White Paper:
Switching Hardware for Automated Test and Measurement Systems
-----------------------------------------------------------------
Keithley Instruments Inc
28775 Aurora Rd
Cleveland, OH 44139
216-248-0400,  800-552-1115,  fax 216-248-6168
-----------------------------------------------------------------

The availability of low-cost PCs with IEEE-488 instrument control
capabilities makes automated test and measurement attractive to a wide
range of users, from laboratory researchers to quality control inspectors.
Signal switching is a key concern in automated test and measurement
systems, whether the goal is testing a single device or thousands of them.
Many measurement accuracy or repeatability problems can be traced to
improperly-configured switching systems. 

There are four basic switch topologies:

Matrix switching. A matrix switch is the most versatile type of switching
system. Any input can be connected to any output by closure of the switch
at the intersection (cross-point) of a row and column. This helps minimize
the need for complex wiring and interconnect systems, and can simplify the
Device Under Test (DUT) interface.

Multiplex switching. A multiplex switch connects one instrument to multiple
devices under test or multiple instruments to one device under test.
Typical applications of multiplex switching include capacitor leakage,
connector/switch contact and insulation resistance test systems. The
multiplex switch is useful in combination with matrix or other switch
configuration to expand switching capacity by sharing electrical paths,
provide additional isolation, reduce cross talk between channels, etc.

Scanner switching. A scanner is a special case of multiplex switching in
which switch closures are sequential or serial, sometimes with the
capability to skip channels. Typical uses of scanner switching include
burn-in testing of components, monitoring time and temperature drift in
circuits and acquiring data on system variables such as temperature,
pressure and flow.

Isolated switching. The isolated switch configuration consists of
individual uncommitted relays, often with multiple poles. Isolated
switches are not connected to any other circuit. Therefore, they are free
for building very flexible and unique combinations of input/output
configurations with the addition of some external wiring. This type of
switch can be useful for creating additional isolation between circuits,
providing safety interlock, actuating other relays or circuits, or
building special topologies such as binary ladders and tree structures.

Builders of automated test and measurement systems face a variety of
technical, timing, space and budgetary challenges when choosing and
configuring their switching hardware. Increasingly, switching hardware
manufacturers are tailoring their new offerings in response to users'
demands for equipment that simplifies the configuration process and
minimizes costs, while maximizing system throughput. 

System Builder Challenge #1 -- To configure a user interface for the
switching system that's easy to understand, program and operate. When
choosing switching equipment, it's critical to know in advance how complex
it will be to get the system up and running for a given application. Once
a switching system is in place, it should also allow the user to make
modifications readily as needs expand or change. The user interface will
have a major impact on how easy the process of programming and operation
will be. 

For example, one of the hardest parts of getting a switch system up and
running is keeping track of channel open/closed status. This can be a
concern with systems of all sizes, not just high point count applications.
The test programmer goes through several phases as the application
software is developed. In the learning phase, the programmer gets a feel
for the mainframe's operation, experimenting with controlling functions
via the front panel and over the IEEE bus. Usually, one of the first
experiments is to close some channels and route a few signals back to the
measuring instrumentation. A switching mainframe that provides channel
status information on the front panel allows the operator to get instant
feedback on the operation of the switch topology, and speeds the learning
phase.

During checkout/debug, most of the system has been programmed and must be
tested. As in the learning phase, easy access to channel status
information is very valuable. It not only provides instant feedback that
the correct channels (or incorrect channels) are being closed and opened,
but also makes it possible to verify that closures are occurring in the
proper sequence.

The last phase is actual operation. Here, channel status information helps
indicate the test is running properly and is not locked up. It can also
assist in on-going maintenance of the system.

Many switch mainframes require the user to manipulate front-panel controls
or send special commands over the bus to determine the current status of a
particular channel. However, some newer mainframes have user interfaces
that employ a graphical pattern on the front panel display that shows the
open/closed status of all channels simultaneously. Systems with high
channel counts sometimes provide the same capability via an interactive
channel status LED display. By making it easier to get the switching
system running quickly and simplifying modifications, these types of
displays can reduce programming complexity dramatically.

As mainframe designs have changed, so have the designs of switching cards,
with a growing number of high density switching cards now available.
Inserting any of these cards in a compatible mainframe automatically
configures the display to show the available channel capacity for that
card.

Although many mainframes allow users to program and control switching from
an external controller via the IEEE-488 bus, some of the newest offerings
can also program and control a scan (channel spacing, scan spacing, number
of scans) without direction from a computer with just a few key presses on
the front panel user interface. This is due to the fact that scanning
functions are built in as opposed to being available from the front panel.
This frees up the computer to process data or monitor other aspects of the
test. In some of the new high channel count mainframes, an optional light
pen allows users to "point and click" the desired channels or crosspoints
on a channel status LED grid. Users can build scan lists, create matrix
patterns, open or close channels, and store patterns in memory simply by
selecting the LEDs for the desired channels or range of channels, then
clicking the appropriate light pen keys.

System Builder Challenge #2 -- To maintain signal integrity through the
switching hardware. When configuring a system, many engineers focus all
their energies on choosing and assembling the optimum mix of sources,
measurement instruments, controllers, and software for their application.
But if they fail to lavish the same kind of attention on the quality of
the switching configuration, the measurement integrity of their entire
system becomes suspect.

Each time an engineer adds another component or instrument to a test
system, he increases the potential for introducing measurement errors.
That means that, when adding switching capabilities to a test system,
engineers must take into account any factors that could degrade overall
signal quality. Even in so-called "non-critical" applications, degradation
due to offset voltage, isolation resistance, and leakage current can
become a significant cause of error. To ensure high switching integrity no
matter what the signal level, users should look for a system that offers a
broad range of compatible cards. This ensures the user the flexibility to
choose the best card for the task without sacrificing signal integrity.

System Builder Challenge #3 -- To get the highest switching value for the
hardware investment. Many system builders must work within tight budgetary
limitations when choosing switch hardware. In these situations, one of the
most critical considerations is the equipment's switching density -- in
other words, the channel capacity for a given space. Switching density
becomes especially important as the number of test points rises. The lower
the switching density, the higher the number of separate cards and
mainframes that will be required to handle a particular application.
Conversely, the higher the equipment's switching density, the less
hardware will be needed (and typically, the lower the cost to solve the
application). The size of the mainframe itself is also important in
situations where space is also limited -- a low-profile, high-density
mainframe is particularly advantageous in production test applications,
where rack space is often at a premium.

In order to handle 80 channels of two-wire switching, some systems require
two mainframes with eight separate switching cards installed. The extra
hardware and wiring needed when using such low-density equipment can not
only increase system complexity, but may also complicate the configuration
task unnecessarily. In contrast, some newer designs can accomplish the
same task with a single half-rack mainframe and two cards, simplifying
system set-up and minimizing equipment costs.

The channel capacity of the switching cards and slot capacity of the
scanner mainframes involved are not the only factors that determines a
system's switching density. A number of mainframe manufacturers, including
Keithley, have developed scanners with an analog backplane that
automatically makes the intercard connections inside the mainframe when
cards are installed. The analog backplane provides greater configuration
flexibility and virtually eliminates the need for card-to-card wiring for
many applications.

Inter-product compatibility is another consideration that can help system
builders economize on switching equipment purchases without sacrificing
performance. The ability to use the same accessories and companion
products with a number of different mainframes is a good illustration of
this. For example, many applications, such as research and development,
require relatively low channel counts. As the needs of the application
change and grow, system builders typically want to avoid buying lots of
new switching cards for their new, higher capacity mainframes. Software
compatibility between products is also important because it makes it easy
to adapt programs developed for lower channel count mainframes for use
with higher channel count models.

System Builder Challenge #4 - To speed and simplify the process of
triggering multiple instruments. There's more to system speed and
throughput than the readings per second specification. It also means being
able to trigger, switch, and measure without having to determine all the
timing parameters of the system empirically.

Controlling trigger timing in PC-based test and measurement systems with
multiple sources and instruments can tricky, even for experienced system
designers and users. The latencies and timing uncertainties involved in
coordinating the operation of multiple sources and instruments have long
limited the speed and accuracy of traditional techniques such as software
triggering via the IEEE-488 bus or the use of external trigger inputs and
outputs.

Fortunately for system integrators and users, a new type of triggering
hardware, sometimes referred to as a system trigger controller, has
emerged. These products function as programmable trigger sources and
provide multiple trigger input and output channels. The trigger inputs and
outputs of each controlled device are connected to an independent set of
inputs and outputs on the trigger controller, allowing each device to be
controlled and monitored independently.

The user programs the system trigger controller to generate the required
triggering relationships. The benefits of this approach include: 

* Triggering relationships may be changed without altering
  physical connections.
* Control of all devices is centralized, and the state of the
  measurement process is easily observed.
* One or more triggering pulses may be issued at a precise rate,
  or in response to any arbitrary combination of trigger pulses.
* A delay may be inserted between trigger outputs, allowing the
  user to compensate for devices that do not provide a status
  output.
* Any arbitrary sequence of trigger commands and delays can be
  specified through a trigger program, and the measurement
  process may be started and stopped precisely.
* Recent benchmark testing indicates these new triggering
  approaches may allow users to improve their test throughput
  by up to 8x when compared to software triggering.

In order to achieve maximum throughput, delays introduced by the system
trigger controller must be negligible in comparison with those introduced
by the instruments themselves. One of these products, the Trigger Master
from Keithley, employs a custom microsequencer with an 8MHz clock and
special I/O circuitry to minimize latency and improve timing accuracy.

The system trigger controller architecture provides the user with many
features that were previously unavailable, including:

* Multiple TTL-compatible triggering channels, which may be
  designated as inputs or outputs (and changed dynamically). 
* Ability to output a trigger signal or a series of pulses with
  100ppm timing accuracy and negligible timing jitter.
* Ability to detect a specified number of transitions of either
  polarity on any combination of the triggering channels, and to
  generate trigger outputs in response to a trigger input in less
  than 2.5 usec.
* Memory may be loaded with a series of operations to be carried
  out by the microsequencer, allowing the PC to assume other
  tasks during the measurement process.
* Ability to define up to two levels of loops within the trigger
  control program, allowing a group of operations to be executed
  repetitively.
* A crystal-controlled timebase that may be used to generate a
  wide range of time delays (1 usec. - 65.535 sec.) to be
  inserted between trigger events with high resolution and
  accuracy.
* Internal status registers that may be read at any time by the
  host PC in order to determine the exact state of the
  measurement process.
* An internal flag register that may be written to by the trigger
  control program, allowing the host PC to trace its execution.
* Ability to generate a hardware interrupt to signal the need for
  service. This capability is useful when the setup (measurement
  range, sensitivity, etc.) must be changed in the middle of a
  test.
* Significantly reduced trigger latency due to the
  microsequencer's high clock speed.

The Trigger Master is connected to the sources and instruments being
controlled through a bus known as the Trigger-Link. Trigger-Link-equipped
instruments, such as Keithley's Model 7001 and 7002 scanner mainframes and
the Model 2001 Digital Multimeter, feature dedicated circuitry that allows
external trigger inputs to initiate actions directly, reducing triggering
latencies. Trigger-Link's features and capabilities include:

* A total of eight conductors - six trigger "channels",
  two grounds.
* Interfaces through an eight-pin "Micro DIN" connector.
* Devices with Trigger-Link provide two connectors to permit
  "daisy-chaining" of the bus through several devices. The
  Trigger Master resides at one end of the bus and provides
  a single connector.
* Devices with Trigger-Link may be programmed to source and
  accept triggers on any of the six channels.
* An optional adapter box allows non Trigger-Link compatible
  instruments to be interfaced through BNC connectors. A separate
  BNC is provided for each Trigger-Link channel.
* Triggering rates of up to once every 10 us (100 kHz) are
  possible.
* Trigger-Link instruments feature dedicated circuitry that
  allows external trigger inputs to initiate actions directly.
  This approach further reduces triggering latencies. Other
  instruments frequently use their internal microprocessors
  to monitor the state of the external trigger signal, and
  therefore, cannot initiate actions as quickly as those with
  Trigger-Link capabilities.
* The SCPI (Standard Commands for Programmable Instruments)
  standard, adapted by all major instrument companies, contains
  a well-structured trigger programming scheme. Trigger-Link is
  built around this SCPI trigger model, so access and control are
  easy and well-coordinated.

System Builder Challenge #5 -- To choose a switch system that offers broad
flexibility to match a variety of applications. Many research and
engineering organizations simply don't have the financial resources
necessary to purchase new switching hardware for each new system built. In
these cases, existing switching equipment from old ATE systems must be
applied to new systems. Therefore, the switching hardware chosen must be
adaptable to a broad range of operating environments. For example, in many
engineering applications, users are involved in design- or
development-related work that requires testing devices or assemblies over
a wide variety of conditions. Modifications to the test system are made
frequently as users learn more about a particular device. Many different
versions of the same device may also require testing. Therefore, the
ability to configure a system quickly and to make wiring changes easily is
critical to engineering users, since they can't always know in advance
what their exact test requirements may be.

The ability to make changes to an experiment rapidly is equally important
in research applications. Research users often must be able to mix
different types of wire on the same card in order to handle a wide range
of inputs.

Production test users, those involved in some aspect of testing, quality
assurance, or inspection related to the manufacture of a product, have
different interconnect needs.

Their applications are usually permanent and unlikely to change once the
system is configured and in use.

The newest mainframe designs have been developed to satisfy the needs of
all three types of users with a choice of interconnect options. Many offer
a detachable screw terminal connection board, which simplifies wiring
connections and making modifications. The screw terminals also allow
research users and engineers who require maximum flexibility to combine
different types of wire, such as shielded cable, ribbon cable, twisted
pair, etc. on the same card. For production test applications, optional
multi-pin connectors allow the user to disconnect cards quickly without
removing them from the mainframe.

 ============================================================
 From the  'New Product Information'  Electronic News Service
 ============================================================
 This information was processed from data provided by the
 above mentioned company. For additional details, contact 
 the company at the address or telephone number indicated.
 OmniPage Pro is now used for converting all printed input! 
 ============================================================
 All submissions for this service should be addressed to:
 BAKER ENTERPRISES,  20 Ferro Dr,  Sewell, NJ  08080  U.S.A.
 Email: RBakerPC (AOL/Delphi), rbakerpc@delphi.com (Internet)
 ============================================================
