The arrival of various surface mount technologies for semiconductor packaging more than two decades ago resulted in very dense board assemblies that are difficult to test or debug at component level, mainly because of the greater difficulty with which the components' external electrical connections could be accessed. Furthermore, the problem only got worse with time as surface-mount packages continued to shrink.

'Bed-of-nails' testing, which refers to the use of hardware, jigs, and fixtures to test a semiconductor device by accessing its pins directly, was no longer a practical option for 'modern' packages sprung from these newer technologies. The need to resolve problems associated with the testability of these surface-mount packages triggered the formation of the JTAG in 1985, and the subsequent definition of the boundary-scan testing standard.

Boundary-scan testing basically involves embedding of special test circuits at chip level that would facilitate board-level testing of the chip and the board itself.  These additional circuits allow input and output signals to be scanned into and out of the I/O circuits of a device in a serial manner, letting the device to be tested with a small number of accessible pins (just 4 pins, to be exact). An industry standard since 1990, the JTAG standard is now one of the most popular and widely-used Design for Test (DFT) techniques today. 

The benefits offered by boundary-scan testing include: 1) significant reduction in the required number of physical test points on the board; 2) increased board component density; 3) corresponding reduction in costs associated with test fixtures; 4) shorter time-to-market; 5) on-board testing and programming capabilities; 6) reduced in-circuit testing time; and 7) higher production efficiency.

A JTAG-compliant device has: 1) a boundary cell on each of its I/O pins; 2) a 'scan path' or 'scan chain' that connects these boundary cells together in serial manner; 3) 4 or 5 additional pins to handle the JTAG signals; 4) a Test Access Port (TAP) for controlling the JTAG signals used in boundary-scan testing; and 5) a 16-state TAP controller or state machine that controls the states of operation of the boundary-scan testing.

Under normal operating conditions, the boundary cells simply let the input/output signals pass through them, into and out of the I/O pins. When the device is placed under the 'Test' mode, however, these boundary cells become 'active' for use in the direct capture or control of the signals going into and out of the I/O circuitries of the device, circumventing the device's normal input and output connections. The boundary cells basically consist of multiplexed shift-registers that are located around the chip's periphery (hence the name 'boundary' cell).

The test access port is simply a serial interface port defined by the IEEE 1449.1 to consist of at least 4 pins (optionally 5) in a JTAG-compliant device. These 5 pins, which are used to implement JTAG's serial protocol for boundary-scan testing, are: 1) the TCK pin, which is a clock signal that synchronizes the internal tap controller state machine operations; 2) the TMS pin, which is a mode select signal sampled at the rising edge of TCK to determine the next machine state; 3) the TDI pin, which is the data input pin; 4) the TDO pin, which is the data output pin; and 5) the TRST pin (optional), which is an asynchronous reset pin.

The properties and capabilities of a JTAG-compliant device's boundary-scan logic are defined by an external file known as the Boundary-Scan Description Language (BSDL) file.  BSDL files are provided by the manufacturers of  JTAG-compliant devices for use in the definition and generation of algorithms for the boundary-scan operation of their devices.

Testing a device or circuit board using boundary-scan technology may consist of the following basic steps: 1) the external tester applies diagnostic input signals to the input pins of the device; 2) the boundary-scan cells at the input pins capture the input signals; 3) the input data are scanned or serially shifted into the core via the TDI pin; 4) the resulting output data are serially shifted out of the core via the TDO pin; and 5)  the external tester compares the output data of the device against expected results.  Board defects such as open pins, missing devices, misoriented components, or dead devices can be found by such simple tests.

Boundary-scan technology has also been used by some companies in software debugging functions for CPU's. Using proprietary techniques complemented by adequate on-chip support, the JTAG interface was utilized to download code from a CPU, execute it, and examine register and memory values.  These functions are enough to handle most of the low-level debugging tasks expected from a typical software debugger. Aside from debugging, the JTAG interface can also be used for emulation, benchmarking, and profiling applications.