Topics for ETS2 2016

Discuss the generic proposals for SJTAG
Post Reply
User avatar
Ian McIntosh
SJTAG Chair
Posts: 429
Joined: Mon Nov 05, 2007 11:49 pm
Location: Leonardo, UK
Contact:

Topics for ETS2 2016

Post by Ian McIntosh » Thu Jan 07, 2016 1:48 pm

We need to have a synopsis for our ETS2 submission before the end of the month. That probably means we need to settle on the text at the meeting of Jan 20.

Use this thread to suggest topics, and perhaps propose wording for the synopsis. Remember that our idea here is to present the issues arising from practical system level applications as "open" and seeking solutions.

I'll start off by suggesting a couple of things that come to mind:
  • The conflict between optimally terminating the JTAG bus for board level test compared to system level with a multidrop bus.
  • Making a system field reprogrammable ("covers on") - managing and addressing boards/chains, configuration control, gang-programming multiple instances of a board.
What other ideas are there? I think I'd like to see something with a more "embedded" angle than the above and Michele suggested also explaining why the recent standards aren't doing enough.
Ian McIntosh
Testability Lead
Leonardo UK

User avatar
Ian McIntosh
SJTAG Chair
Posts: 429
Joined: Mon Nov 05, 2007 11:49 pm
Location: Leonardo, UK
Contact:

Re: Topics for ETS2 2016

Post by Ian McIntosh » Tue Jan 19, 2016 6:18 pm

There has been some email exchange on the draft of the abstract for ETS2, so this post is just to try to bring it together in one place.
Original draft wrote:System level testing often focuses entirely on functional tests which, while offering good fault detection is frequently poor at fault diagnosis, and is unable to exploit the investment in board level tests used during manufacturing. Why should this be the case?

The JTAG (IEEE 1149.1) test standard has continued to evolve since its inception and has spawned additional standards and techniques that address device programming, JTAG chain management, embedded instrument control, etc., but the needs of the system level user still seem not to be addressed in any coherent manner that allows the system test integrator to take a repeatable approach to test development using off-the-shelf software tools.

These test development issues can be compounded if the system design incorporates SoCs or COTS boards where there is IP involved that the vendor wishes to protect. SJTAG (System JTAG) aims to offer a standard for system test applications, but what does it need to do to draw together the existing test standards and reconcile conflicts?
Michele wrote:it looks like a good summary, and well in line with the ETS2 format of "provoking a discussion".

There is one point that IMHO is fundamental in SJTAG and that could get a line in the second paragraph: existing standards target the Design phase of circuit, so they can prescribe specific and coherent DfT approaches. On the other hand, System Level comes at the integration phase with little or no control on the DfT of the individual components, so SJTAG must try and be as inclusive as possible: provide ways to efficiently reuse existing and potentially incoherent DfT elements and strategies.
Ian wrote:I agree with your statement that SJTAG needs to be able to make use of what it "inherits" from the board levels, but I'm also of the opinion that there's a failure to design for the system level of testing. In fact, I'd argue that in many cases there's a complete failure to "design the system" - I think that what often happens is we design boards/modules that fit into a pre-conceived architecture (or decomposition) that typically makes up the product we're designing. So we design boards that conform to some interface and function definition without thinking too much about what the system needs to do, and so "system test" gets overlooked.
Michele wrote:That is exactly the kind of added value for SJTAG that we must put forward!
Heiko wrote:I think you both brought up great points for a discussion at ETS2.

Perhaps it would be good to have a slide or two that introduces the use cases for SJTAG we have identified in the past, to give attendees a context and to see if someone has additional use cases.
Ian McIntosh
Testability Lead
Leonardo UK

Post Reply