Ever-increasing circuit performance challenges
all aspects of test technology. Getting
a clean and accurate signal from the tester
electronics to the board under test is
critical for high speed test. Fixture
wiring can be a major contributor of distortion
and noise to the signal transmission path.
QA's Double-Ended Socket addresses the
limitations of fixture wiring by eliminating
the wire.
To better understand the possibilities
of wireless fixturing, QA has examined
the high frequency performance of Double-Ended
Sockets and probes on .100, .075, and
.050 inch centers. An RF Network Analyzer
was used to measure the frequency response
characteristics of a wide variety of probe
configurations. The frequency range of
these measurements was 300 KHz to 3 GHz.
A TDR Oscilloscope was also used to look
at the impedance of the signal path through
the socket and probe.
Test fixtures were constructed for .100,
.075, and .050 inch products. The fixtures
consisted of a .250 inch G-10 socket mounting
plate, a .062 inch G-10 socket spacer
plate and two electrical interface boards
attached to the socket mounting plate
with phenolic standoffs. The interface
boards provided the SMA connectors for
the test equipment and copper traces to
contact the various probe/socket configurations.
Configurations consisted of different
spacings for the ground and signal probes,
multiple ground probes and arrangements
to measure cross-talk where one pair of
probes was "driven" and the "pick up"
measured.
The following graphs study the performance
of the .100 inch centers assemblies. Comparable
data for .075 and .050 inch centers Double-Ended
assemblies appear in Appendices A and
B.
|
Figure 1 shows the frequency
response of two .100 inch probes
on 1.00 inch centers. This may
be representative of the signal
probe to ground probe separation
on an IC package. Note the roll-off
just above 100 MHz. |
| |
|
|
In figure 2, the probes are on
.100 inch centers. On these closer
centers, frequency response up
to 300 Mhz is achieved. This improvement
results from the more closely-spaced
probes providing a better match
to the impedance of the 50 Ohm
test environment. |
| |
|
|
A TDR Oscilloscope allows measurement
of the impedance of a transmission
at any point along its length.
Figure 3 shows the impedance of
two .100 inch probes on .100 inch
centers. |
In this TDR graph, the transmitted signal
had a rise time of 35 picoseconds which
equates to a 10 GHz test frequency. The
impedance extremes are exaggerated by
the very high bandwidth of the TDR step
signal; at the lower frequencies the differences
would be less apparent. These high frequency
measurements show three distinct physical
regions: the socket below the G-10 mounting
plate, the socket passing through the
G-10 plate and the probe extending above
the G-10 plate. The dip in impedance when
the socket penetrates the G-10 mounting
plate is due to the change of dielectric
material separating the sockets; i.e.
the transition from air to G-10 and back
to air.
|
Figure 4 shows the performance
of a three-probe in-line configuration
on .100 inch centers with the
signal probe placed between two
grounds. Although this configuration
may not always be practical, its
performance to greater than 2
GHz is exceptional. |
For a three-probe configuration (signal
between two grounds) excellent performance
to more than 2 GHz was achieved.
Cross-talk in a conventional fixture
is a complex function of many variables:
the characteristics of the test signals,
the length and type of wiring used, how
the wiring is (or isn't) dressed, and
the relative locations of the probes themselves.
Wiring problems are the reason for the
existence of double-ended sockets. Replacing
wiring with a translator board provides
a more repeatable and controllable environment
for routing test signals between the UUT
and the test electronics.
|
The test signals and probe
locations are driven by the needs
of the UUT. For reference purposes,
a plot of the cross talk between
two .100 inch double-ended assemblies
on .100 inch centers appears in
figure 5. Similar plots for .075
and .050 inch centers assemblies
appear in Appendices A and B. |
Cross talk for two pairs of .100 inch
centers double-ended assemblies on .100
inch grid.
A double-ended socket and probe assembly
is capable of delivering excellent high
frequency performance. Signal-to-ground
probe spacing plays a major role in determining
the quality of the transmission path;
in general, the closer, the better.
Replacing fixture wiring with a translator
board allows the test engineer greater
control of length and impedance characteristics
of the signal path to the unit under test.
This results in cleaner, distortion-free
test signals and higher performance testing.
|
Frequency
response of two .075 inch double-ended
assemblies (signal and ground)
on 1.00 inch centers. |
| |
|
|
Frequency
response of two .075 inch double-ended
assemblies (signal and ground)
on .075 inch centers. |
| |
|
|
For
a three-probe configuration (signal
between two grounds) excellent
performance to more than 2 GHz
was achieved. |
| |
|
|
Impedance
of the transmission line created
by two .075 inch double-ended
assemblies (signal and ground)
on .075 inch centers. Note: the
35 picosecond rise time equates
to an effective test frequency
of 10 GHz. |
| |
|
|
Cross
talk for two pairs of .075 inch
centers double-ended assemblies
on .075 inch grid. |
|
Frequency
response of two .050 inch double-ended
assemblies (signal and ground)
on 1.00 inch centers. |
| |
|
|
Frequency
response of two .050 inch double-ended
assemblies (signal and ground)
on .050 inch centers. |
| |
|
|
For
a three-probe configuration (signal
between two grounds) excellent
performance to 3 GHz was achieved. |
| |
|
|
Impedance
of the transmission line created
by two .050 inch double-ended
assemblies (signal and ground)
on .050 inch centers. Note: the
35 picosecond rise time equates
to an effective test frequency
of 10 GHz. |
| |
|
|
Cross
talk for two pairs of .050 inch
centers double-ended assemblies
on .050 inch centers. |
| |
|
|
