the natural frequency obtained
will be too high, thus if B = 5 M we obtain

fo = fM v' 1 + 0.2 = 1.1 fM

i.e. a value which is 10% high.
For the miniature accelerometer the ratio of weights of the seismic mass to the
weight of the base is about 1 : 1 and therefore the free hanging resonance will
be about 1.5 times the mounted resonance and the resonance obtained on
the 180 grams block of steel is o.S% higher than that which would be obtained
on an infinite mass.
A small difference in resonance frequency is obtained by considering the
damping present in the accelerometer. Thus undamped natural frequency is
defined as the frequency at which the difference between the motion of the
mass and base lags the motion of the base by a phase angle of 90°, whereas
damped natural frequency is the frequency of decaying oscillations after an
initial excitation. The difference is negligible for high Q piezoelectric accelero-
meters such as the ones manufactured by Bruel & Kjrer.
The undamped natural frequency given on the accelerometer calibration
charts is obtained with the accelerometer fixed firmly to a steel block of
weight 180 grams.

14
 2. Factory Calibration
Each individual accelerometer has before leaving the factory undergone a
very thorough ageing, testing and calibration procedure to ensure the user of
a high quality product. A typical calibration chart as supplied with the accelro-
meters is shown in Fig. 2.1.

C• ll braHon Ch•rtfor
Ac:c.ltotolll.t.rTYftot4Jl6

S.ri• l - f<M~.- Br~,'••~.!~.,,.

"':!...:r::.~=~~~~~- -l 0 0 0 0 0 0 0 0 0 0 0 0 0 D 0 0 0 0 0 0 -·-
0 0 0 0 0 0 0 0 0 0

•-5-""'lfr•t ~-~-~-e.l•n _ Zt _ oc
a n41ftC'hHIInl
c-~,_..,. of _./97. ______ "' ,
'l'...... s-.tll.tly .J.~~"'Y/1"
a....,.s-~~twt~y __u_~--- i>Cit;
C.t>oc~ty (l•cluclt~IU"'-) -- 3Zf ___ pl'"

~~~Wo•'-1-s.-.NMtya\ 80c./e __~6- '/o

~M.-.!....,._... ,_,_ J?" .....__ II.o/a J~-..t~oitrO
£~~E::,:=:~~~::~::r:+~::~~ ,~Mm=mmm,~
·...,+10

- ·.
••• - oeo.e c"''!"~· ~::,~- ;;;
.,r,o
..._
Fig. 2.1. A typical calibration chart.

Cable Capacity.
The capacitance of the accelerometers are given including the capacitance
of the ordinary B & K connecting cable, so that when other cables are used,
the change in cable capacitance must be taken into account, as this deter-
mines the effective sensitivity of the accelerometer. To find the capacitance
of the accelerometer alone, deduct the cable capacitance given on the calibra-
tion chart. The standard length of cable included is 1.2 m (4 ft).
The capacity of the 1.2 m cable is around 105 pF (micromicrofarad) but the
individual value is given on the calibration chart.

Voltage Sensitivity.
The voltage sensitivity of the accelerometers with connecting cable is deter-
mined at a frequency of 50 Hz and at room temperature (appr. 20°C).
Fig. 2.2 shows the instrumentation employed.
The accelerometers are calibrated at 50 Hz, the frequency being monitored by
a crystal controlled electronic counter, and the periodic time of the vibration
is kept within 200ooo ± 100 flSec. Thus the frequency accuracy is ± o.05 %.
The shaker providing the mechanical excitation is mounted on a large concrete

15
block which is supported by steel springs, giving the whole assembly a natural
resonance frequency of about 1 Hz. This ensures isolation from background
vibration.

Frequency
Analyzer
Type 2107

Unknown
Accelerometer

265726

Fig. 2.2. Set-up used for calibration of absolute sensitivity of the accelero-
meters.

The calibration of all units is achieved by comparison between a standard

accelerometer and the unknown in a back to back mounting as shown in
Fig. 2.2. The standard is frequently checked with a microscope observing the
displacement directly, with an accuracy of ± 1 %.
The distortion in actual vibration units is kept below 2 % and max. influence
on peak to peak amplitude below o.S %.
The output from the accelerometer is measured with a B & K Frequency
Analyzer set to 50 Hz and calibrated with a reference voltage of exactly the
same frequency as the signal to the shaker, giving a total voltage measuring
accuracy of ± o.15 %.
The total accuracy on the sensitivity calibration of the accelerometers Is thus
better than ± 2 %.
The calibration vs. time history is closely followed for a number of representa-
tive units taken from production and they show less than 2 % change per year.
Note that the calibration is valid for the accelerometer with its individual
connecting cable, and remember to take any change of cable capacitance Into
account when different cables are employed.

16
Also the input capacity of the preamplifier must be taken into account. The
influence of this capacity in the calibration set-up is eliminated with a substitu-
tion measurement. In practice the capacity will be less than 10 pF using a
normal B & K amplifier and plug such as B & K Cathode Follower Type 2615,
Source Follower 2616 or 4 pF in the case of Preamplifier 2623.
Note also the units employed in the calibration. The sensitivity is given in
mV/g = mVr.m .• ./g,.m.s. = mVp•• k/gp••k· If, for example, the sensitivity is required
in mVr.m.• ./gP••k the sensitivity given should be multiplied by o.707.

Charge Sensitivity.
This sensitivity is calculated from the voltage sensitivity and the equivalent
capacitance of the accelerometer with normal connecting cable.
Charge sensitivity is expressed in pico-coulomb/g and is independent upon the
capacitive loading on the accelerometer. It is determined by multiplying the
voltage sensitivity with the total capacity used in the calibration, i.e. capacity
of transducer and cable.
Scharge = S vo ltoge (C.+ Cc)
where c. = accelerometer capacity and c. = Cable capacity in nF.
Since the voltage sensitivity is given as mV/g the charge sensitivity will be in
pico-coulomb/g.

Capacity.
The capacity of the accelerometer is given including the cable capacity as
specified on the calibration sheet. Accelerometer capacity comes into the
question when the low frequeny cut-off of the measuring system is computed,
as It determines the effect of loading on the accelerometer.
The capacity is measured at 1ooo Hz in a capacitance bridge, comparing with
a standard capacity equal to the nominal capacity of the accelerometers.
Accuracy ± o.5 %.

Transverse Sensitivity.
Transverse sensitivity is the maximum sensitivity to a transverse acceleration
expressed in percent of the reference sensitivity in the intended measuring
direction. It is primarily due to irregularities in the ceramics, and limitations in
the mechanical coupling between ceramic and metal parts. Careful mechanical
machining helps to minimize transverse sensitivity.
Transverse sensitivity can be regarded as a result of the axis of the accelero-
meter making a small angle with the direction of the maximum sensitivity as
shown in Fig. 2.3.
Imaging the accelerometer placed in a rectangular coordinate system, as
shown, the vector representing the maximum sensitivity can be resolved into
two components: the main axis sensitivity which is the one called voltage or
charge sensitivity on the calibration chart, and the transverse sensitivity given
on the calibration chart as a percentage of the main axis sensitivity.

17
 Maximum
Main axis sensitivity
sensitiVity I
I
100%
l
l
l
l
I
I
I
I
I

\
\
\
\ Minimum
, sensitivity
in this
direction

Fig. 2.3. Graphical illustration of transverse sensitivity.

The voltage obtained from an accelerometer is a product of the acceleration

vector and the component of the sensitivity in the same direction, so that in
a certain transverse direction, at right angles to both the main axis and the
max. transverse sensitivity axis the accelerometer output will be minimum.
In order to maintain a low transverse sensitivity the user should always mount
the accelerometer on a flat, clean surface and keep the accelerometer mount-
ing surface free from burrs and scratches. Also he should avoid very large
shocks in the transverse direction, such as might be caused by dropping the
accelerometer on the floor, and large temperature shocks.
If the accelerometer is properly handled and mounted, the transverse sensitivity
will normally be below 10% up to 5ooo Hz. Above 5ooo Hz it is very difficult
to establish useful test facilities for transverse measurements. The type of
mounting of the acceleror:neter will, however, undoubtedly set a limit to the
good performance of the accelerometer in the transverse direction at a fre-
quency somewhat lower than that for operation in the main direction.
The transverse sensitivity of the accelerometers is specified as being less than
8 % of the main axis sensitivity .However, the individual maximum transverse
sensitivity is given on the calibration chart.
In the factory calibration the transverse sensitivity is measured at approximately
30Hz on a rotating table. whose motion sideways is more than 100 times larger
than the motion in the vertical direction. See Fig. 2.4.
The accelerometer is first mounted with its main axis parallel to the direction
of motion of the table and the deflection on the meter is adjusted to a

18
 Fig. 2.4. Part of the arrangement for obtaining transverse sensitivity.

reference mark (100 %). Then the accelerometer is mounted with its main axis
at right angles to the motion and the table is rotated slowly. The meter is set
to indicate full scale deflection for 10% transverse sensitivity and the maximum
value is noted.
The factory calibration of transverse sensitivity is carried out after 24 hours
of storage at room temperature.

Undamped Natural Frequency.

The undamped natural frequency is the frequency at which the motion of the
seismic mass element lags 90 ° behind the motion of the accelerometer
housing.
The damping ratio, which is the ratio between the actual damping of the
accelerometer and the damping for which the system would be critically
damped, is very low: in the order of o.02. Therefore the damped natural fre-
quency, fd, is nearly identical to the undamped natural frequency, fu, since

fd = fu V 1 - 0.022
The undamped natural frequency is measured electrically by finding the lowest
frequency at which the 90° phase shift occurs when an electrical signal is
applied to the terminals in series with a 300 pF capacitor, and with the
accelerometer mounted on a block of steel of weight 180 grammes.
The mounted resonance frequency is the frequency at which the sensitivity of
the pick-up is a maximum when mounted on a stainless steel block of approx-

19
imately 180 grammes. As the mounted resonance given is obtained under
actual operating conditions with the best possible contact between the ac-
celerometer and the vibrating steel block this resonance can be taken as a
practical upper limit. In practice the mounting will generally be less effective
and a lower resonance frequency is obtained.

Frequency Response Curve.

An individual frequency response curve is supplied with each accelerometer.
This is taken with the transducer mounted on a steel block of weight 180
grammes and with a plane mounting surface.
The instrumentation employed is similar to the set-up described in chapter
"Calibration" and is based upon the B & K Calibration Exciter Type 4290.
In actual practice such a frequency response is to be expected from the
accelerometer, provided the method of fixing to the vibrating specimen is
satisfactory. It is generally accepted that the piezoelectric transducer may be
used for frequencies up to about 1/3 of the resonance frequency shown by this
curve for less than 1 dB error, or to 1/5 of the resonance frequency for less
than 5 % error, but poor mounting methods will lower the resonance frequency
and therefore also lower the upper frequency limit of the operating range.
The error at any frequency can be easily calculated from the formula for the
relationship between the relative displacement of the mass spring system to
the displacement of the base, viz.

Relative Displacement of Mass-spring

Displacement of base

where
w = 2 n X forcing frequency and
wo = 2 n X undamped natural frequency.
From this equation it may be seen that at 1/3 the resonance frequency the error
will be 1 dB and at 1/5 the resonance frequency the error will be about o.5 dB.

Temperature Sensitivity.
Changes in accelerometer capacitance, charge sensitivity and voltage sensitiv-
ity with temperature are given for the range 20-250°C (70-480° F) with
reference to the values given at room temperature.
These are individual values obtained while heating the accelerometers to the
appropriate temperatures, reaching a steady temperature before measurement
is taken.
The information is given in curve form so that interpolation can be readily
performed.
When an accelerometer is used as a voltage generator, the transducer output
in mV/g will depend upon the temperature of the piezoelectric and on the
external loading. The calibration curve given is valid for the accelerometer
with 1.2 m cable and thus for a capacitive loading of about 105 pF. Any change

20
in loading capacity will change the voltage sensitivity variation with tempera-
ture. It may therefore be possible to optimize the voltage sensitivity versus
temperature characteristic.
The leakage resistance may also vary and alter the low frequency response.
This effect is, however, negligible for the B & K accelerometers, since the
leakage resistance is larger than 20ooo Mohm for all temperatures in the
operating range.
Charge amplifiers eliminate any influence of parallel capacity on the charge
sensitivity versus temperature curve. Only series capacity should be con-
sidered, if present.

Current
jtoG.t
jwGt
Cp Ct

Accelerometer Cable Preamplifier.

a) b)

Fig. 2.5.
a) Equivalent circuit tor accelerometer and preamplifier input.
b) The same ctrcuit simplified tor normal operating frequency range.

Considering the accelerometer as a charge generator it is easily seen from

Fig. 2.5 that the voltage sensitivity at any temperature with any amount of
cable capacity is

where
E, = voltage sensitivity at temperature t
Q, = charge sensitivity at temperature t
C, = total capacity in the circuit at temperature t

Now, the increase in charge sensitivity is given on the calibration chart In dB,
so that knowing the charge sensitivity at room temperature the charge sen-
sitivity at temperature t is easily found by using the conversion table for dB
to ratio given in the appendix.
Ct is also easily found when the capacity external to the accelerometer is
known. The increase in accelerometer capacity with temperature is given on
the calibration chart, as well as the capacity at room temperature.

21
Therefore
Ot =OoX~
Ct = C. X K.: + C.,.t - Cc
where
Oo = charge sensitivity at room temperature
~ = charge sensitivity factor .
c. = capacity of accelerometer plus associated cable
(1.2 m) at room temperature
Kc =capacity factor
C.. t = total capacity external to the accelerometer
(long cable plus preamplifier)
Cc =capacity of accelerometer cable used during factory
calibration (1.2 m)
Then we have

C. X Kc + C..t - Cc

5
dB
4

3
2

•---=•-_-
- - - - - - - -•::...:;:...::::• Capacitance
1.8 dB
_ •:::-:::-::::•--
0+--4~~:L~~---+-----+-----+--­
Charge Sensitivity
0.9dB
-o
- o - 0 -ovoltage Sensitivity
2

3
50 100 150 200

Fig. 2.6. Temperature sensitivity curves from the calibration chart on Fig. 2.1.

Example:
What is the voltage sensitivity of the accelerometer Type 4336, Serial No. 148396
(Fig. 2.1) at a temperature of 250°C and with a total external capacity (long
cable plus preamplifier) of 1ooo pF.

Solution:
Frol'n=the calibration chart we find
Oo = 1.16 pCoulomb/g
l<q =1d8=1.12
C. = 321 pF ,
K.: ~= "1.6 dB= 1.2
Cc , = 107 pF

22
Then the voltage sensitivity at 250°C
E250 1.16 X 1.12 I
0 001016
321 X 1.2 + 1000-107 = · Vg
E2so = 1.02 mV/g

Calibration Accuracy.
The accuracy of the above factory calibrations is better than ± 2 Ofo for charge
sensitivity, voltage sensitivity and capacity. The transverse sensitivity may
change temporarily when the accelerometer is exposed to large shocks,
especially sideways. Normally, however, it will return to the original value
within the next 24 hours.

23
 3. Mounting Methods
A proper mounting of the accelerometer to the specimen is of utmost im-
portance when measurements are taken, especially at higher frequencies. The
frequency response curve given on the calibration chart is for the best possible
mounting of the accelerometer. (Screwed tightly with a steel stud onto a
polished metal surface). When other methods are used the resonance fre-
quency will generally be lower. How much lower will be determined by the
mass of the accelerometer and the stiffness of the mounting.

Mounting Thread.
The mounting thread used for the Miniature Accelerometer Type 4336 is M3,
both for fixing the accelerometer to the specimen and for the cable con-
nection.
There are several methods available for fixing the B & K accelerometer to the
specimen on which the vibration is to be measured. The following accessories
for fixing purposes are included in the Accelerometer Set Type 4316:
Three M3 threaded studs and nuts
One cementing stud
Wax
One M3 screw tap
Allen key for steel studs.

Mounting of Accelerometer.
The possible ways for applying the accelerometer to the vibrating specimen
are depleted in Fig. 3.1.
Type 1 mounting is the best solution frequency response wise, approaching a
condition corresponding to the actual calibration curve supplied with the
accelerometer. If the mounting surface is not quite smooth it is a good idea to
apply a thin layer of silicon grease to the surface before screwing down the
accelerometer. This Increases the mounting stiffness.
Type 2 mounting Is convenient when a cementing technique Is desirable, while
the possibility of removing the accelerometer from the vibrating surface from
time to time is retained. This mounting can also be used to electrically
isolate the accelerometer from the vibrating body. The stud is cemented onto
the vibrating surface, possibly with a fine glass mesh in the cement to make
sure that the stud is not touching the surface, and the accelerometer is
screwed onto the threaded end of the stud.

24
 r
Steel Cementing
yr:tud
Cement
~··
Wax

CD 0 ® 26612?

Fig. 3.1. Different ways of mounting the accelerometers.

1. With steel stud.
2. With cementing stud.
3. Accelerometer stuck on with wax.

Type 3 mounting employs a thin layer of wax for sticking the accelerometer
onto the vibrating surface. The wax is delivered with the Accelerometer Set
Type 4316. Due to the low mass of the accelerometer this mounting method
may be used to quite high vibration levels, at ordinary room temperatures.
For minimum weight and optimum performance of the mounting one may also
recommend the Eastman 910 cement, marketed by the Armstrong Industry, or
Tixo K-1 manufactured by Tiox-Tinten und Klebstoffwerk G.m.b.H., Vienna.
Dental cement and epoxy resins are also very useful, especially In connection
with the cementing stud which is intended for use in applications where mount-
ing by cementing techniques is preferred, while retaining the possibility of
removing the accelerometer itself.
Soft setting glues or gum should be avoided because of decoupling and bad
frequency response.

With steel stud

11 " 1111 11 u., Cementing stud +wax
- -- - - Cementing stud + dental cement
· · · ·. · ·• · · •· • Cementing stud + Eastman 910

100 500 1000 2000 5000 10000 20000 50000

&66123

Fig. 3.2. Frequency response of an accelerometer for various types of

mountmg.

25
Fig. 3.2 shows some frequency response curves obtained for various types of
mounting. The importance of the mounting method used should be obvious.

Wax, tape, epoxy or other

type of clamping 2657f8

Fig. 3.3. Clamping of the cable to the vibrating specimen.

Mounting of Cable.
The connecting cable should always be given particular attention. Especially
at lower frequencies it may give rise to noise voltages, due to bending or
cable w~ip. The B & K accelerometer cable is designed and treated for noise-
less operation, however, it is always good policy to clamp the cables as firmly
as possible in order to avoid relative movement. See Fig. 3.3.

26
 4. Measuring Systems
An accelerometer always works in connection with amplifiers and indicating
instruments in order that the signal on the output terminals can be of some
use to the operator.
A vibration measuring system usually consists of the following parts:
Accelerometer
Preamplifier
Integrating networks
Amplifier
Filter network
Indicating device.
In the Bruel & Kjrer systems the amplifier, filter network and indicating device
are often contained in one instrument. A block diagram of possible instrument
combinations from the B & K program Is given in Fig . 4.1.
' '332, '333, '33,, ZR 0020 2612, 2613, 261, . 2801, 2802 2603, 2604. 2107.; 2305 "20
'335, '336, '3'0 2615,2622,2623, 2,112, 2203,2211,
2630, 1606,2616 2'09, 2'17., 2502

1612,1613, 1620

Fig. 4.1. Possible instrument combinations for vibration measurement and

analysis.

Preamplifiers.
Preamplifiers are used for conversion of the rather weak transducer signal Into
a stronger signal which can be handled by the succeeding storage or read-out
equipment. The signal from the piezo-electric accelerometer appears as a
voltage across a capacitive Impedance. The charge generated Is proportional
to the acceleration.
We have the choice of making the total capacity In the circuit as small as
possible and thus obtain the highest possible voltage into the preamplifier, or
to load the accelerometer so heavily with a shunt capacity that we have a
system independent of small changes in cable capacity due to different lengths
of cable. The first solution is called a voltage amplifier and the second a
charge amplifier.

27
Voltage Preamplifiers.
When used as a voltage source the accelerometer must be loaded by an
extremely high impedance in order to retain its sensitivity versus frequency
characteristic. Capacitive loading reduces the sensitivity over the whole fre-
quency range, while conductive loading reduces the sensitivity at low fre-
quencies. This can be seen from the following:
The equivalent circuit of an accelerometer with external loading is drawn
in Fig. 4.2.

c R=..L
G

266o33

Fig. 4.2. Equivalent circuit of accelerometer and preamplifier.

Q = charge induced across the accelerometer capacitive element

Sq = charge sensitivity of the accelerometer (Coulomb/g)
A = acceleration to which the accelerometer is subjected (d)
C = total capacity in the circuit, including accelerometer, cAble and
preamplifier
R = 1/G where G is the total conductance in the circuit, including
accelerometer, cable and preamplifier.
Assuming a sinusoidal acceleration of angular frequency w we Rave the
current flowing in the circuit
I =jwQ
and the voltage output
I jwQ
(1)
E = G+jwC G+jwC
This shows that when G ~ jwC i.e. when the shunt resistance in the circuit Is
very high or at high frequencies the output voltage depends only upon the
capacitive loading:
E =jwQ = _9_
jwC C ~· : .> '"

It is also seen that the output is directly proportional to 1/C. This must be
taken into account when long accelerometer cables are employed.
From equation (1) it can also be seen that when G ~ jwC, i.e. for low fre-
quencies or for low shunt resistance the output is frequency dependent:

E =jwQ =jwRQ
G

28
 1.0 0
~-o-1--""'
~
~ dB
0.9 ~
./
:-:: I-
_,
Relative input voltog~
0.8
to preamplifier OS
a function of v
/ -2

0.7
frequency and /_ -I-
-3

0.6
loading
1-
ll----
--
f---
f-,--

-
--
- -I-
--~-

--
1---- -

----1-
1- -4
-5
I)
0.5 -6
I
0.4
I
v
v R= preamplifier resistance
C= total capacity in
0.3

0.2 /
v circuit (preamplifier+
cable +accelerometer)

Ol ./
v
~

0.01 0.02 0.05 0.1 0.2 0.5 1 fRC 2

266002

Fig. 4.3. Chart for finding required input impedance when accelerometer
capacity and low frequency cut-off is given.

This means that the output falls off at the same rate as the frequency at the
low frequency end . The corner frequency where the output is 3 dB down is
where /G/ = /jwC/
1
I.e. fc = 2nRC
where fc is called the "cut-off frequency".
The internal resistance of the accelerometers is extremely high, always ex-
ceeding 20ooo Mohm at room temperature (""" 20°C). The resistance of the
piezoelectric material is lower at high temperatures but usually still higher
than 20ooo Mohm at 250°C. This large resistance gives a low frequency cut-off
value of o~04 Hz for the accelerometer alone and thus makes measurement
possible down to practically DC with specially designed amplifiers. Fig. 4.3
gives a chart for finding the required input resistance for a given lower limit-
ing frequency and a given accelerometer capacity.
Example:
Find the required input impedance for a 1 dB cut-off at 1 Hz using an
accelerometer with capacity 300 pF.

29
Solution:
A 1 dB cut-off is seen to give a value of about o.3 for fRC, I.e.
fRC = o.3
0.3 0.3
R = fC = 1 X300X10-12
R = 1ooo Mohm
The minimum acceptable input resistance of the preamplifier is 1ooo
Mohm .

Cable Capacitance Sensitivity Correction.

The capacity of the cable connecting the accelerometer to the preamplifier
will reduce the voltage sensitivity of the accelerometer. When the charge
sensitivity of the accelerometer is given, as is the case with all B & K accelero-
meters, the easiest way to obtain the voltage sensitivity for any particular
system is to divide the charge sensitivity by the total shunt capacity In the
input circuit, i.e. accelerometer capacity, cable capacity and preamplifier
input capacity. The capacity of the B & K accelerometer able is about 90 pF/m
(27 pF/ft).
Example:
An accelerometer has a charge sensitivity of 1.5 pCoulomb/g and a
capacity without cable of 200 pF. What is its voltage sensitivity with a
connection cable of 4 m length? Preamplifier input capacity negligible.
Solution:
Cable capacity 4 X 90 = 360 pF
Total capacity 360 + 200 = 560 pF

Voltage sensitivity ~:o =o.00268 V/g = 2.68 mV/g

Charge Amplifiers.
The charge amplifier is garnrng widespread acceptance mainly because of
its simplicity of operation. It eliminates the effect of shunt capacity in the
input circuit, so that the operator can work without attention to variable ac-
celerometer cable lengths. The only information required is the charge
sensitivity of the accelerometer.
An equivalent circuit diagram for accelerometer, cable and - charge pre-
amplifier is given in Fig. 4.4. By some calculation it is found that the output
voltage is

C.+Cc+C,- C, (A-1)
which means that as long as c, (A- 1) >
c.+ Cc + c, the output voltage is
- QA/C, (A -1) independent upon cable capacity.

30
 jwQ

265716

Fig. 4.4. Equivalent circuit diagram for accelerometer ,cable and charge
amplifier.

The charge sensitivity is given on the calibration chart for the 8 & K accelero-
meters, but when this is not given it can easily be calculated by multiplying
the voltage sensitivity with the total capacity (accelerometer plus cable) used
in the calibration.
Note that the charge sensitivity of piezoelectric accelerometers does in general
change with temperature. An individual charge sensitivity versus temperature
curve is therefore supplied with each 8 & K accelerometer, covering the range
20-250 o C (70-480 o F).
Although the charge amplifier ·has advantages for certain applications, the
following points should be considered:
The large amplification necessary for proper operation of the charge
amplifier requires more components than what is necessary for a voltage
amplifier. In practice this means higher costs and less reliable operation
in severe environments.
More serious Is the inferior rise time and recovery time obtained for most
charge amplifiers. This is esp_ecially important for shock-measurements.
One of the arguments in favour of the charge amplifier is the possibility
of using long cables between transducer and amplifier. This is, however,
just as possible with voltage amplifier, if the gain is high enough to offset
the loss in sensitivity, and in addition to this the voltage amplifier may
have a superior signal to noise ratio. Charge amplifiers are therefore not
recommended when low signals are to be measured.
One may conclude that voltage preamplifiers are to be preferred for the
majority of applications. For special installations where long cables are used
in different lengths, for example on vibration test sites or for measurements on
large:_structures, the charge amplifier would be a natural choice.

31
B & K Preamplifiers.
Bruel & Kjrer produce several types of preamplifier for use with the accelero-
meter in vibration measuring set-ups. These include:
1. Cathode followers (several types)
2. Preamplifier Type 1606
3. Preamplifier Type 2622
4. Battery Driven Preamplifier Type 2623
5. Battery Driven Preamplifier Type 2616
1. The microphone cathode followers fitted with input adaptors are excellent
preamplifiers for the accelerometers when the measurement Is one of
acceleration only. Type 2615 is recommended. It has an input impedance
of 200 Mohm in parallel with 3 pF which gives a low-frequency cut-off of
approximately 2.5 Hz. The low-frequency cut-off of the system will there-
fore usually be determined by the following instruments. The cathode
follower attenuation is approximately o.9 dB. The Integrator ZR 0020 may
be screwed directly onto the 1" cathode followers, so that measurement
of velocity and displacement may be carried out in addition to accelera-
tion. See description below.
2. The Preamplifier Type 1606 is especially designed for vibration measure-
ments. It contains integrating networks for measurement of velocity or
displacement as well as acceleration. The gain is variable from 0 to about
38 dB and the input impedance is 200 Mohm in parallel with some 50 pF.
The low-frequency limit is lower than 3 Hz for zero gain of the amplifier.
A small shaker table on the preamplifier may be used to calibrate the
whole vibration measuring set-up at a vibration level of 1 g peak ( = 980.6
cm/sec 2 peak). The preamplifier is powered from the microphone input
socket of one of the B & K microphone amplifiers or analyzers.
3. The Preamplifier Type 2622 is designed especially for use in vibration test
set-ups in connection with other B & K vibration generators and control
equipment, but may equally well be used on its own as an ordinary
accelerometer preamplifier. It is a two-channel instrument which can be
used either as a voltage or as a charge amplifier. Each channel has a
three-decade attenuator which is set to the voltage or charge sensitivity
of the accelerometer in use, within the range 10-100 mV/g or 10-100
pCoulomb/g. The output signal is then scaled to 10 mV/g.
The lower limiting frequency can be chosen as 2-5-10 or 50 Hz. Apart
from the two outputs there is a third output which can be automatically
switched between the two channels, used for example for changing
between two control accelerometers on a shaker table.
4. The Preamplifier Type 2623 is an impedance conversion device of an
extremely small and rugged construction. See Fig. 4.5.
An extremely high input impedance has been achieved through the use of
a "field effect" transistor in the input circuit. The preamplifier is designed

32
 Clamp

~ 4336 ~
10-32 NF
Stud

. 2623
Fig. 4.5. The Preamplifier Type 2623 clamped near to the accelerometer.

to withstand severe environmental conditions and its sensitivity to vibra-

tion is extremely small.

Special features:
Input impedance: Min. 2ooo Mohm at 25°C
Min. 200 Mohm at 100°C
Output resistance: 40ohm
Voltage gain : 0 ± o.OSdB
Frequency range: o.5-500ooo Hz
Noise: Max.15 p,V, 2-40ooo Hz for 1ooo pF across input
Output voltage: Max. 20 V peak-peak
Rise time: < 10 p,sec.
Recovery time: <50 p,sec.
Dimensions: 45 mm x 14 mm dia.
(1.77" x o.55" dia.)
Weight: 20.6 grammes
5. The Preamplifier Type 2616 is a battery driven three-stage transistorized
impedance conversion device, suitable for use with high impedance
transducers. Due to its small size it can be placed close to the accelero-
meter, thus minimizing sensitivity loss and noise from the connecting cable.
See Fig. 4.6.
The preamplifier is battery driven from six built-in mercury cells or
supplied from an external DC source, 6-35 volts.
Special features:
Input impedance: 1200 Mohm, 10 pF
Frequency range: o.5-500ooo Hz
Dynamic range: 96.5 dB
Attenuation: 0 or 40 dB (fixed)
Amplification: + 2 to -20 dB (continuously variable)
Output voltage: Max. 2.8 V peak-peak

33
 Output

4336 Preamplifier Type 2616

266/29

Fig. 4.6. The battery driven Preamplifier Type 2616.

Rise time: <10 flSeC.

Recovery time: <50 flSeC .

Integrating Networks.
Apart from the integrating networks contained in the Preamplifier Type 1606,
Bruel & Kjrer produce a separate unit called the Integrator ZR 0020.
The Integrator, containing two stages of integration, is designed for screwing
directly onto a B & K Precision Sound Level Meter Type 2203, effectively
converting this into a handy, portable vibration meter, capable of indicating
levels of acceleration, velocity and displacement when an acceleration pick-up
is employed as a vibration transducer. A slide rule is delivered with the
Integrator which may be set to the acceleration pick-up sensitivity and used
for direct conversion of dB-readings to units of vibration (metric and British).
Accelerometer sensitivities from 10 to 1ooo mV/g are covered.
The components of the RC integrating networks have been chosen to give a
low-frequency cut-off (-3 dB point) at about 5 Hz. This is sufficiently low,
since the Precision Sound Level Meter itself has a low-frequency cut-off of
about 5 Hz. The high-frequency limits are determined by the capacitive
coupling between input and output and are about 15 kHz for velocity and 4 kHz
for displacement measurements. These ranges are sufficiently large for the
majority of applications.
It is also quite possible to use the Integrator with the 8 & K 1" Cathode
Followers, whereby measurement and analysis of acceleration, velocity and
displacement may be carried out with any of the 8 & K analysing/indicating
instruments, i.e. 2107, 2112, 2603 or 2211.

Amplifiers and Indicating Instruments.

Bruel & Kjrer produce a number of indicating amplifiers for use in vibration
measuring set-ups. They include:

34
 1. Microphone Amplifier Type 2603, 2604
2. Frequency Analyzer Type 2107
3. Audio-Frequency Spectrometer Type 2112
4. Sound Level Meter Type 2203
5. Vacuum Tube Voltmeter Type 2409
6. Vacuum Tube Voltmeter Type 2417
7. Level Recorder Type 2305.
1. Microphone Amplifiers. These are convenient to use both with cathode
followers and with the preamplifiers. They measure overall vibration level,
but are equipped with filter input and output sockets for operation with
external filters in frequency analyzing set-ups. Frequency range 2603:
Flat to within ± o.5 dB from 2 to 40ooo Hz. 2604: Flat to within ± o.5 dB
from 10 to 200ooo Hz.
2. Frequency Analyzer Type 2107. This is a microphone amplifier similar to
Type 2603 but with a frequency selective filter network built-in. In vibra-
tion measurements it is an excellent means for analyzing the frequency
contents of the vibration signal. The analyzer is of the constant percentage
bandwidth type and the bandwidth is selectable from about 6 % to 29 %
of the tuned-in frequency. Using the 6 % bandwidth gives a clear indica-
tion of each frequency component of the vibration. An important advantage
is the possibility for driving the frequency tuning mechanism from the
Level Recorder Type 2305, to give automatic frequency spectrum analysis
curves on precalibrated paper, when the two instruments are synchronized.
Frequency range 2-40ooo Hz for overall measurements (frequency selective
part switched out), 20-20ooo Hz for frequency analysis.
3. Audio Frequency Spectrometer Type 2112. This is again a microphone
amplifier but with 1/3 and full octave, contiguous fixed filters built-in. The
frequency range covered is 22 to 45ooo Hz when used sele·ctively, 2 to
40ooo Hz when used for overall measurements. When used with the Level
Recorder Type 2305 the filters may be switched automatically from the
recorder and frequency spectra are recorded on preprinted, calibrated
paper as either third octave or octave analysis.
4. The Precision Sound Level Meter has its own cathode follower built-In
and may therefore be used with the accelerometer connected directly to
its input terminals. Using the Integrator ZR 0020 described above the
Precision Sound Level Meter is effectively converted into a light, port-
able vibration meter, for measurement of acceleration, velocity and dis-
placement. The Octave Filter Set Type 1613 may be attached to the
Sound Level Meter for a rough frequency analysis of the vibration. Fre-
quency range for overall measurements: 20 to 20ooo Hz, and for octave
analysis: 22.5 to 20ooo Hz.
5. Vacuum Tube Voltmeter Type 2409. This voltmeter is an excellent instru-
ment for overall vibration measurements, used with for example the Micro-

35
 phone Power Supply Type 2801 and a Cathode Follower Type 2615. The
frequency characteristic of the voltmeter is flat to within ± o.2 dB from
2 to 200ooo Hz with true rms, peak or arithmetic average indication of
the vibration signal. Using the Preamplifier Type 1606 one can measure
velocity and displacement as well as acceleration level.
The Vacuum Tube Voltmeter may also be used with an input directly from
the accelerometer, since the input impedance is as high as 10 Mohm in
parallel with 20 pF. This means that vibration measurements may be
carried out down to as low as 50 Hz employing the Voltmeter alone.
6. Random Noise Voltmeter Type 2417. This meter may be used as the above
Type 2409, but the meter circuit employs selectable time-constants from
o.3 to 100 seconds, which is used to obtain a stable reading when the
vibration is irregular or "random" at low frequencies. An output socket is
provided for use with the Level Recorder Type 2305. The output signal is
DC, proportional to the rms level of the vibration, and the Level Recorder
must be switched to DC operation. The frequency characteristic of the
voltmeter is flat to within ± o.2 dB from 2 to 20ooo Hz.
7. Level Recorder Type 2305. All the amplifying-indicating instruments de-
scribed above have an output terminal for feeding for example the B & K
Level Recorder Type 2305. This provides a permanent written record of
whatever is measured , as a function of time, or when the Recorder Is
coupled to a frequency analyzer, as a function of frequency. The Level
Recorder is in itself a very versatile measuring instrument, capable of
measuring rms, peak and average values of signals from 2 to 200ooo Hz
or DC signals. Its dynamic range is variable from 10 dB to 75 dB, and
the writing width is 50 or 100 mm.
The Statistical Distribution Analyzer Type 4420 may be coupled to the Level
Recorder to give a level/time distribution of the measured signal,
The sampling frequency is variable from o.1 to 10 seconds and the width
of the amplitude window is one tenth of the writing width of the Level
Recorder.
As the counters employed go up to 999,999, an analysis time of more than
24 hours is possible even with the highest sampling frequency.

Ground Loops.
In set-ups where ground loops are disturbing or where ground currents may
be carried in the shield of the transducer cable, thus superimposing a noise
voltage on the signal, one should first of all consider the use of isolated
mounting. This may be obtained with a cementing technique, either using
the cementing stud er cementing the accelerometer directly onto the vibrating
surface. If one of these methods cannot be employed, a battery driven system
may be the solution, or a battery driven preamplifier connected to AC driven
equipment via a balancing transformer such as the B & K Type Tl 0001. See
Fig. 4.7.

36
 Measuring Analysing Ampl.

•
:::j ::

..::~ -~i~'' .2. ..a..

@o o@

-~m !
Battery Driven Balancing
Accelerometer Preamplifier Transformer
~
4332- 35 2616 Tl 0001 2107
4336 2112
4340• etc.

Fig. 4.7. Use of Balancing Transformer to eliminate ground loops .

37
 5. Calibration
When absolute vibration level is measured it is necessary to know the absolute
sensitivity versus frequency characteristic of the measuring instrumentation.
As a rule piezoelectric accelerometers of careful design and manufacture are
extremely stable, but after exposure to high temperature or shock environ-
ments it may be wise to recalibrate.
Equipment for relatively quick and simple calibration of vibration pick-ups is
available from BrOel & Kjrer. This consists of a small electrodynamic shaker
Type 4290 with an accurately calibrated control accelerometer built in. Using
the control accelerometer in a feedback loop with a B & K feedback controlled
sweep oscillator the vibration of the shaker table is held constant in the
frequency range 100 to 30ooo Hz, and when an accelerometer is fixed onto
the shaker table its frequency response is easily measured, e.g. with a Level
Recorder Type 2305, or point by point. A suitable set-up is shown in Fig. 5.1.
The Level Recorder is used for obtaining a permanent record of the frequency
characteristic of the accelerometer. The absolute sensitivity of the accelero-
meter is found by comparing the output with the output from thi built-in

B.F.O.

266710

Fig. 5.1. Set-up suitable for the calibration of accelerometers. There are many
possible alternatives to the various instruments shown.

38
calibrated control accelerometer. Since the shaker table itself is part of the
control accelerometer, the vibration is exactly the same for the two accelero-
meters.
The comparison method gives an accuracy on the absolute calibration of better
than ± 1 dB.
When higher accuracy is needed, or when it appears necessary to check the
sensitivity of the control accelerometer, the same shaker may be used for
reciprocity calibration. The procedure is described fully in the instruction
manual for the Calibration Exciter Type 4290. The accuracy of this method
is better than ± o.5 dB with careful procedure.
An alternative method is to keep a "secondary standard" for reference calibra-
tion only. For this purpose any one of the accelerometers 4332-33- 34-35
may be used. If this is kept for reference only and is not exposed to extreme
mechanical shocks or temperatures, it will serve as a stable reference over
a long period of time. Comparison between the reference and an unknown
may easily be carried out on for example the 4290 shaker table. 2 % absolute
accuracy is obtainable this way.

39
 6. Instrument Combinations
The following are a few suggestions for instrument combinations which may
be used in some common accelerometer applications. It is in no way suggested
that this list covers more than just a fraction of the possibilities one might
meet in practice. Nor does it necessarily give the one and only solution to
the given problem. The selection of instrument combinations should always
be given careful attention with due regard to frequency response, dynamic
range, operating environment, accuracy, read-out device etc. It is hoped that
the foregoing chapters will be of some assistance to the person responsible
for such a selection.
Further information about the various B & K instruments is found in their
respective Instruction and Application books.

Portable Instrumentation.
The Precision Sound Level Meter Type 2203 is an excellent amplifier for vibra-
tion measurements when portable instrumentation is required. Used with the
Integrator ZR 0020 it can measure velocity and displacement in addition to
acceleration. An octave filter set Type 1613 can be joined to the Precision
Sound Level Meter for frequency analysis of the measured vibration quantity.
See Fig. 6.1.
The frequency range of this system is from about 10 Hz to 20 kHz.

J ZR0020 2203 1613

Accelero- Integrator Indica-ting Octave
me'ter Amplifier Filters
fl65709

Fig. 6.1. Portable, battery driven instrumentation for measurement and analysis
of vibration.

Such instrumentation is also well suited for collecting vibration data in the
field with the aid of a portable tape recorder. Any desired vibration signal may
then be recorded on the tape and brought back to the laboratory for subse-
quent analysis with more powerful instrumentation systems.

Frequency Analysis.
Frequency analysis of vibration may be carried out with the system depicted
in Fig. 6.2. The frequency Analyzer Type 2107 has a selectable bandwidth

40
 Battery

4336
Accelerometer Preamplifier

Fig. 6.2. Instrumentation suitable for frequency analysis of vibrations in the

range 20-20ooo Hz.
from about 6 Ofo to 29 Ofo and the centre frequency of the pass-band is con-
tinuously variable from 20 Hz to 20 kHz in six subranges. If a Level Recorder
Type 2305 is used for driving the tuning mechanism of the Frequency Analyzer,
the frequency spectrum can be recorded on preprinted, frequency and ampli-
tude calibrated paper.
Overall acceleration can be measured in the range 2 Hz to 40 kHz.
The Frequency Analyzer may be substituted by a Spectrometer Type 2112
which has contiguous 1/3 and 1/1 octave filters in the range 22 Hz to 45 kHz.

Shocks and Transients.

Theoretical analysis of shock waveforms suggests that in order to measure
a shock, a very wide frequency range is desirable. For a general pulse of
duration T the system time constant, RC, should be at least 20 T for less than
5 % error. The system time constant is made up of the preamplifier input
resistance, R, and the total capacity, C, in the input circuit, i.e. accelerometer,
cable, and preamplifier input capacity all added in parallel.
The high frequency limit of the measuring system, including the accelerometer,
should be at least 10/T Hz and the low frequency characteristics of the ampli-
fiers should be flat down to 8/1000 T Hz. See B & K Technical Review No. 3-
1966.
Phase Distortion.
In order to obtain an undistorted display it is also necessary that the various
frequency components are phase delayed by an amount proportional to the
frequency, i.e.
rp = k f
where rp is the phase distortion of the system at frequency f.
The system may be used for shock measurements when k = 0, i.e. no phase
distortion, or when k is some positive or negative constant.
In shock measurements extremely high signal peaks may be encountered, and
it is necessary to make absolutely certain that the dynamic capacity of the
measuring system is sufficiently large.

41
High shock pulses, resulting in saturation of the amplifiers, may cause block-
ing of the amplifiers, and a certain "recovery time" will elapse before the next
pulse can be measured correctly. This recovery time is to a certain extent
dependent upon the low frequency cut-off of the amplifiers, so that the lower
one's low frequency cut-off is, the more careful one must be not to overdrive
the amplifiers. For pulse train measurements this is especially important.
With its very wide frequency range and large dynamic range, the Miniature
Accelerometer is particularly well suited for shock measurements. Used to-
gether with one of the preamplifiers Type 2633 or 2616 the frequency range
is from o.5 Hz to 40 kHz, and the rise time and recovery time are 10 p,sec and
50 p,sec respectively. Rated conservatively such a system can be used for
measuring shock pulses varying in length from 25 p,sec to 16 msec. This
covers most practical cases.

DC Supply

4336 2623
Accelerometer Preamplifier Oscilloscope
zut3t

Fig. 6.3. Instrumentation suitable for shock measurements.

J
Another preamplifier well suited for shock measurements is the Cathode
Follower Type 2615 connected as shown in Fig. 6.4.
This instrument combination gives a very large dynamic range and an excellent
signal to noise ratio. The low-frequency cut-off is around 2.5 Hz (accelerometer
capacity 300 pF, cathode follower input resistance 200 Mohm). As no phase-
shift is introduced by the cathode follower the shock waveform is reproduced
faithfully on the oscilloscope screen.

0
4336 2615 2801
Accelerometer Cathode Power Oscilloscope
1!i!Ul3;1.
Follower Supply

Fig. 6.4. Cathode Follower Type 2615 used for shock measurements.

42
VIbration Tests. Fig. 6.5 shows a typical arrangement for vibration testing of
mechanical components. The component under test is mounted on an electro-
dynamic vibration exciter (shaker) which is fed from the Sine-Random Genera-
tor Type 1040 via a power amplifier. To maintain constant vibration level on
the shaker table, independent of frequency, the output from the Sine Random
Generator is regulated by the signal from an accelerometer (Type 4333)
mounted on the table. The Preamplifier Type 2622 is used in the feedback loop
to match the impedance of the accelerometer to that of the Vibration Meter
Type 2501, and to standardize the output signal from the accelerometer to
10 mV/g.
Miniature accelerometers may be placed at various points of the tested
structure for measurement of the response.

Sine Random
Generator

_!_ ;

·· :
• • 5i. 0 ; · ·~-
•0 •. ··:e:it>
- ·- ·--
:e: • •
"• , .• : ,. ,, • : :e:: -•... :. :e . ~Jie.
2122
Vibration Meter Preamplifier Level Recorder
266133

Fig. 6.5. Vibration test set-up.

Quality Control. In the last few years it has been realized that a vibration
measuring system may be used for quality control of such items as electric
motors, vacuum cleaners, refrigerator units etc. The Bruel & Kjaar Noise Limit
Indicator has been designed especially for production control of mass pro-
duced items with regard to their noise and vibration output. The instrument
has two alternative inputs for B & K Cathode Followers or Preamplifiers Type
1606. 12 parallel output amplifiers may be fitted with plug-in filters, e.g. 1/3
or 1/1 octave filters, and a control panel with red lamps will indicate when
the input signal level exceeds any preset value in each frequency band. On
the basis of noise or vibration investigations on satisfactory production items,
these limits may be set, and a quick production line control may be set up. In
case of a fault in the item being tested, the vibration may be high in one or
two particular frequency ranges. By looking at the frequency range in which

43
 00
''!!!
· =--
2211
ZH 0001
Noise Limit lndicatDr Control Panel
..26+S9f

Fig . 6.6. The Noise Limit Indicator Type 2211.

the preset level is exceeded, it is often possible to tell straight away what
the fault is, and it may be corrected without loss of time.
This method of quality control is covered more fully in B & K Technical Review
No. 2-1963, " Quality Control by Noise Analysis" . The Technical Review is
mailed free of charge upon request.

44
 7. What to Order
The Miniature Accelerometer Type 4336 is available as Accelerometer Set Type
4316 containing the following:
One accelerometer,
One low noise cable for ordinary temperatures,
One low noise cable for high temperatures,
Three threaded steel studs and nut (M3),
Cementing stud,
Wax,
Screw tap for 3 mm thread,
Allen key for steel studs,
Microplug to B & K plug adaptor,
Individual calibration chart.
See Fig. 7.1.

Fig. 7.1.

45
The Miniature Accelerometer is also available as Accelerometer Package
Type 4356 containing
Five Miniature Accelerometers
Five Low noise cables for high temperatures
Five Low noise cables for ordinary temperatures
Five Steel studs, M3
Five Individual calibration charts
Any extra accessories must be ordered separately.

Accessories.
The following accessories are available for mounting of the accelerometers
and the transistorized preamplifiers.

For 4336:
AO 0051 1.2 m (4 ft.) of mmmoise cable for operation to 100°C (212° F)
fitted with miniature plugs. Individually calibrated capacitance
value.
AO 0052 1.2 m (4 ft.) of mininoise cable for operation to 260° C (500° F)
fitted with miniature plugs. Individually calibrated capacitance
value.

For the transistorized preamplifiers Type 2616 and 2623:

UA 0142 1 set of clamping magnet containing 5 permanent magnets UA 0070

with isolated mounting. Used for Preamplifier Type 2623.
UA 0129 Set of 20 miniature plugs JP 0012 with tools and instruction for
mounting of the plugs on cable.
UA 0130 Set of 25 miniature plugs JP 0012.
AO 0037 1.2 m (4 ft.) of mininoise cable for operation to 100°C (212°F)
fitted with miniature plugs. Individually calibrated.
AO 0038 1.2 m (4 ft.) of mininoise cable for operation to 260° C (500° F)
fitted with miniature plugs. Individually calibrated.
AC 0010 Mininoise cable up to 600 ft. in one length.
90 pF/ m or 30 pF/ft., for operation to 100°C (212°F).
AC 0005 Mininoise cable up to 600 ft. in one length.
90 pF/ m or 30 pF/ft., for operation to 260°C (500° F).
JP 0028 Adaptor plug, microplug to B & K coaxial.

46
 8. Conversion Charts, Tables etc.
The following table is given in order to facilitate the conversion from dB to a
(voltage) ratio. It is used as follows:
Subtract a whole number of n X 20 from the dB value to be converted which
gives a positive remainder between 0 and 20. Look up the ratio in the table
corresponding to the remainder. The value sought is then 10" x value from the
table.
Example: Convert 65.3 dB re. 1 g into units of g.
65.3 = (3) X 20 + 5.3.
5.3 gives from table 1.841. The g-value is then 103 X 1.841 = 1841 g.
With negative values the procedure is the same, e.g.:
Convert -31.8 dB re. 1 g into units of g.
-31.8 = (- 2) X 20 + 8.2.
8.2 gives from table 2.570. The g-value is then 10- 2 X 2.570 = o.02570 g.

Table for Converting Decibels Into (Voltage) Ratio.

dB .0 .1 .2 .3 .4 1 .5 .6 .7 .8 .9
I I I I I I I
0 1.000 1.012 1.023 1.035 1.047 1.059 1.072 1.084 l 1.096 1.109
1 1.122 1.135 1.148 1.161 11.175 1.189 1.202 1.216 1.230 1.245
2 1.259 1.274 1.288 1.303 11.318 1.334 1.349 1.365 1.380 1.396
3 1.413 1.429 1.445 1.462 11.479 1.496 1.514 1.531 : 1.549 1.567
4 1.585 1.603 1.622 1.641 . 1.660 1.679 1.698 1.718 1.738 1.758
5 1.778 1.799 1.820 1.841 i 1.862 1.884 1.905 1.928 1.950 1.972
6 1.995 2.018 2.042 2.065 2.089 2.113 2.138 2.163 t2.188 2.213
7 2.239 12.265 2.291 2.317 12.344 2.371 2.399 12.427 2.455 2.483
8 2.512 2.541 2.570 2.600 12.630 2.661 2.692 2.723 ' 2.754 2.786
9 2.818 2.851 2.884 2.917 ' 2.951 2.985 3.020 3.055 3.090 3.126
10 3.162 3.199 3.236 3.273 3.311 3.350 3.388 3.428 3.467 3.508
11 3.548 3.589 13.631 3.673 3.715 3.758 3.802 3.846 13.890 3.936
12 3.981 4.027 4.074 4.121 4.169 4.217 4.266 4.315 4.365 4.416
13 4.467 4.519 4.571 4.624 : 4.677 4.732 4.786 4.842 4.898 4.955
14 5.012 5.070 5.129 5.188 5.248 5.309 5.370 5.433 5.495 5.559
15 5.623 5.689 5.754 5.821 5.888 5.957 6.026 6.095 6.166 6.237
16 6.310 6.383 6.457 6.531 6.607 6.683 6.761 6.839 6.918 6.998
17 7.079 7.161 7.244 7.328 7.413 7.499 7.586 7.674 7.762 7.852
18 7.943 8.035 8.128 8.222 8.318 8.414 j 8.511 8.610 8.710 8.810
19 8.913 9.016 9.120 9.226 9.333 19.441 • 9.550 9.661 9.772 9.886

47
Frequency, Acceleration, Velocity, Displacement Nomograph. (RMS-values).

100 1000 Hz(cfs) 10000

2657"4-

48
Temperature:

9 5
F= - C+32 C= - (F - 32}
5 9

Single Degree of Freedom System.

M = mass (kg}
K = stiffness (Newt/m)

wo = V~ = 2 :rr: X resonance frequency

_ 1/_g_ where Ll., = static deflection

26570~ Wo - v LJ st Of the maSS.

For Single Frequency (Sinusoidal) Vibration.

Acceleration
I Velocity

1
I Displacement

1
a cos wt - - a sin w t - -OJ2 acos wt
w
1
-w vsin wt v cos w t - - v sin w t
OJ

- w2 d cos w t - w d sin w t d cos w t

RMS Values.

A Alw Alw 2

wV v V/w

w2 D wD 0

49
Conversion of Length.

m em mm
I ft in

1 100 1000 3.281 39.37

0.01 1 10 0.0328 0.3937

0.001 0.1 1 0.00328 0.03937

0.3048 30.48 304.8 1 12

0 .0254 2.54 25.4 0.0833 1

Conversion of Acceleration.

g
I m/sec2
I cm/sec 2
I ft/sec2 in/sec2

1 9.81 981 32.2 386

0.102 1 100 3.281 39.37

0.00102 0.01 1 0.0328 0.3937

0 .03109 0.3048 30.48 1 12

0.00259 .).0254 2.54 0.0833 1

Conversion of Weight (Mass).

kg gram lbs. oz.

1 1000 2.205 35.3

0.001 1 0.0022 0.0353

o.4536 453.6 1 16

0.02835 28.35 0.0625 1

50
 9. Specifications
Accelerometer Type
I 4336

Contained in Accelerometer Set 4316

Contained in Accelerometer Package 4356

*Sensitivity (mV/g) 4-6

*Charge Sensitivity (pC/g) 1- 2
*Free Resonance (kHz (kc/s)) 125
--~ -
*Mounted Resonance (Solid Steel) 125
(kHz)
-- - ~
- ~

*Capacity lnclud. Cable (pF) * * 300

*Transverse Sensitivity (%) L: 8
0
Max. Ambient Temperature ( C) 260
-~
--
*Temperature Stability (dBr C) 0.02
---
Lowest Leak Resist. at 20 ° C (MQ) 20ooo
Typical value at 250 ° C (MQ) 20ooo
Magnetic Sensitivity (uV/ Gauss)
--- ~-~--

Acoustic Sensitivity (fl,VI fl bar)

~~
1
L: o.5
I
Torque Sensitiv., 2- 20 kgcm (%) L: 1
--
Max. shock (g's) 14ooo

* Freq. Range (Hz (c/s))

2% 2- 25000
10% 2--40000
Type of Connection Side
Height (mm) 9.2
Weight (grams) 2
~

Material of Base Titanium

Provision for Water Cooling No

Mounting thread
I M3

* Individual values given on the calibration chart.

• * With standard low-noise cable , 1.2 m (4 feet) long.

51
B K

November 1966
Orboe Print - Copenhagen