1
= R1 x C1 (Seconds) and
2
= R2 x C2 (Seconds)
Consider 3 possible relationships between
1
and
2
:
1.
1
<<
2
In this case, the lower cutoff (-3db) frequency for the system is:
.16
0
= -------- (Hz) (Eq. 1)
1
The lower -5% frequency is:
-5%
= 3 x
0
(Hz) (Eq. 2)
The sensor is controlling the low frequency entirely in this case.
2.
1
>>
2
In this case, the output load and coupling capacitor determine the
low frequency response as follows:
The lower cutoff frequency (-3db) is:
.16
0
= -------- (Hz) (Eq. 3)
2
The lower -5% frequency is:
-5%
= 3 x
0
(Hz) (Eq. 4)
3.
1
=
2
In this case, where
1
and
2
are equal or close in value, the
combined time constant,
3
= (
1
+
2
) / 2
The -6db frequency is:
.16
-6db
= ------ (Hz) (Eq. 5)
3
The -3db frequency is:
0
= 1.6 x
-6db
(Hz) (Eq. 6)
The -5% frequency is:
-5%
= 1.6 x
-6db
(Hz) (Eq. 7)
These values are approximate and are to be used as a guide only.
Getting the Most From the Low Frequency Response
of the
Accelerometer
To measure ultra low frequencies with a very long TC LIVM accelerometer
where the AC coupling TC of the power unit is the limiting factor, a DC coupled
LIVM power unit (Model 4115B) is available. This unit utilizes a direct-coupled
summing amplifier to null the DC bias of the accelerometer by summing an
equal absolute value negative DC voltage at the input stage. The result is a zero
DC voltage level at the output, achieved with no coupling capacitor.
Using this power unit, the accelerometer discharge time constant alone deter-
mines the low frequency response of the system in accordance with previously
mentioned equations 1 and 2.
High Frequency Response
Another important consideration in selecting an accelerometer may be its high
frequency response.
Figure 4: High Frequency Response Comparison
Figure 4 shows the typical high frequency characteristics of four Dytran
accelerometers. These curves illustrate the undamped 2nd order system
response characteristic of the accelerometers and the bar graphs illustrate the
comparative useful frequency range of each model, the comparison criterion
being the +5% deviation from the 100 Hz reference sensitivity.
The high frequency response of any accelerometer is sensitive to mounting
techniques and may be modified by any anomaly that reduces the mechanical
coupling between accelerometer and mounting surface such as the use of an
adhesive, magnetic or ground isolation base, dirty or non-flat mounting sur-
face and too thick glue lines in adhesive mount installations. Follow the
mounting instructions outlined in the manual supplied with each accelerome-
ter for best results.
Sensitivity Standardization
The reference sensitivity (mV/g) of all Dytran vibration accelerometers is mea-
sured at 100 Hz at an input amplitude of 1g, RMS unless otherwise specified.
This is measured by the back-to-back comparison method. The sensitivity of
shock accelerometers (such as series 3200B) is determined by a drop-shock
technique developed by Dytran. All calibrations are NIST traceable.
Standardized models are considered to be those models whose sensitivities
3200B
LOG f
3
2
0
0
B
3
0
3
0
B
3
0
1
0
B
3030B
3
1
0
0
B
M
A
G
N
I
T
U
D
E
3010B
3100B
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com
are specified to be within 2% of the nominal sensitivity value at 100 Hz.
Shock accelerometers, because of their nature, are not standardized. Consult
the product data sheet to determine which units have standardized sensitivity.
Piezodyne
tm
Technology
Dytran has perfected an advanced patented concept in LIVM technology that
increases the voltage output from piezo crystals using a feedback technique
with the standard unity gain IC LIVM amplifier. This concept, called
Piezodyne
tm
, (Patent no. 4,816,713) spawned a line of miniature, high sensitivi-
ty, high resolution accelerometers.
Because there is no gain amplifier used in Piezodyne, output noise does not
increase in proportion to the increase in output signal amplitude. The result is
a 6db improvement in signal-to-noise ratio and up to 8 times increase in sensi-
tivity.
RMS to Peak Conversion
The output voltage generated by an LIVM accelerometer has a direct correlation
with input acceleration. A 1g RMS sinusoidal input will produce a 1g RMS out-
put signal as illustrated in Figure 5. A 100 mV/g accelerometer (Model 3100B)
is used here as an example. Refer to figure 5.
Figure 5: Input/Output Waveforms
For sinusoidal vibration input, it is convenient to read the output with a true
RMS reading AC voltmeter. To convert this value to peak gs, simply multiply by
1.414. Example:
gs peak + 1.414 x gs RMS. and,
gs peak-to-peak = 2.828 x gs RMS
Shock Accelerometers
Shock accelerometers are designed to measure very rapidly changing high level
unidirectional transient acceleration inputs as might be generated by pyrotech-
nic devices, crash tests, impact tests, etc. They are characterized by small size,
high stiffness (for high natural frequency) and ruggedness. Model 3200B is one
such accelerometer.
The resonant frequency of series 3200B shock accelerometers is greater than
100 kHz resulting in excellent rise time and minimal ringing. These rugged 6
gram instruments feature integral 10-32 or 1/4-28 threaded integral mounting
studs (6 mm is also available) and hardened 17-4 steel housings. The sensing
element utilizes an exclusive 2-piece element base for stain isolation and high
natural frequency.
+g
0
1g RMS
t
t
Corresponding output voltage
signal from 3100B @ 100 mV/g
Input acceleration to 3100B
1g Peak
141.4 mV Peak
282.8 mV pk-pk
2.828 mV, pk-pk
-g
-v
0
+v
100 mV RMS
Input Acceleration to 3100B
Corresponding Output voltage
Signal from 3100B @ 100 mV/g
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com
Introduction to Charge Mode Accelerometers
Dytran charge mode accelerometers are designed to measure shock and vibra-
tion phenomena over a broad temperature range. These accelerometers, unlike
the Low Impedance Voltage Mode (LIVM) types, contain no built-in amplifiers.
Dytrans charge mode accelerometers utilize high sensitivity piezoceramic crys-
tals, of the lead zirconate titinate (PZT) family, to produce a relatively high
charge output in response to stress created by input vibration or shock acting
upon the seismic system.
Because of the high impedance level of the charge mode signal generated by
the crystals, a special type of amplifier, called a charge amplifier, is used to
extract the very high impedance electrostatic charge signal from the crystals.
The charge amplifier has the ability to convert the charge signal to a low
impedance voltage mode signal without modifying it.
Figure 1: Typical compression design, charge mode accelerometer
Figure 1 is a cross-section of a typical charge mode compression design
accelerometer, model 3100C6. The sensitivity is 100 pC/g (pC = pico coulomb
= 1 x 10
-12
Coulomb) and the useful frequency range is up to 5 kHz. The
3100C6 operates at temperatures up to +500F.
A heavy metal seismic mass is preloaded against the piezoceramic crystals with
an elastic preload screw. The mass converts the input acceleration into analo-
gous stress on the crystals producing an output charge signal in direct propor-
tion to instantaneous acceleration.
When to Use Charge Mode Accelerometers
The question may be asked, When should I consider using charge mode
accelerometers vs. LIVM types with built-in electronics?
Charge mode accelerometers should be considered:
1. when making measurements at temperatures above +250F, the maximum
temperature for most LIVM instruments,
2. when the versatility of the laboratory charge amplifier is desired for system
standardization, ranging, filtering, integrating for velocity and displacement,
etc. and,
3. when adding or replacing accelerometers where existing charge amplifiers
must be used for economic or other reasons.
Two System Concepts
Charge mode accelerometers may be combined with a variety of electronic
components to create two basic measurement system classifications:
1. The conventional charge mode system, and
2. The Hybrid system.
The conventional charge mode system utilizes a sophisticated laboratory
charge amplifier while the hybrid system features simple dedicated range
miniature in-line charge and voltage amplifiers operating in conjunction with
LIVM current source power units.
The Conventional Charge Mode System
The versatile laboratory charge amplifier is the main feature of the convention-
al charge mode system. This section will familiarize you with the theory, oper-
ating characteristics and features of the basic laboratory charge amplifier.
Figure 2: Elements of the conventional charge mode system
Figure 2 illustrates a laboratory charge amplifier, model 4165 in use with a
model 3100C6 charge mode accelerometer. Series 6019A low-noise coaxial
cable is used to minimize triboelectric noise generated by cable motion. This is
a very versatile system whose signal conditioning options include:
1. Standardization of system sensitivity
2. Full Scale range selection
3. Discharge time constant choices
4. Filter options
5. 0-10 VDC out for sinusoidal input
6. Overload indication
7. Instant system zeroing (or reset)
8. External calibrate signal insertion
9. Front panel meter for observation of DC level of output signal
PIEZO CRYSTALS
ELECTRODE
PROTECTIVE CAP
SEISMIC MASS
PRELOAD SCREW
10-32 ELECTRICAL
CONNECTOR
MOUNTING STUD
MODEL 6200
10-32 TAPPED HOLE
STRAIN ISOLATION
BASE
AC
DC
LIVM
CHG
1K
+10
DC VOLTS
-10 0
20
90
ELECTROSTATIC
CHARGE AMPLIFIER
AC VOLTS-RMS
2 1 0
O/L
3 0
80
0
10
RANGE
UNITS/VOLT
S
L
M
5 4 3
10
100
500
OUT
IN
10
7 0
1
FILTER
1K
100
200
20
2K
20K
ZERO
TC
PWR
6 0 4 0
10
50
50K
1.0 to 11.0
SENSOR
SENSITIVITY
pC or mV/UNIT
.1 to 1.1 5 0
CHARGE MODE
ACCELEROMETER
CHARGE AMPLIFIER 4165
METER
5K
10K
RESET
USA INSTRUMENTS, INC.
LOWNOISE CABLE
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com
Charge Amplifier, Basic Theory
A charge amplifier is a special high gain, high input impedance inverting volt-
age amplifier with capacitive feedback. The amplifier is usually an operational
amplifier (op-amp) with near infinite voltage gain.
Figure 3: The charge amplifier
Referring to Figure 3, the input charge q
in
is applied to the summing junction
(inverting input) of the charge amplifier and is distributed to the input capaci-
tance of the amplifier C
A
and the feedback capacitor C
f
. We may write the equa-
tion:
q
in
= q
A
+ q
f
Eq 1
Using the electrostatic equation q = Cv and substituting in equation 1:
q
in
= v
A
C
A
+ v
f
C
f
Eq 2
Using equation 2 and making the appropriate substitutions and solving for the
output voltage of the amplifier in terms of input charge, amplifier loop gain,
and input and feedback capacitance we have:
-q
in
-q
in
1
V
out
= = x Eq 3
C
A
/ A + C
f
(A + 1) C
f
(1 + 1 / A) 1 + C
A
/ C
f
(A + 1)
where A is the open loop gain of the op-amp.
Now, letting gain A approach infinity, we have:
-q
in
V
out
= Eq 4
C
f
This result (Eq 4) shows clearly that the transfer function (gain) of a charge
amplifier is a function only of the value of the feedback capacitor C
f
. Notice
that input capacitance C
A
has no effect on the sensitivity of the charge amplifi-
er. This means that cable capacitance, for example, has no effect on the sensi-
tivity, a significant find when switching cable lengths and types.
Adding Versatility to The Charge Amplifier
Standardization of system sensitivity, say to exactly 10.00 or 100.00 mV/g is
accomplished by adjusting the amount of feedback with a potentiometer as
shown. Standardization is often necessary because accelerometers are rarely
ever made to exact sensitivities. The use of a precision multi-turn potentiometer
with turns counting dial allows the standardization of the system sensitivity by
dialing in the accelerometer sensitivity.
Referring to Figure 3b, changing the range of the charge amplifier is accom-
plished by switching various values of feedback capacitor into the feedback
path. This is accomplished with a rotary switch which has maybe 10 values of
precision capacitor arrayed around it.
Again referring to Figure 3b, feedback resistor R
f
gives DC stability to the circuit
and establishes the discharge time constant (TC) of the amplifier thereby set-
ting the low frequency response of the amplifier. A momentary reset switch (S1)
discharges the residual charge in the feedback capacitor returning the system
output to zero.
The circuit shown in Figure 3b represents only the first stage of a rudimentary
charge amplifier. One or more stages of filtering, integration and other features
can be added all in one compact package.
The Hybrid System
A Dytran hybrid system combines charge mode accelerometers with miniature
in-line fixed sensitivity charge amplifiers. These charge amplifiers are powered
by standard 2-wire LIVM power units.
Figure 4: A hybrid system with in-line charge amplifier
The hybrid system, (refer to Figure 4) is ideal for field use because of its small
size and rugged construction of the miniature charge amplifiers, for example,
Models 4751B and 4505A. These amplifiers are powered by conventional LIVM
constant current power units and transmit the output signal over the same two
wires as do conventional LIVM systems. The power unit separates the signal
information from the DC bias of the amplifier and couples it to the readout
instrument. As with conventional charge mode systems, low noise coaxial cable
is used to couple the accelerometer to the charge amplifier to minimize tribo-
electric noise.
S1
Cf
+
(a)
-v
out
Standardization
Potentiometer
A
C
-A
-
f R
0
10
C
A
f
S1
+
(a)
-v
out
q
in
C
-A
q
in
Rf
-
OPEN SHORT
24
NORMAL
0 12
USA
INSTRUMENTS, INC.
4 1
2 3
PWR ON
INPUT
S/N XXXX
MOD 4705AXX
SENSOR BIAS VDC
4114B
CURRENT SOURCE
USA INSTRUMENTS, INC.
CHANNEL
MONITOR
CHARGE MODE ACCELEROMETER
SIG/PWR
LIVM POWER UNIT
GENERAL PURPOSE COAXIAL
CABLE
SERIES 4705A
IN-LINE CHARGE
AMPLIFIER
LOW NOISE COAXIAL CABLE
(a) (b)
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com
When to Use The Hybrid System
The hybrid system should be considered when:
1. The accelerometer is used to measure events at a temperature which is above
that recommended for LIVM instruments, i.e., instruments that have built-in
amplifiers and the environment is not favorable to laboratory charge ampli-
fiers.
2. System cost is an important factor. Per channel cost of the hybrid system is a
fraction of that for the conventional charge amplifier system.
3. LIVM current source power units are in-hand and must be utilized.
4. Ruggedness and small size of the measurement system is imperative.
5. A dedicated system without the versatility of a laboratory charge is sufficient
for the measurement task.
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com
An accelerometer is an instrument that senses the motion of a surface to which
it is attached, producing an electrical output signal precisely analogous to that
motion. The ability to couple motion, (in the form of vibration or shock), to
the accelerometer with high fidelity, is highly dependent upon the method of
mounting the instrument to the test surface. For best accuracy, it is important
that the mounting surface of the accelerometer be tightly coupled to the test
surface to ensure the duplication of motion, especially at higher frequencies.
Since various mounting methods may adversely affect accuracy, it is important
to understand the mechanics of mounting the accelerometer for best results.
Calibration
Throughout the article we will refer to back-to-back calibration at times. It
will be informative to explain what is meant by this and to show how this type
of calibration is performed at Dytran.
Figure 1: Back-to-back calibration set-up
Figure 1 illustrates the components of a simple accelerometer calibration sys-
tem utilizing the Dytran Model 3120BK back-to-back accelerometer calibration
system, a small electrodynamic shaker, a signal generator, a power amplifier
and the readout instruments.
To perform a calibration, the test instrument is attached to the top surface of
the back-to-back standard accelerometer, (model 3120B) using the method to
be used in the actual application, i.e., adhesive or stud mount. At each frequen-
cy of interest, the input amplitude (in gs RMS) is set precisely by the back-to-
back standard system and the corresponding output from the test system is
recorded. To learn more about this topic, refer to the article Back-to-Back
Accelerometer Calibration in this series.
For purpose of analysis, a piezoelectric accelerometer may be considered to be a
second order spring-mass system with essentially zero damping. (Refer to
Figure 2).
The spring (K) is the crystal stack and the mass (M) is the seismic mass that
stresses the crystals to produce an electrical output proportional to acceleration.
The dynamic characteristics of this system determine the frequency response of
the accelerometer.
Figure 2: The accelerometer as a spring-mass system.
Figure 2a illustrates the accelerometer. Its spring-mass analogy is Figure 2b
and Figure 2c is a typical frequency response plot for such a system. The plot is
obtained by graphing accelerometer output vs. frequency with input vibration
level held constant at each frequency setting. Every such system has a mounted
resonant (or natural) frequency, f
n
characterized by a very high peak of output
at resonance. The solution for the differential equation of motion yields the
definitive expression for the resonant frequency as follows:
(Eq 1)
where: f
n
= system natural frequency (Hz)
K = spring constant of the crystal stack (lbs/in)
M = mass of the seismic system (Slugs)
Examination of the response graph (Fig 2b) shows that the lower frequency
portion of the curve is sufficiently flat to provide a useable range up to approxi-
mately 1/3 of the resonant frequency. This will not be the case however, if, dur-
ing the mounting, other springs are inadvertently interposed between mating
surfaces creating secondary spring-mass systems with lower natural frequencies
than that of the accelerometer itself. The following section is an attempt to
explain how this can happen if care is not exercised during mounting of the
test accelerometer. We start by exploring the various mounting methods com-
monly used to mount accelerometers.
Stud Mounting
The preferred method of mounting an accelerometer to the test object is the
stud mount method. (See Figure 3). The stud may be integral, i.e., machined
as part of the accelerometer or it may be separate (removable). The stud mount
method yields the best results because when the instrument is installed in this
fashion, the accelerometer and the test surface are essentially fused together
by virtue of the high clamping force of the stud, ensuring the exact duplication
of motion of both bodies at all frequencies.
The inclusion of a thin layer of silicone grease between mating surfaces aids in
the fidelity of motion by filling in any voids due to slight imperfections in the
mounting surfaces.
Accelerometer Mounting Considerations
OSCILLOSCOPE
CH A CH B
10 0
USA INSTRUMENTS, INC.
20
OPEN NORMAL
TEST
SENSOR BIAS VDC
CURRENT
SOURCE
4119B
SHORT
STD
ON PWR
DIGITAL
FREQUENCY METER
USA
OPEN
FUNCTION
GENERATOR
24
SHORT
0
SENSOR BIAS VDC
NORMAL
INSTRUMENTS, INC.
12
TEST ACCEL.
POWER UNIT
CURRENT
SOURCE
4110C
ON PWR
TEST
ACCELEROMETER
MULTIMETER
MOD 4119B STD.
SYSTEM PWR.
UNIT
BK-BK STD.
ACCELEROMETER
10 TURN POT
ATTENUATOR
2-POSITION
SELECTOR SWITCH
VIBRATION EXCITER
(SHAKER TABLE)
POWER
AMPLIFIER
n
LOG f
a
f
n
K
1/3 f
M
(b)
M
A
G
N
I
T
U
D
E
(c)
QUARTZ CRYSTALS
(k)
SEISMIC MASS (M)
(a)
n
f =
1
2
K
M
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com
The Mounting Stud
The mounting stud itself is a very important factor of the performance of the
accelerometer. Most Dytran mounting studs are fabricated from heat treated
beryllium copper because of its high tensile strength and its low modulus of
elasticity. This means that the stud will be very strong and relatively elastic, a
perfect combination for the task of holding two surfaces together under a high
preload.
The collar that is machined into the stud (see Figure 3a) prevents the stud
from bottoming in either mounting hole. This ensures that the stud will be
centered between the two mounting holes so that both sides have adequate
thread engagement.
All Dytran accelerometers have a recess machined into the mounting surface to
accommodate this collar allowing both surfaces to be in intimate contact.
When installing the stud, it is best to first thread the stud into the accelerometer
to ensure that the stud enters the threaded port fully, then thread the
accelerometer into the mounting port until the surfaces meet and torque in
place.
In the design of miniature accelerometers such as the models 3030B, 3144A
and the 3200B et.al., interior space is at a premium and the only alternative for
stud mounting is the integral stud as shown in Figure 3b. This style of
accelerometer with reasonable care, will provide a long lifetime of normal
operation.
When Mating Surfaces Are Not Flat
As previously stressed, flatness of mating surfaces between accelerometer and
mounting surface, is of prime importance for best frequency response.. Here we
will examine the mechanics of a poor mount and its effect on frequency
response.
Figure 4: Non-flat accelerometer mounting surface
Figure 4a illustrates schematically, a condition where the accelerometer has
acquired a dished shape thru heavy usage. The mechanical analogy of this is
a leaf spring with spring rate K
m
as shown in Figure 4b. There are now two
spring-mass systems with this type of anomaly and both will affect frequency
response.
The new spring-mass system is formed by spring K
m
and the mass of the entire
accelerometer M
m
. The resonant frequency f
m
of this new system will most like-
ly be lower than that of the accelerometer and may affect the response curve as
illustrated in Figure 4c.
Even though the new resonant frequency is higher than the actual resonance of
the accelerometer, its effect will be to increase the output of the accelerometer
at the high frequency end of the accelerometer response.
Figure 3: The threaded stud mount
Two stud mount designs are illustrated in Figure 3, the separate stud in Figure
3a and the integral stud in Figure 3b. The separate stud style accelerometer is
the most popular for several reasons:
1. The removable stud allows easy access to the mounting surface of
the accelerometer for restoration of surface flatness should this become neces-
sary. Even with normal care, in time, after many installations, the mounting
surface of the accelerometer may become worn or damaged to a point where it
is no longer flat enough to affect a satisfactory mount and frequency response
will be compromised. It is a simple matter to restore flatness if the stud can be
removed and the accelerometer base can be applied directly to a lapping plate
for restoration of flatness. When the stud is integral and cannot be removed,
refurbishment of the mounting surface becomes very difficult and can only be
performed at the factory.
2. If the integral stud is broken or the threads become stripped or
otherwise damaged, the instrument may be essentially destroyed. On the other
hand, the separate stud can be easily replaced.
3. At times, with radial connector style accelerometers like the
model 3100B, it is important during installation, to orient the connector so that
nearby obstacles may be avoided. By exchanging mounting studs, the desired
orientation may be obtained.
4. The separate stud type accelerometer may be adhesive mounted
without using a mounting adapter, should this be desired.
M
m
K
m
ARTICLE VIII FIG 4
OCT 23 2000
(C)
(b)
M
n m f f
LOG f
M
A
G
N
I
T
U
D
E
(a)
K
LOWER SUB RESONANCE
ORIGINAL RESPONSE
NEW RESPONSE
ERROR DUE TO FAULTY SURFACE
ACCELEROMETER
(b) (a)
REMOVABLE STUD
TAPPED HOLE
INTEGRAL STUD
MOUNTING
SURFACE
(c)
(b) (a)
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com
We have chosen for purpose of this explanation, a hypothetical non-flat condi-
tion to illustrate the mechanics of response degradation. This analogy can be
extended to include other situations where mating surfaces are precluded from
intimate contact such as when foreign particles are entrapped between mating
surfaces or when other types of surface irregularities exist. The results of all
such imperfections will be more or less similar in nature to the example chosen
here.
Surface Preparation
It is difficult to overemphasize the importance of flatness of mating surfaces in
the mounting of piezoelectric accelerometers, especially with regard to frequen-
cy response. All Dytran accelerometer mounting surfaces are lapped optically
flat where possible or machined to very tight flatness tolerances. The test object
surface must be as carefully prepared. Although lapping is usually not possible,
other machining processes such as spotfacing, grinding, milling, turning, etc.,
can produce acceptably flat mounting surfaces (flat to .001 TIR).
After machining the surface and preparing the tapped mounting hole, clean
the area thoroughly with compressed air and a solvent to remove all traces of
metal chips, cutting oil, and any other surface contaminants. Before installing
the accelerometer, spread a light coating of silicone grease on either mating
surface. The grease will lubricate the surface and ensure intimate contact by
filling in tiny surface imperfections, maximizing high frequency transmissibili-
ty to the accelerometer.
Mounting Torque
Although every Dytran accelerometer is designed to minimize the effect of
mounting torque variations on sensitivity, it is good practice to set the torque
level, using a torque wrench, to the value recommended on the installation
drawing provided with the instruments. This will ensure that the instrument is
properly mounted and will preclude the expense and delays that may result
from overtorquing and breaking or stripping the threads of mounting studs.
This practice will also eliminate one of the main causes of calibration inaccu-
racy.
Adhesive Mounting
Situations often arise where the stud mount method is impractical, even impos-
sible, such as when mounting the accelerometer to thin sheet metal or to other
surfaces where drilling a mounting hole is not allowable. In such cases, an
adhesive mount installation can be the only practical way to install an
accelerometer.
Some accelerometers are designed to be adhesive mounted directly to the test
surface. (models 3115A, 3105A, 3053A, etc.). Others utilize mounting adapters
or bases for adhesive mounting. These adapters are normally first glued to the
test surface, then the accelerometers are stud mounted to them.
Figure 5: The adhesive mount, direct and with adapter
Figures 5a and 5b illustrate two adhesive mount installations, one direct mount
and the other with adhesive adapter. Figure 5c shows the undesirable thick glue
line and figure 5d illustrated the mechanical analogy of the thick glue line
mount. The thick layer of adhesive is actually a spring and has the effect of cre-
ating a new spring mass system as previously described in the section When
mating surfaces are not flat, with a similar result as shown in figure 4c.
To avoid the thick glue line, we recommend the use of a cyanoacrylate adhe-
sive, sometimes known as Instant Bond adhesives. These types of adhesives
are readily available and are recommended because:
1. They set very quickly,
2. Not much adhesive is required for a strong bond so glue lines will necessarily
be very thin,
3. Cleanup is easy because these types of adhesives are easily dissolved with ace-
tone.
Some users report good results with dental cement. Because of its high rigidity,
acceptable transmissibility can be obtained even with the slightly thicker glue
line that results. However, the problem with dental cement lies with its tenacity.
We know of no solvent that readily dissolves it so removal of the accelerometer
can result in damage to the instrument.
In conclusion, when using the adhesive mount method, expect problems at
high frequencies in direct relationship to the mass of the accelerometer. If pos-
sible, calibrate your accelerometer using a back-to-back accelerometer system
(such as the 3120BK calibration system) using the exact adhesive that will be
used in the actual test. In this manner, you can determine the precise behavior
of your measurement system at the expected frequencies.
Removal (unmounting) of Adhesive Accelerometers
Many accelerometers and adhesive adapters have been damaged or destroyed by
improper removal. The only sure way to avoid such damage is to torque the
accelerometer or adapter with a wrench using the flats provided. Adhesives are
generally weakest in shear strength and will yield under steady torque. Under
no circumstances should you strike an accelerometer or adapter to remove it.
The accelerometer would most likely sustain damage and may, at best, change
calibration after such trauma. All Dytran adhesive mount adapters have hex or
other flats to facilitate removal.
Electrical Isolation Bases
Isolation bases are used to electrically insulate the housing of an accelerometer
from the test surface. This may be necessary to avoid annoying ground loops
which can interfere with the measurement process when the test surface is an
elevated electrical potential.
Be aware of the fact that the use of any such base will effect the high frequency
response in the same manner as previously described in the section The
Adhesive Mount. Again, we recommend calibration with the actual adapter to
determine the effect on high frequency response.
The model 6220 is an example of a well-designed isolation adapter with some
exceptional features. The design incorporates stainless steel upper and lower
bases with an insulating anodized aluminum disc sandwiched between them
under high preload. The lower base has an integral threaded stud and the
upper has a 10-32 tapped hole. The upper and lower bases are interlocked
(d)
M
K
(c) (b) (a)
Thick Glue
Line Analogy
Thick Glue
Line
Direct
Adapter
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com
together to withstand high levels of mounting torque without damage. Both
upper and lower bases can be refinished to restore flatness without affecting
insulation.
Several anodized aluminum bases are also available for less demanding appli-
cations, (models 6226, 6244, 6245, and 6261 for example). These bases must
be handled carefully to avoid scratching the anodized surfaces that will com-
promise the insulating properties.
Magnetic Mounting Adapters
In general, magnetic adapters should be used with caution and rarely trusted at
frequencies above 1 kHz. Expect response degradation in direct proportion to
the weight of the accelerometer. There are some things the user can do to
ensure the best possible accuracy from the magnetic mount installation:
1. If possible, attach the magnet to a flat, bare, ferromagnetic metal surface
(See Figure 6a). A thick layer of paint on the test surface will lessen the holding
force of the magnet and could lower the effective high frequency response.
2. Clean the mounting area to remove any oil, grease and other foreign matter
which could preclude the intimate contact necessary to ensure a strong mag-
netic bond.
3. Select a flat area if possible, to achieve maximum surface contact. Avoid situ-
ations as illustrated in Figure 6b.
4. Attach the magnet to the test surface CAREFULLY. Remember that the pull of
a magnet rises sharply just before contact with the ferromagnetic surface and
this force could pull the assembly from your grip resulting in a very severe
metal-to-metal impact. This could overrange the accelerometer beyond its
maximum shock range and permanently damage it.
If possible, calibrate the accelerometer/magnet assembly by use of the back-to-
back calibration method.
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com
Mounting Wax
Mounting wax is very convenient to use but we do not recommend this method
as a viable means of mounting an accelerometer. It should only be used when
no other alternatives are feasible. The inconsistency in thickness and the low
modulus (rigidity) of wax make the results unreliable at higher frequencies. As
previously mentioned, calibration with the exact wax to be used will give the
best indication of the expected results.
Figure 6: Magnetic mounting adapters
Magnetic mounting adapters are used to attach accelerometers to ferromagnet-
ic surfaces such as machinery and structures where the instrument is to be
moved quickly from place to place. The accelerometer is attached to the mag-
netic adapter (usually by stud mount) and the assembly is applied to the test
surface. While this method is certainly convenient, the user may be misled by
this convenience.
(a) (b)
Magnetic
Adapter
(a) (b)
To calibrate a vibration accelerometer is to accurately determine its sensitivity
(in mV/g or pC/g) at various frequencies of interest. The ISA approved back-to-
back comparison method is probably the most convenient and least expensive
technique.
At Dytran, back-to-back calibration involves coupling the test accelerometer
directly to a (NIST) traceable double-ended calibration standard accelerometer
and driving the coupled pair with a vibration exciter at various frequencies and
acceleration (g) levels. The assumption here is that since the accelerometers
are tightly coupled together, both will experience exactly the same motion, thus
the calibration of the back-to-back standard accelerometer can be precisely
transferred to the test accelerometer.
The Dytran model 3120BK vibration calibration system used in conjunction
with a small electrodynamic shaker, a signal generator, a frequency meter and
several other pieces of equipment provides an inexpensive means to set up a
calibration facility. The 3120BK may also be used with more sophisticated com-
puter driven automatic calibration systems.
The 3120BK Back-to-Back Calibration System
The model 3120BK vibration calibration system consists of a double ended cali-
bration accelerometer, (model 3120B), a standardization amplifier, (model
4119B), and the necessary interconnect cables and accessories. (See figure 1).
Figure 1: Model 3120BK system
Model 3120B Back-to-Back Standard Accelerometer
Figure 2 is a representative cross section of the model 3120B back-to-back stan-
dard accelerometer. This type of accelerometer is also known as a double
ended standard because of its two mounting surfaces. The lower surface
attaches to the shake table armature and the test accelerometer is attached to
the upper surface.
Figure 2: Model 3120B back-to-back calibration accelerometer
The quartz shear seismic element in the 3120B is mounted directly to the
underside of the upper mounting surface to position it in closest possible prox-
imity to the unit under test. This location ensures the tightest possible coupling
to the test accelerometer. The excellent strain isolation of the quartz shear ele-
ment serves to minimize the effect of the mass of the test accelerometer on the
sensitivity of the standard. Subsequent sections of this article will address this
phenomenon, known as mass loading.
Within the 3120B, the electrical output of the self generating quartz shear seis-
mic element is connected directly to the input of an integral IC impedance con-
verting amplifier. (See the article Introduction to LIVM Accelerometers for a
complete treatment of the Dytran internal amplifier concept). This amplifier
buffers the signal making it impervious to outside interference and to cable
generated noise.
The electrical connector of model 3120B is the convenient 10-32 coaxial type
which has become the industry standard.
Model 4119B Standardization Amplifier
The line-powered model 4119B supplies constant current power to operate the
IC amplifier in the 3120B and standardizes the system sensitivity to precisely
10.00 mV/g at 100 Hz. It also provides the necessary low-pass filtering to sup-
press the rising high frequency characteristic of the 3120B to provide flat fre-
quency response to 10 kHz. (See Figure 3).
Figure 3: Block Diagram Model 4119B amplifier
Back-to-Back Accelerometer Calibration
4119B
SENSOR BIAS VDC
0
SHORT
10.00 mV/g OUTPUT
CURRENT SOURCE
NORMAL
24
OPEN
USA
12
ON
INSTRUMENTS, INC.
PWR
6010A10 COAXIAL
CABLE
6200 MTG. STUD
(TYP)
3120B BACK-TO-BACK STANDARD
ACCELEROMETER
MODEL 4119B
STANDARDIZATION
AMPLIFIER
6020A05 COAXIAL
OUTPUT CABLE
IC AMPLIFIER
10-32 COAXIAL
CONNECTOR
TEST UNIT MOUNTING
SURFACE
10-32 TAPPED HOLE
TOP AND BOTTOM
QUARTZ SHEAR MODE
SEISMIC ELEMENT
VIBRATION TABLE
MOUNTING SURFACE
10.00 mV/g
OUTPUT
CURRENT
LIMITING
DIODE
INPUT FROM
3120B
A
-
+
FILTER STAGE
STANDARDIZATION
STAGE
A
-
+
C
+20
VDC
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com
The constant current source, a 2 mA current limiting diode, is powered by an
internal 20 VDC power supply. A coupling capacitor C blocks the DC bias volt-
age which exists on the 3120B line, and connects the vibration signal (AC) to
the input of the standardization stage.
This variable gain stage adjusts the system sensitivity to exactly 10.00 mV/g at
the 100 Hz reference frequency. The next stage of the 4119B is a second order
Butterworth low-pass filter with adjustable frequency characteristics. This filter
is adjusted to exactly match the high frequency characteristics of the 3120B.
The rolloff characteristics of the 4119B cancel the rising characteristics of the
3120B at higher frequencies.
Performing the Calibration
Assemble the system elements as shown in Figure 4. Couple the test accelerom-
eter to the top surface of the 3120B. By setting the vibration frequency and the
amplitude (using the output of the 3120BK system) a frequency response curve
may be plotted for the test accelerometer. At each frequency, set the amplitude
(in RMS gs) and read the corresponding amplitude from the test accelerometer
(in RMS mV).
Figure 4: The complete calibration system
Mass Loading Compensation
It is appropriate at this time to discuss a very important but little emphasized
phenomenon associated with back-to-back accelerometers known as mass
loading.
The 3120B is initially calibrated with a single ended accelerometer, model
3010B, which has been calibrated by an NIST certified calibration station. This
accelerometer weighs 19 grams. When accelerometers (or velocity pickups),
weighing considerable more that 19 grams, are placed atop the 3120B, the
increased inertial loading due to the increase in mass, actually changes the
effective sensitivity of the 3120B inserting a small calibration error. This error
increases at higher frequencies.
These errors are negligible when calibrating units weighing up to 30 grams but
over this weight, correction curves should be constructed to compensate for this
effect. The following section shows how to establish the compensation curve for
test units of varying weights.
Mass Loading Compensation Curves
The mass loading effect is frequency dependent as illustrated in Figure 5a. This
figure shows a typical family of correction curves as plotted with various masses
atop the 3120B.
To plot a mass loading correction curve for a model 3120BK system, proceed as
follows:
1. Select a single-ended accelerometer to use as a transfer standard (preferably
a model 3010B). Weigh the instrument precisely and record this weight, in
grams.
2. Attach this accelerometer to the 3120B and determine its sensitivity at all fre-
quencies of interest using the 3120B as the standard. Record the sensitivity at
each frequency.
3. Weigh the new instrument to be calibrated. (If it weighs less than 30 grams,
you do not need mass loading correction.
4. If it weighs more (say 50 grams) subtract the weight of the transfer standard
from 50 grams and record. This is the needed weight of the compensation
mass.
5. Calculate the dimensions of a steel (or tungsten) cylinder required to equal
the result of step 4 and fabricate a compensation mass as shown in Figure 5b.
Note: It is important that the mating surfaces of the compensation
mass be very flat (optical flatness is preferred). This degree of flatness is best
obtained by a lapping process. Dytran has the equipment and skills to produce
compensation masses at reasonable cost.
6. Attach the transfer accelerometer and compensation mass together as shown
in Figure 5b, placing a light coating of silicone grease between all mating sur-
faces. Torque in place.
Figure 5: Typical correction curves and a compensation weight
OSCILLOSCOPE
CH A CH B
10 0
USA INSTRUMENTS, INC.
20
OPEN NORMAL
TEST
SENSOR BIAS VDC
CURRENT SOURCE
4119B
SHORT
STD
ON PWR
DIGITAL
FREQUENCY METER
USA
OPEN
FUNCTION
GENERATOR
24
SHORT
0
SENSOR BIAS VDC
NORMAL
INSTRUMENTS, INC.
12
TEST ACCEL.
POWER UNIT
CURRENT SOURCE
4110C
ON PWR
TEST
ACCELEROMETER
MULTIMETER
MOD 4119B STD.
SYSTEM PWR.
UNIT
3120B BK-BK STD.
ACCELEROMETER
10 TURN POT
ATTENUATOR
2-POSITION
SELECTOR SWITCH
VIBRATION EXCITER
(SHAKER TABLE)
POWER
AMPLIFIER
W
E
I
G
H
T
G
R
A
M
S
150
75
50
19
10,000
9.00
9.50
10.00
S
E
N
S
I
T
I
V
I
T
Y
m
V
/
g
100 10 1
LOG f
(a)
1000
COMPENSATION
MASS
TRANSFER STD.
ACCELEROMETER
TEST
ACCELEROMETER
MODEL 3120B
SHAKE TABLE
ARMATURE
(b)
EQUIVALENT
WEIGHTS
(c)
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com
7. Using the sensitivity of the transfer standard (obtained in step 2) to deter-
mine the amplitude at each frequency point, determine the loaded sensitivity of
the back-to-back standard accelerometer and record each of these values. These
new sensitivity values plotted against frequency represent the correction curve
for that particular mass of test instrument.
8. Mount the test instrument atop the 3120B, as shown in Figure 5c and, using
the new values obtained in step 7, proceed to calibrate the test instrument by
setting the amplitude at each frequency using the corrected output from the
3120BK system and reading the corresponding output from the test instrument
at each frequency point.
NOTE: When using an RMS reading voltmeter to read amplitude
values, you may convert to equivalent peak g levels by multiplying the RMS
values by 1.414. This will only be necessary when calibrating velocity or dis-
placement pickups.
Figure 6: System dimensions
MODEL 3120B
6200 MOUNTING
STUD (REF)
1.07
1.79
USA
PWR
CURRENT SOURCE
NORMAL SHORT
INSTRUMENTS, INC.
4119B
OPEN
SENSOR BIAS VDC
10 20 0
ON
(b)
MODEL 4119B
7.75
5.50
.83
1.00
.625
(a)
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com
The LIVM Sensor/Power Unit System
In order to fully understand the purpose of current source power units, it will
be necessary to first understand the operation of the low impedance voltage
mode (LIVM) sensors with which the power units are designed to operate.
Most Dytran sensors utilize a common drain unity gain MOSFET integrated cir-
cuit (IC) amplifier to reduce the impedance level of the self generating piezo-
electric crystals by 10 orders of magnitude (see Figure 1). Some Dytran sensors
have voltage gain ICs, others utilize miniature charge amplifiers. All operate
on the same two-wire principle. All of the amplifiers are located within the sen-
sor housing.
Figure 1: Unity gain LIVM sensor and power unit schematic diagram
The amplifier circuit illustrated in Figure 1 is called a source follower since the
source voltage essentially follows the gate voltage dynamically. This is a unity
gain amplifier.
The source load R
S
is remote, i.e., it is located in the power unit and not within
the sensor housing. It is apparent upon examination of this circuit that the
power to the sensor amplifier and the signal from the sensor amplifier are car-
ried over the same two wire conductor. This fact makes possible the convenient
two conductor operation which characterizes LIVM operation.
The voltage gain of the sensor amplifier shown in Figure 1 is:
V
out
Y
fs
G = ------------- = -------------- Eq. 1
V
IN
Y
fs
+ 1/R
S
Where:
Y
fs
= Forward transadmittance of the common drain FET. (Mhos)
R
S
= Dynamic impedance of the source element. (Ohms)
Using 40,000 Mhos for Y
fs
and 100,000 Ohms for R
S
and solving for voltage
gain g, we have:
40,000 x 10
-6
G = ---------------------------------------- = .99 Eq. 2
40,000 x 10
-6
+ 1/ 100,000
The result of equation 2 shows that the gain of the follower circuit is very close
to unity and indicates that the value of source resistor R
S
must be necessarily
very high to keep the gain close to unity. Yet this element (shown as a resistor)
must be able to pass enough drive current to give the system the ability to drive
the shunt capacitance of long cables when necessary. These demands pose a
dilemma for the source element RS.
The Current Source
The constant current diode possesses the necessary qualities to satisfy the
requirements of the source element. A drive current level to 20 mA and a
dynamic impedance of 100k Ohms.
At Dytran, we utilize two types of constant current devices . (see Figure 2)
A 2 mA JFET constant current diode (Figure 2a) is used in our battery operated
power units to keep power drain to a minimum to prolong battery life.
In our line-operated power units, we utilize a clamped base bi-polar transistor
constant current circuit. (see Figure 2b) The value of emitter resistor R
e
sets the
value of constant current. Dytran line operated power units with the variable
drive current feature utilize a multi-turn potentiometer for R
e
to allow adjust-
ment of sensor drive current over the range of 2 to 20 mA.
Figure 2: Two Constant Current Elements
The DC Power Source
In the portable Dytran power units, the energy to operate the LIVM sensor
amplifier is supplied by two 9-Volt dry cell batteries. In the line-powered units,
a well regulated DC power supply is utilized.
Several of our rack mountable multi-channel power units such as the 4116,
4120 and 4121 have provision for mains power and external battery power.
Terminals are provided on the rear panel to allow the connection of batteries in
field situations where AC power is not available or when lowest sensor noise
level is desired.
Introduction To Current Source Power Units
R
S
COUPLING CAP
Vout
+
in
S
G
PIEZO
ELEMENT
SENSOR
R V
POWER UNIT
CABLE
CAPACITANCE
DC POWER
SOURCE
CABLE
IC AMPLIFIER
in
V D out
ARTICLE X FIG 2
OCT 31, 00
PHYSICAL DIODE
SCHEMATIC OF JFET
CIRCUIT
SCHEMATIC SYMBOL
OF CURRENT DIODE
(a)
(b)
R
e
i
CONSTANT
CURRENT OUT
THE CONSTANT CURRENT DIODE
A CONSTANT CURRENT CIRCUIT
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com
Sensor Bias Voltage
When the LIVM sensor is supplied with constant current from a current source
type power unit in the prescribed range of 2 to 20 mA, and +18 to +30 VDC
compliance level, the sensor amplifier will turn on at a bias level established
by the specific amplifier used in that unit. This bias level can range from +8
Volts to +13 Volts DC. To the power unit, the sensor amplifier looks like a zener
diode, i.e., the bias voltage is constant, unaffected by the current level applied
once the diode knee has been exceeded.
Figure 3: Bias Voltage Levels in the LIVM System
Referring to Figure 3, the signal from the piezo element applied to the gate of
the FET is at zero Volts (ground) quiescent level. At the source terminal of the
FET, the bias voltage is at the previously mentioned turn on level of +8 to
+13 Volts DC. To restore the output voltage level to zero volts quiescent, a sim-
ple DC blocking coupling capacitor C
c
, (usually 10F) is used in most Dytran
power units to couple the signal to the readout instrument.
This capacitor and the 1 Megohm pulldown resistor in parallel with the readout
instrument input impedance constitute a high-pass filter which may limit the
low frequency response of the measurement system. Refer to the section
System Low Frequency Response in the article Introduction to LIVM
Accelerometers for a more complete discussion of this topic.
Output Voltage Swing
Most LIVM sensors are ranged for a full scale output voltage swing of 5 Volts.
An overrange capability exists (the point where the signal clips) in most instru-
ments and it is informative for the user to know where this point is in both the
positive and the negative directions.
The maximum output voltage swing, without clipping of the signal, is con-
trolled by two parameters:
1. Power supply (compliance) voltage.
2. Sensor bias voltage.
Note: The exact bias voltage is reported on the calibration sheet supplied with
each LIVM instrument.
Figure 4: Output signal waveform clipping
Figure 4 illustrates the three different situations of output signal status as seen
at the source terminal of the LIVM amplifier using a hypothetical sensor whose
bias voltage we have set at +10 VDC. The waveform illustrated is typical of that
generated by an accelerometer subjected to a sinusoidal input stimulus.
The first waveform (Fig. 4a) represents a normal full scale output signal at 5
Volt level with no clipping of the signal.
The second waveform (Fig 4b) illustrates the overrange limit when using a
battery powered current source power unit such as the Dytran model 4102B
which has a +18 Volt supply voltage level. As seen in this example, overrange is
limited on the positive swing where the signal clips at the +18 Volt level, 8 Volts
above the +10 Volt bias level. The negative side remains unclipped since the
signal peak amplitude is well below the +10 Volt swing available before clip-
ping.
The third waveform illustrates the result on voltage clipping when using a line
operated power unit like the Dytran models 4110C and 4114B and others, with
+24 Volt compliance. This shift changes the overrange on the positive side to
+24 minus the +10 Volt bias or 14 Volts before clipping occurs.
Please note that in this illustration, we used a sensor with a bias voltage of +10
Volts. Dytran produces sensors with many different bias voltages. The overrange
capability of each specific model will vary dependent upon the specific sensor
bias chosen. Now you have the tools to accurately determine these overrange
points.
Sensor Drive Current Explained
The IC amplifier in most standard Dytran LIVM sensors can operate normally
with drive currents anywhere within the range of 2 to 20 mA. Battery operated
power units such as models 4120B, 4103B and 4105B utilize 2 mA constant
current diodes as the current source element. Line-operated power units such as
the 4110C, 4114B, 4120, 4121, 4116, etc., use variable constant current circuits
(as described in Figure 2b) to allow setting the sensor drive current within the
range of 2 to 20 mA.
A logical question that may come to mind is: Why do you fix drive current in
some power units and make it variable in others? Another question often
asked is: Why do you offer variable drive current and what level of current
should I use in my application if it is variable?
To answer the first question, we us a fixed 2 mA constant current diode in the
battery powered units simply to conserve battery life. For most laboratory or
field use when relatively short cables are in use, 2 mA of drive current is suffi-
cient to give satisfactory results while keeping battery life high.
The answer to the second question is a bit more complicated and the next sec-
tion is an attempt to answer it.
Driving Long Cables
Referring back to Figure 1, it can be seen that cable capacitance
appears directly across the source terminal of the LIVM amplifier.
Cable capacitance can become a significant factor when long cables
are being used. This capacitance loads the amplifier and can cause
signal distortion especially at higher frequencies. (See Figure 5a) This
type of distortion is caused by slew rate limiting as seen in opera-
tional amplifier circuits when this specification is exceeded.
0
-V
+V
INPUT RESISTANCE
OF READOUT
INSTRUMENT
+
cc
0
0
+V
-V
-V
+V
+10V BIAS VOLTAGE
in R
G
S
D
10V
t
(c) SIGNAL CLIPPING
SENSOR BIAS
(a)
0
14V
8V
(b)
10V
S
U
P
P
L
Y
V
O
L
T
A
G
E
+5V
+24V
+18V
+15V
+10V
SIGNAL CLIPPING
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com
Figure 5: Slew rate limiting as seen in LIVM sensors
Figure 5 illustrates two manifestations of this effect when insufficient drive cur-
rent is supplied to the LIVM amplifier to charge the cable capacitance at a high
enough rate. Figure 5a illustrates the type of distortion as it might appear on a
sinusoidal waveform.
Figure 5b shows a result of insufficient drive current as it might affect the result
of a waveform that might be generated by a shock tube wavefront acting on a
LIVM pressure transducer.
The solution to these problems is to increase the sensor drive current sufficient-
ly to eliminate the distortion. The instruction manuals supplied with the power
units shows how to adjust the sensor drive current.
The distortion illustrated in Figure 5 is a function of several different system
parameters. These are:
1. Cable length as previously discussed. (This determines capacitive load).
2. Desired maximum output voltage swing. (This is determined by the
sensor sensitivity and the amplitude of the measurand).
3. The needed system high frequency response. (This is set by the rise
time or highest frequency parameters).
The interaction of drive current on these three parameters is illustrated in the
following chart, Figure 6. This chart will be helpful as a general guide to
choosing optimum drive current for your specific application and to illustrate
limitations in the system performance for various situations.
Figure 6 gives the sinusoidal frequency response of a typical LIVM sensor with
three different lengths of cable, four different drive current settings and two dif-
ferent full scale output voltage swings. The effects of these parameters can be
seen vs. drive current.
Please note that this chart represents the capability of the LIVM amplifier alone
and does not show the limitations which may be imposed by the particular sen-
sor element. Overall performance will likely be limited by other factors inherent
in the design of the particular sensor. For example, pressure sensors in general,
will have higher frequency capability than most accelerometers, etc.
Consult the specification sheet in the manual for the performance characteris-
tics of your particular sensor.
The chart, Fig. 6 is based on a cable capacitance of 30 pF/ft, the approximate
capacitance of RG58/U coaxial cable. Adjust chart values accordingly for cables
with differing capacitances per ft.
Figure 6: Cable Driving Parameters vs. Sensor Drive Current
NOTE: The combined lengths of input (sensor to power unit) and output
(power unit to readout) cables must be used when entering the chart for
unbuffered (passive) power units such as the 4102B, 4110C, 4121, 4116, etc.
Only input cable lengths need be considered for the buffered units such as
models 4105B, 4115B, 4125B, 4126, 4112B, 4119B, etc. These units have op-
amp circuits which can drive long output cables independent of sensor drive
current. Consult the sales department if you need help in determining which
units are buffered.
Why Not to Use Maximum Allowable Drive Current
Unless Necessary
The factory setting for all Dytran line operated power units with variable drive
current is 5 mA. There are several reasons why we do not set this current to 20
mA (the maximum allowable for LIVM units) and why you should increase this
current only when necessary.
The integrated circuit (IC) amplifier used in most LIVM sensors is necessarily
very small and its heat dissipation properties are limited. Low values of drive
current keep junction temperatures low, minimizing thermal stress and pro-
longing amplifier life. This is especially important when the sensor is used at
elevated temperatures close to the +250 maximum allowable temperature.
Another factor associated with drive current is that the background noise level
of the IC amplifier is lowest when drive current is low. Therefore, if maximum
(b)
SENSOR (OUTPUT)
WAVEFORM
SENSOR (OUTPUT)
WAVEFORM
INPUT WAVEFORM
INPUT
WAVEFORM
(a)
1000
100
10
500 kHz
700 kHz
25 kHz
80 kHz
500 kHz
Drive
Current
Cable
Length
@30 pF/ft
8 kHz
600 kHz
150 kHz
300 kHz
1300 kHz
40 kHz
70 kHz
16 kHz
200 kHz
300 kHz
900 kHz
150 kHz
100 kHz
1V Ft. mA
2
5
10
20
10
100
1000
1000
100
10
1000
100
10
Frequency Response 5%
Output Signal Amplitude
5V
20 kHz
10 kHz
50 kHz
5 kHz
1.7 kHz
50 kHz
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com
resolution is important, it is best to operate with the lowest possible sensor drive
current.
Another reason to keep drive current low is to minimize transient thermal
effects during very low frequency measurements. In such situations, high ther-
mal dissipation from the IC amplifier, due to high drive current, can alter the
thermal equilibrium within the sensor and may cause annoying baseline shift
when the orientation of the sensor changes. This could show up as a drift on
the output signal.
See the section Extending The Low Frequency Response in the article
Introduction to LIVM Accelerometers for more information on the topic of
low frequency response.
The Fault Meter as a Troubleshooting Tool
Most of Dytrans current source power units incorporate a front-panel mounted
voltmeter which monitors the sensor bias voltage. On battery powered units, the
meter serves as a battery voltage monitor also.
Figure 7: Various fault meters
Figure 7a illustrates, schematically, where the meter is located in the power
unit. Figure 7b shows the low current DArsonval meter used in the battery
powered units and Figure 7c illustrates the 10-segment bar type LED voltmeter
used on most line-powered units. Although they look different, both meters
serve the same basic function.
These meters are handy system troubleshooting tools and they serve as confi-
dence builders for LIVM systems, especially for first time users of such systems.
Normal operation of sensor, power unit and cables is indicated by an approxi-
mate mid-scale reading on the meter (in the NORMAL area of the meter
scale). The reading is approximate because the normal tolerance on the bias of
standard LIVM sensors is from +9 to +12 Volts. The actual indication will thus
vary from sensor to sensor. This is normal.
If a short circuit exists across the SENSOR terminal, a zero Volts reading will
result. The meter will read in the SHORT section of the meter face.
NOTE: If a SHORT indication exists, do not worry about the power unit burn-
ing up. The SHORT current is limited by the setting of the constant current
source and will not exceed 20 mA.
By systematically disconnecting the cables, first from the sensor to the power
unit, then from the output, the fault can be located. If the fault disappears
when the 10-32 sensor cable is disconnected, look for slivers of metal which
can peel off of cable connectors and short across the cable ends. A small brush
will be helpful in removing these slivers of metal and correcting the short con-
dition.
If an OPEN (full scale) reading is noted, check for an open sensor cable.
Disconnect the sensor from the cable and leave the cable attached to the power
unit. Now short across the cable end with a metal object and observe the meter.
If the cable is good, the meter will go to SHORT when the cable end is short-
ed. If the meter does not go to SHORT when the cable end is shorted, the
cable is faulty and should be replaced. When a SHORT or OPEN indication
exists after the cable has proved to be intact, the problem may lie in the sensor
itself. Faulty sensors are not field repairable and must be returned to the factory
for evaluation and possible repair or replacement.
With no sensor attached to the power unit, the meter will read in the OPEN
area (full scale on the meter), verifying that the power source is functioning
and set at the proper compliance voltage.
With the battery powered units, the meter will indicate when it is time to
replace the batteries. When the BATT TEST momentary pushbutton is
depressed and held, this measures battery voltage and it should indicate above
the BATT OK line on the meter scale if the batteries are serviceable.
The Zero Clamp (DC Restorer)
When measuring relative unidirectional pulses such as may be gener-
ated by impact forces from a punch press, it is often desirable or even
necessary to maintain the baseline (zero) reference voltage during the
test. The Dytran zero clamp (standard on the model 4110C only and
available on special order on some other models) will retain the base-
line voltage during such testing.
Figure 8: Baseline shift and the clamped waveform
Figure 8a indicates what occurs to the baseline of the output waveform from
the typical LIVM AC coupled power unit under conditions as described here.
Starting at time t
0
, the pulse train baseline drifts slowly downward until the
area enclosed by the waveform above the zero line will exactly equal the area
below the baseline. This is the normal quiescent condition of a typical AC cou-
pled system. The zero baseline is lost.
Figure 8b illustrates the results under the same set of conditions as in Figure 8a
but using the Dytran zero clamp feature.
Dytran has designed an adjustable zero clamp circuit with a potentiometer to
make up for variations in the readout instrument characteristics and precisely
zero the baseline. Consult the catalog specification information covering LIVM
power units or contact the factory to determine which model power units (other
than the 4110C) are available with the zero clamp option.
SENSOR
MONITOR
METER
COUPLING CAP.
+
R
c
OUTPUT
OPEN
BATT.
O.K.
SHORT
10
NORMAL
(a)
20 0
SENSOR BIAS
DC VOLTS
NORMAL SHORT
24
(c)
(b)
12 0
OPEN
SENSOR BIAS VDC
-V
+V
0 t
(a)
+V
0 t
(b)
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com
Introduction
A transfer function of a mechanical system describes its dynamic behavior in
response to an applied stimulus (excitation). There are six major transfer func-
tions used in mechanical systems analysis.
These are:
Dynamic Mass (Force/Acceleration)
Mechanical Impedance (Force/Velocity)
Dynamic Stiffness (Force/Displacement)
Compliance (Displacement/Force)
Accelerance (Acceleration/Force)
(Note that each of these transfer functions is itself a function of Force.)
Each transfer function describes the dynamic behavior of the mechanical sys-
tem in a different fashion and each involves the ratioing of the output
response to the input stimulus. Response can be measured most conveniently
with accelerometers placed at important points throughout the structure under
test. Spectrum analyzers perform the necessary mathematical ratioing on input
and response signals to produce, almost instantaneously the desired transfer
function.
The input stimulus (forcing function) may be applied to the structure by vari-
ous methods. One common way to excite structures is by use of an electrody-
namic or hydraulic shaker. A force sensor attached to the armature driving the
test object may be used to define input force amplitude and a signal generator
controls frequency. When the test object is too large to be excited in this fash-
ion, or when it is physically impossible to do so, there is another method to
consider.
An alternative method available to excite a structure without the complication
and expense of a shaker system, involves striking it with a calibrated dynamic
impulse hammer. A force sensor mounted in the head of the hammer trans-
forms the input force pulse into an analogous waveform that contains the nec-
essary amplitude and phase information to completely describe the forcing
function. Impact tip material stiffness helps determine the frequency content of
the input forcing function by controlling the impact pulse duration. By defin-
ing the frequency and amplitude of the forcing function, impulse hammers
present a fast, simple way to excite structures in a well defined fashion.
Spectrum analyzers in conjunction with hammer systems (with accelerome-
ters) can instantaneously plot transfer functions greatly simplifying testing and
saving valuable man hours. A branch of mechanical system analysis called
Modal Analysis, uses software to further refine the frequency response data to
describe in detail each mode (resonance and anti-resonance) of the structure.
The Impulse Hammer
The dynamic impulse hammer has two primary functions:
1. It must excite the test structure with constant force over the frequency
range of interest and,
2. It must produce an analog voltage pulse which is an exact representa-
tion of the input impulse (Ft) both in amplitude and phase.
A piezoelectric impulse hammer consists of a head assembly containing a force
sensor and a handle with rubber grip, interchangeable head extenders and sev-
eral different impact tips (steel, aluminum, hard plastic and soft plastic).
Figure 1 illustrates the construction of one of several models of Dynapulse
hammers produced by Dytran, the model series 5800B. The two other
Dynapulse hammers are the series 5801B and the model 5850B.
Figure 1: Impulse Hammer Structure
Dytran Dynapulse
note 180
phase shift
from displ.
note 90
phase lead
dv
dt
a = ---- = - A sin t
2
x = A sin t
A
A
A = Peak Displacement
v = ---- = A cos t
dx
dt
ACCELERATION (a)
VELOCITY (v)
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com
Acceleration Compensation
A design incorporating components within the sensor to cancel the effect of
motion (vibration) on the sensor output signal.
Acceleration Sensitivity
In a pressure transducer or impulse hammer, this refers to the unwanted out-
put signal from the sensor in response to vibration in the sensitive axis of the
sensor. This parameter is specified in terms of psi/g for pressure sensors and
Lbs/g for impulse hammers. Certain Dytran sensors such as the Series 2200 and
2300 pressure sensors and the Dynapulse series of impulse hammers are
acceleration compensated to minimize this effect.
Charge Mode
Sensors which contain piezoelectric crystals but no built-in electronics. These
sensors may use quartz or piezoceramics to generate electrostatic charge sig-
nals in response to input measurands.
Current Source Power Unit
A power supply expressly designed for use with Low Impedance Voltage Mode
(LIVM) sensors consisting of an energy source (batteries or DC power supplies)
and a constat current element. These units are characterized by stable constant
current output and high dynamic impedance. They also serve as signal decou-
pling devices.
Discharge Time Constant (TC)
The time required for the output voltage from a sensor or system to discharge
to 37% of its original value in response to a zero rise time step function input.
This parameter determines low frequency response.
Fault Monitor Meter
A DC Voltmeter incorporated into LIVM power units to read the DC voltage at
the output of the current source. If a sensor is connected, the meter reads the
sensor bias voltage and can be used to indicate open and shorted cables and
sensors or to verify normal system operation.
Frequency Response
The highest and lowest frequencies at which measured deviations from a refer-
ence sensitivity (usually 100 Hz) lie within a specified error. The deviations are
usually specified as -5%, but in some cased -3dB is specified.
LIVM
Low Impedance Voltage Mode, Dytrans Trademark describing its line of piezo-
electric sensors with built-in impedance converting electronics.
Linearity
Actually the non-linearity or deviation from a straight line in a plot of output
amplitude vs. input amplitude of a sensor or system. At Dytran, we use the pop-
ular zero based best straight line method of determining linearity. Full scale is
determined, a zero based best straight line plot of output vs. input is made, an
error band is created using a specified percent of full scale. All points on the
curve must fall within this error band.
Mass Loading
The change in actual sensitivity of a back-to-back accelerometer when loaded
with masses significantly different from the transfer standard with which it was
originally calibrated. Correction curves of actual sensitivity vs. frequency are
referred to as mass loading correction curves.
Normalization
See Standardization.
Phase Shift
The difference in phase angle between input measured and output electrical
signal, measured in degrees. The response may lag or lead the input.
Picocoulomb
A measure of electrostatic charge output from piezoelectric sensors.
1 Picocoulomb= 1 x 10
-12
Coulomb.
RMS
Root Mean Square. In AC theory, this is the heating value of an AC wave-
form and it is equal in amplitude to .707 times the peak value of the waveform.
Most AC meters are calibrated to read in RMS. To obtain the peak value of the
waveform, multiply the RMS value by 1.414.
Resonant Frequency
Also called natural frequency. The lowest frequency of a second order system
(spring-mass system) which satisfies the solution to the differential equation of
motion.
At this frequency, the amplitude increases by factors of as much as 100. The
useable upper frequency of a sensor is determined by this parameter. All sensors
are spring-mass systems with intrinsic natural frequencies
Rise Time
The time required for a sensor or system to respond to an instantaneous input
step function, measured from the 10% to 90% points on the response wave-
forms.
Sensitivity
The scale factor of a sensor or system, measured in terms of change in output
signal per change in input measured. Accelerometer sensitivity is expressed in
terms of mV or pC/g, pressure sensor sensitivity is expressed in terms of mV or
pC/psi and force sensor sensitivity is expressed in terms of mV or pC/LbF.
Sensor Bias
When LIVM sensors are supplied with constant current, a DC voltage bias exists
at the center conductor of the sensor connector. The output signal is superim-
posed on this DC bias. The power unit extracts the signal from this bias by vari-
ous means, effectively blocking it. The normal range for this voltage is +9 to
+12 volts.
Standardization
As applied to systems, refers to the application of gain or attenuation as needed
to correct the sensitivity of a sensor to the exact nominal (or design) sensitivity.
As applied to sensors, refers to the process by which the sensor sensitivity is
brought close to the nominal value. In specification sheets, it is the highest
acceptable deviation of the measured sensitivity from the nominal value,
expressed in percent.
Transient Thermal Response
The time history of the change in output (voltage or charge) from a sensor
resulting from a sudden change in ambient temperature.
Transverse Sensitivity
The output of an accelerometer in response to motion in directions orthogonal
to its sensitive axis, expressed as a ratio of measured output to cross axis input,
in percent.
Thermal Coefficient of Sensitivity
The measured change in the sensitivity (of scale factor) of a sensor, from its
room temperature (reference) value to the value at a higher or lower stabilized
ambient temperature. This parameter is measured in percent of change in sen-
sitivity per degree of temperature deviation.
Triboelectric Noise
The unwanted generation of electrical charges by the chafing together of insu-
lation layers inside electrical cables during flexing caused by vibration and
shock induced motion.
Zero Shift
The change in baseline level of the output voltage of an accelerometer immedi-
ately after a mechanical shock.
GLOSSARY OF TERMS
n
f =
1
2
K
M
k
21592 Marilla Street, Chatsworth, California 91311 Phone: 818.700.7818 Fax: 818.700.7880
www.dytran.com For permission to reprint this content, please contact info@dytran.com