Most digital electronics designs now feature some form of JTAG/boundary-scan to aid structural board testing or in-system device programming. However, accessing the JTAG feature in a reliable and repeatable manner has always been an aspect somewhat neglected by boundary-scan technology and programmable device vendors. Until recently it has been largely the responsibility of the end user to liase with either the functional test system builder or maybe an In-Circuit Test (ICT) bureau to make sure that JTAG tester hardware is  ‘production ready’.

Expanding capabilities

Now though we see the co-incidence of two capabilities that can combine to make a powerful, adaptable and low-cost bench-top ATE (Automatic Test Equipment) system.  With the introduction of low-cost, multi-function JTAG testers such as the ‘MIOS’series from JTAG Technologies you get a compact boundary-scan based tester complete with mixed signal IO channels (voltage, measure, source and frequency) that can be embedded directly into a re-configurable test fixtures – see figure 1). Add  power supplies and any other measurement resources you might require and your ATE hardware is complete.

Test all board types (not just JTAG/boundary-scan)

Such is the flexibility of this type of system, it is easy to build a 256+ I/O channel tester that can be configured for the testing of all board types, not just those equipped with JTAG on-board. By means of personality adapters (aka cassettes) users can extract the maximum benefit from their tester resources.

Software options

All PCB tester systems require a developer and test executive platform. Fortunately MIOS hardware is supported by numerous options to suit all budgets.  Therefore in order to develop and execute  high-integrity JTAG tests using the MIOS hardware (e.g. scan path infrastructure, interconnections, logic clusters &  memory clusters), ISP (In-System Programming), ADC/DAC, PSU and clock tests system builders can choose from:

1. Python scripting language – a favourite among engineers and scientists that comes with a host of open-source libraries to control instruments (e.g. GPIB, USB), serial ports, JTAG controllers and more. Its flexibility allows easy construction of loops, branches, variable handling and file IO. You may have  to undertake more coding than some other methods, however dozens of test modules already exist for devices and logic clusters accessed by JTAG/boundary-scan and there is even a VISA control library in existence.
2.  LabVIEW –  for scientists with a preference for graphical icon-based programming, National Instruments LabVIEW has become a staple of the test world especially within functional mixed-signal testers. Control icons (VIs) exist for prepared test execution and also low-level control of JTAG device pins.
3. JTAG ProVision  – for those with no desire to undertake coding (graphical or textual), then a high-level auto-generation system like simply imports CAD data and device models from a supplied library featuring 10 000s of parts and exports a complete test program set. You can still augment these tests with scripted applications to get the ultimate in flexibility and fault coverage.

Furthermore there is also .NET, TestStand and ATEasy support, and other options depending on the current method of working or individual preferences.

For hardware interfacing (fixtures) the open architecture of the  MIOS tester hardware allows it to be added to almost any popular cassette based fixture. To date there have been successful applications created with ATX, DD Mechatronics, ECT, Ingun and MG hardware but it’s straightforward to marry this potent test capability with any local or international fixture supplier.

Conclusion

Test equipment for PCBAs is an often neglected responsibility for many small and medium businesses. Relying on a CEM (contract manufacturer) that may have little or no knowledge of a board’s functionality can be ‘haphazard’ to say the least – even if your board is tested do you know to what extent (fault coverage)? Taking ownership of, or at least engaging with the test problem will give a better understanding of process failures and the fault spectrum that can then be more specifically addressed. Without this you will inevitably be paying for the board bone-pile either directly or as a hidden cost.