Error detection/correction and fault detection/recovery – Pulse or data error handling – Transmission facility testing
Reexamination Certificate
1999-01-04
2001-10-02
Tu, Christine T. (Department: 2133)
Error detection/correction and fault detection/recovery
Pulse or data error handling
Transmission facility testing
C714S712000, C370S245000
Reexamination Certificate
active
06298458
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates generally to the field of self-test structures for communications devices such as transceivers, and more particularly, a system for testing complex integrated circuit communications transceiver devices.
2. Discussion of the Prior Art
Transceiver circuits used in network communication applications, e.g., Ethernet, are designed and built as stand-alone chips with a standardized digital communications port for connection to the digital domain. Due to their stand-alone nature, and the relative simplicity of early transceivers, it was relatively simple to test these chips functionally to verify both the logic portion of the chip, and the analog transmit and receive circuitry. Early transceivers were built using a current driving, wave shaping, digital-to-analog converter as a transmitter and multiple analog comparators as a receiver. Due to their relative simplicity, these structures could alternatively be tested by parts when integrated into a larger piece of silicon with the digital portion being tested via Level-Sensitive Scan Design (“LSSD”), the DAC portion being tested through a series of DC parametric measurements, and the receiver being tested through a latched receiver test.
As the speed and complexity of the transceiver circuits increases, however, so does the complexity of the receiver. Moreover, the complexity of the receiver increases at a faster rate than the complexity of the transmitter and the digital circuitry. The transmitter and the digital circuitry both may still be tested via traditional LSSD and parametric means with little test coverage impact. Newer receivers, however, generally contain variable equalization and gain circuity, PLL driven clock recovery, peak detection, and AC coupled input circuitry with internal DC restoration which makes testing via DC parametric levels ineffectual. As a result, not only has functional test been required to verify the operation of the receiver, but such testing has generally needed to be performed over multiple cable lengths to insure that the equalization and gain range of the receiver can mitigate both signal loss and signal (primarily phase) distortion caused by varying lengths of cable over which the signal may travel.
FIG. 1
illustrates a data flow for an industry standard test methodology
10
such as described in the reference “ICS 1890 Test Plan Revision 3.0” ©1996 Integrated Circuit Systems, Inc. In this prior art method, all portions of the stand-alone transceiver chip are tested functionally at speed with repeated tests over multiple cable lengths, e.g, 1m-100m of cable, to test the ability of the device to adapt to, and successfully receive signals transmitted over varying lengths of cable. As shown in
FIG. 1
, the digital patterns from the tester (not shown) are input to the digital portion
15
of the transceiver under test, which generates signals for testing the analog transmitter portion
20
. Outputs from the analog transmit portion
20
are current signals
21
which are input into passive components
23
a
and coil (magnetic) step-up module
24
a
for converting the current signals
21
into voltage signals suitable for input to the analog receiver portion
40
of the transceiver device. Depending upon the test, these analog voltage test signals are transmitted to the receiver
40
via variable length coaxial cables
30
having a pair of RJ45 connectors
25
a,b
and corresponding passive components
23
b
and step-down coil (magnetic) modules
24
b
for converting the test signals into the proper input signals
26
for the analog receiver.
The reference entitled “System and Method for Analog Test and Trimming of a Transceiver Circuit,” IBM Disclosure BU8-97-0336, 1997, describes an earlier test methodology
80
, the likes of which is illustrated in FIG.
2
. Using this methodology, analog signals brought into the transceiver and converted to digital signatures via RX port
81
could be sampled by register
82
at appropriate intervals during a test window. Composite samples stored in
82
may be selected as data input to the transmitter portion of the transceiver
86
via mux structure
85
and control logic
88
. Measurement using external test equipment of the analog waveforms produced by transmitter
86
respondent to register
82
contents provides a means of both assessing the function of the receiver
81
via analog output measurement, and providing analog feedback to monitor resistor trimming processes common in the tuning of analog integrated circuits. Additionally, a method allowing transmitter
86
to alternately source analog test data which may be provided by a built-in-self-test (BIST) to the receiver
81
via an optional external connection
89
and provide analog signatures for the verification of receiver
81
using sampling register
82
and control logic
88
is described. The methodology also adds Mux Latches
83
to provide for digital test stimulus for the bulk of the receive digital function
84
, and a built-in-self-test unit
87
to provide data patterns, primarily for analog verification of the transmitter
86
. Other digital portions of the transceiver, not shown, are tested using standard digital methodologies such as level-sensitive-scan-design (LSSD), scannable flip-flops, or functional patterns.
A prior art test methodology implemented for IBM's 10T Ethernet core is illustrated in FIG.
3
(
a
). As shown in FIG.
3
(
a
), the transceiver device
50
is segmented into a receive portion
65
, a transmit portion
70
, and a digital portion
16
comprising a transceiver digital logic device
15
, transmit BIST logic device
63
and mux/latch boundaries devices
60
a
and
60
b
, each device being tested separately. The digital portion
16
is tested via LSSD means with mux/latch boundaries
60
a
and
60
b
providing alternate stimulus at the analog-to-digital boundary and digital data sampling at the digital-to-analog boundary respectively. The transmitter
70
is tested via parametric measurement (i.e. current or voltage) with patterns driven by a transmit counter/built-in-self-test (BIST). The receiver
65
is tested via a latch receiver test similar in nature to ASIC receivers whereby the digital results of analog-to-digital conversion performed by receiver
65
on DC input signals are captured at the mux/latch boundary
60
a
for scanout and verification via the tester.
FIG.
3
(
b
) illustrates a test methodology for the IBM token ring transceiver embedded into a silicon chip which includes an on-chip analog loop back path
72
including a transfer gate
73
for providing an internal bridge between the analog transmitter
70
′ and receiver
65
′ in the circuit and used to test the receiver portion of the circuit.
In addition to the references cited above, U.S. Pat. Nos. 5,337,316, 5,402,440, 5,648,972 and 5,675,588 each described methodologies for testing communications devices, with U.S. Pat. No. 5,648,972 particularly directed to testing of a single integrated circuit transceiver.
Functional single IC transceiver testing of the nature as shown in
FIG. 1
, uses the transmitter portion to drive the receiver and requires complex board design to include passive components and coaxial connectors, as well as the ability to switch in and out coaxial cable lengths from 1 meter to more than 100 meters in order to “generate” the receive signal used for test. Generally, because the circuit is allowed to auto-converge to an equalization and gain setting, the test time for each cable length is greatly increased, e.g., to more than four seconds, while the data collection and verification methodology becomes complex, requiring either on-board or on-tester data collection and post-processing to handle the variable latency of the different cable lengths and the high data rate of the communications port. These factors make implementation of this test methodology in a digital test manufacturing environment extremely difficult, especially when multiple port counts which may exceed twe
Cranford, Jr. Hayden C.
Gude Eirik
Iadanza Joseph A.
Owczarski Paul A.
Raymond Jonathan H.
International Business Machines - Corporation
Scully Scott Murphy & Presser
Shkurko, Esq. Eugene I.
Tu Christine T.
LandOfFree
System and method for manufacturing test of a physical layer... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and method for manufacturing test of a physical layer..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method for manufacturing test of a physical layer... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2598949