Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
I~ESCP~[PTION~
INæECTICN OF SOIIDER JOINTS BY
ACOUSTIC IMPEDANCE
Background of the Invention
In my U.S. patent, 4,21~,922, I described two methods for
detecting flaws or unacceptable conditions in solder joints wherein
an acoustic signal, with or without an electrical current, was
applied on one side of a solder joint/ passed through the joint, and
received on the other side. The input was m~dulated by the joint
and a ccmparison between the m~dulated output and the input was used
to identify bad solder joints.
It occurs m~ny times that a discrete element or integrated
circuit is soldered by its lead wires to a conductor pad on a
printed circuit board wherein the "other side" of the solder joint
is inaccessible. For example, this is the case where the leadi is
soldered to a conductor padi which passes through the substrate to
connect elements on the other sidie. Therefore, it would be kenefi-
cial to have a non-destruictive testing methodiwhereby the test could
be run on the solder pad itself Common ultrasonic flaw detectors
which apply an acoustic wave to the joint and receive back an echo
fron the flaw are good for some flaws but not for othiers which are
more subtle.
Summ~y of the Inventlon
It is an object of the invention to provide a non-diestruct-
ive, qualitative method of detecting unsatisfactory soldbr joints in
electrical devi oe s.
.~ , .
. . . ~
-- 2 --
It is also an object of the invention to
provide a method which is quick and reliable and
which can be easily automated.
It is also an object to provide such a
5 method which can be used with a wide variety of flaws
and unsatisfactory bonding.
It is an object to provide an inspection
method which can be applied to but one lead of the
solder joint or merely to the joint itself.
In accordance with the objectives, the
invention is an acoustic method for flaw detection
in solder joints. ~he method utilizes the
characteristic acoustic impedance of the solder
joint which modulates an applied acoustic wave and
15 is observed by a receiving transducer. The
electrical impedance of the receiver as affected
by the acoustic impedance of the solder joint is
monitored over a range of applied frequencies giving
a description of the impedance as a function of
20 frequency. This impedance/frequency description
is also known herein as the spectral response and
is a graphic representation of the condition of the
solder joint.
The method for inspection of defects in
25 a solder joint comprises applying an acoustic
vibration to the solder joint, sweeping over a
frequency range which preferably falls within the
limits of 20 hz to 1 Mhz, or more preferably
sweeping over a range of 150 khz-650 khz. The
30 frequency range should include at least one, and
preferably more than one, natural resonant
frequency of the solder joint. The acoustic
vibrations pass through or are reflected back from
the solder joint and are modulated thereby. The
35 vibrations are then received by a receiving
~ 5~3~6
transducer which is capable of converting the
vibrations into electrical signals. The electrical
impedance of the receiver (which is affected by
the acoustic impedance of the solder joint) is then
5 observed as a function of applied frequency (the
spectral response) and compared to standard
spectral responses of known unacceptable or
acceptable solder joints whereby the latter are used
to characterize the tested solder joint as
10 unacceptable or acceptable.
The acoustic driver applying the acoustic
energy and the receiver may be the same device in
which case the reflected impedance will be observed.
The receiver may alternatively be a second
15 transducer located either near the driver or on the
other side of the solder joint. Preferably the
receiver and the driver are separate elements but
are both contacted directly with the solder joint
as far as possible apart during inspection.
20 Detailed Description of the Invention
In Figure 1, an insulating substrate 1 is
shown with a conductive feed-through 4 connecting
metal conductor layers 9 and 10 on opposite sides
of the substrate. Wires 2 and 3 are respectively
25 soldered to the metal conductor layers 9 and 10 by
solder joints 6 and 5.
Acoustic transducer or driver 7 is used to
apply an acoustic vibra~ion to the solder joint 6
to test whether joint 6 is acceptable. The
30 transducer is connected to a conventional acoustic
generator ~not shown). The transducer may be
piezoelectric, magnetostrictive, or electromagnetic.
These are preferably used for inspection at high,
intermediate, and low frequencies, respectively.
~ 55~5~6
The invention may be used over a wide range of
frequencies which is dependent mainly on the type
and sensitivity of the solder joint and wire leads.
Miniaturized, sensitive joints and leads
5 may dictate the use of frequencies in the 1 r~hz
range, whereas ordinary, heavier-wire joints may
dictate the use of from about 20 hz-200 khz.
Frequencies swept over a range of 150-650 khz were
found most useful in the course of experimenting
10 with the invention.
Returning to Fi~ure 1, acoustic driver 7
is shown at the toe end of the lead wire 2 and
acoustic receiving means 8 is shown near position A
at the heel end thereof. This is the preferred
15 location of the driver and receiver in order to make
more reliable comparisons of one joint wi~h others.
Coupling of the receiver to the solder joint
constitutes an acoustic impedance and should therefore
be kept constant from sample to sample in order
20 that it can be ignored in the comparison. Constant
loading force should be applied to the receiver at
the location at A which is the most stable on the
joint. Other positions for the receiver, such as B
or even C and D, may be used if necessary,
~S however, they are not preferred. Positions C and D
would also involve interference of the impedance of
solder joint 5 into the independent observation of
the impedance of solder joint 6.
The coupling of the receiver to the joint
30 should be under modest pressure. ~f the driver has
a high impedance compared to the joint, defects will
be difficult to detect in a tightly coupled system.
On the contrary, if coupling is too light, only a
small amount of energy can be transferred to the
35 joint and sensitivity will be low.
~ .
~595~;
The receiver 8 may be a conventional
transducer pickup which is responsive to acoustic
impedance and broad-band input. A high-resolution
spectrum analyzer may be used to display or record
5 impedance data for manual inspection or to feed data
to processing means for automated operation. The
receiver should be responsive to input frequencies
over the 20 hz-l ~lhz range~
Description of the Preferred Embodiments
.
Printed circuit boards, each comprising a
substrate and 20, 14-lead flat packs (integrated
circuit elements) were prepared incorporating a
variety of known defects such as inclusions, cold
solder, voids, dewetting, poor filleting, excessive
15 solder, insufficient solder, gold entrapment, bent
lead, cracks, etc. Good solder joints were also
prepared as standards. A setup such as shown in
Figure 1 was used to apply vibrations to the
individual solder joints over a sweep of frequencies
20 in the range of 150 khz-650 khz. The receiver and
driver were located as shown in Figure 1. The
spectral responses obtained over the 150 khz-650 khz
sweep (see Figures 2-4), show that major resonant
frequencies in the transmitter were present at about
25 200 khz, 270 khz, and 520 khz~
The equipment used for testing were:
(1) A Wavetek 114 Sweep frequency generator
to drive the acoustic transducer through a
range of frequencies;
(2) A Tektronix 434 Storage oscilloscope for
monitoring the voltage to the driver.
Voltage was about 1 volt peak to-peak to
keep the stress on the solder to less than
1 psi;
` ' : ''' ~
: .
~5~
(3) A Tektronix 7L5 Spectrum analyzer
(associated with the receiver 8) and a
Tektronix 7613 oscilloscope for
monitoring the response of the system.
The principle on whlch the invention relies
is that each solder joint presents the inspection
apparatus with electrical properties (particularly
impedance) characterizing the physical condit~on of
the joint. Observation and comparison of the
10 impedance at various frequencies with that of
known solder joints enables the characterization of
the sample solder joint consistent with previously
known similar joints.
For exampl~, one mechanism of response of
15 the lead wire and solder joint (as shown in Figure 1)
to the vibrations may be related to the behavior of
an elastic rod constrained along its length. If the
rod is long and slender and constrained to eliminate
flexural modes, it will act like a mechanical filter
20 to pass only those signals at frequencies at or
near its longitudinal resonances. Hence, as the
frequency of the applied wave sweeps through at
least one resonant frequency of the joint, a peak
will occur in the spectral response (impedance
25 versus frequency) curve. The shapes and heights of
the resonance peaks will be a function of the
conditions of the lead and the joint. Losses
at grain boundaries and interfaces between
solder and inclusions or voids, will affect
30 the shape. Defects which affect the stiffness
of the joint will also affect amplitude o
the spectral curve. Generally the broadened
peaks indicate unsatisfactory conditions in the
solder joint. Depressed peaks at the lower
35 frequencies and magnified peaks at the higher
.
., : .
~ , '
-- 7 --
frequencies also seem to indicate unsatisfactory
conditions in the joint.
Figure 2 shows a typical spectral response
of the output impedance (as measured by the voltage
5 drop) of the receiver in the inspection of so~ne good
solder joints as the acoustic transducer applies
vibrations to the solder joint in a sweep over the
range of 150 khz-650 khz. The two peaks on the
left are at about 200 khz and 270 khz. Several
10 small peaks appear in the 520 khz area.
The spectral responses of some defective
solder joints are shown in Figures 3 and 4. The
curves selected represent rather extreme cases.
Other marginally defective joints may produce
15 spectral responses that are not always significantly
different from the good responses ma]cing it appear
that the test is very subjective. However, we
believe that this is occasioned only somewhat by
the apparatus limitations and the test itself but
20 probably more by the limitations in manually
preparing the "good" and "bad" samples. In other
words, some of the attempts at making good and bad
samples actually did not produce the desired good
or bad joint. The test appeared to be a more
25 accurate indicator of good and bad joints than the
preparation method was at preparing samples. The
fact remains that the test easily identified severe
defects and apparently identified marginally
defective joints with good accuracy.
Figure 3 shows the spectral response of a
joint containing minor inclusions. The broadening
of the 200 and 270 peaks as well as the much greater
effect on the amplitude of higher fre~uency peaks
is evident.
~L9 5~ G
Figure 4 also shows the same effects due
to a crack at the heel of the lead in the solder
joint. The stiffness has been severely reduced
causing a large increase in amplitude of the higher
5 frequency peaks.
The samples tested were fairly light leads
and appeared to be most responsive to frequencies
in the neighborhood of 500-600 khz. O~her lead
sizes and solder joints might require lower or higher
10 ranges to be responsive.
The spectral responses of solder joints
are truly characteristic of the conditions therein
but the difference between the responses of a good
joint and a bad joint can be subtle for some joints
15 which are close in their degree o defect.
Classifying the joint condition is therefore largely
a matter of learning and experience on the part of
the operator of the testing equipment. The
experienced operator can use this test to identify
20 poor solder joints with a high degree of accuracy.
