Minutes of Study Group Meeting, 2017-09-25

Meeting called to order: 11:06 AM EDT

The slide references relate to the pack used during this meeting, located here: http://files.sjtag.org/StudyGroup/SG_Meeting_6.pdf

1. Roll Call

Ian McIntosh (Leonardo MW Ltd.)
Heiko Ehrenberg (Goepel Electronics)
Brad Van Treuren (Nokia)
Bill Eklow (Retired)
Brian Erickson (JTAG Technologies) (joined 11:08)
Carl Walker (Cisco Systems)
Adam Ley (ASSET Intertech)
Ed Gong (Intel Corp.)
Bill Huynh (Marvell Inc.)
Joel Irby (ARM) (joined 11:15)
Dilipan Jayachandran (NJIT)
Rajesh Khurana (Cadence Design Systems)
Roger Lin (Via CPU Platform Inc.)
Teresa McLaurin (ARM)
Mukund Modi (NAVAIR Lakehurst)
Richard Pistor (Curtiss-Wright)
Russell Shannon (NAVAIR Lakehurst)
Naveen Srivastrava (Nvidia)
Jon Stewart (Dell) (joined 11:06)

By email (non-attendees):

Eric Cormack (DFT Solutions Ltd.)
Peter Horwood (Firecron Ltd.)
Terry Duepner (National Instruments)
Louis Ungar (ATE Solutions)

2. IEEE Patent Slides

  • {Slides 5-9}

3. Review and Approve Previous Minutes

  • {Slide 10}
  • September 18
    • Updated draft circulated 9/20/17
    • No further corrections
    • Brad moved to approve, seconded by Heiko, no objections or abstentions. Approved.

4. Review Open Action Items

  • {Slide 11}
  • [5.1] Ian to re-circulate link to forum thread on Scope and Purpose. Complete.

5. Discussion Topics

a) Scope and Purpose

  • {Slide 12, 13, 14}
  • Continuing discussion of "Need":
    • Need to remove inference that we are focused on vector based testing. This arose from trying to condense wording that attempted to capture that both legacy vector based tests a dynamic cases (e.g. where flow control may be determined by an intermediate result) needed to be supported. The resultant wording is no longer clear.
    • Debugging requires information from the component manufacturer on how to initiate tests and interpret results. May want to check either that the component is working or that it is correctly fitted to its board. Possibly start off a BIST and wait for the result. Manufacturers may be willing to put that into a specification.
    • Although autonomous tests are possible, it may not always be possible to run e.g. multiple MBIST tests simultaneously due to power limitations, so co-ordination needs to be managed externally.
    • May be introducing "classifications":
      • Autonomous tests than can be executed entirely within the component,
      • Other tests that require external stimulus or interaction.
    • While the pin-level of a SERDES link may be covered by 1149.6 and data link test would need higher level access and information that the device manufacturer may not provide (or even be willing to provide). System-level testing may be covered by the system-level application.
    • The standard could separately describe attributes associated with structural tests and how to access them and attributes associated with functional tests. However, control and management of the resources are likely to be the same in both cases.
    • We may get information on how to access a device at a black box level but functional access may be harder as manufacturers tend to block "back-door" access.
    • In 1687, iScan hides the intent of an access, so permits some level of black-box testing.
    • The standard needs to allow the device manufacturer to define the level of access the component offers. There may be different levels described depending on the type (depth) of testing that vendor is prepared to support.
    • The group's current title is based on "management of test access"; Notion was that the intent of the test didn't matter (that was for the tools or test application to deal with) and that SJTAG would manage getting the data to/from the right endpoints at the right times.
    • Probably need to brainstorm what the problems are, what the important features we want are, then sift them. Not all ideas will fit. For the Scope and Purpose we should be working out what we would like to achieve. The working group will need to solve the problems; the study group needs to define the problem space that the working group should address. Could define an architecture/framework - what features we want to have.
    • We could work out what access/control is lost as you move from component to board level and then from board to system level and then see what you could do to compensate. In addition to what is "lost", further constraints that need to managed or factored in are introduced as the assembly level increases.
    • The initiative group originally used a "requirements" spreadsheet to try to identify essential features before drafting the currently proposed Scope, Purpose and Need texts (http://files.sjtag.org/Brad/SJTAG%20Requirement%20List%2020170118.xlsx).
    • Perhaps need to define what a "system" and "application" are. This may depend on individual perspectives of the assembly hierarchy. A system could be several board installed on a mainboard, or could be a stacked die. May depend on who we perceive to be our target audience.
    • Where is the handoff between SJTAG and other standards. P1687.1 are already questioning this. P1838 is possibly in a similar position. P1838 is handling the "elevator" up through the die stack but the interface is essentially open.

6. Today's Key Takeaways

  • {Slide 15}
  • Brainstorm the problems that emerge as test moves from chip-level to board-level to system-level.
  • What constitutes a system? How is it differentiated from an "application"?
  • Where are the hand-offs between SJTAG and other standards?
  • SJTAG as "management of access" was to try to limit the scope (effort/time) of the standard.

7. Glossary Terms from This Meeting

8. Topic for next meeting

  • Scope and Purpose - continued
    • Aim to have a more focussed topic by the time the agenda is sent out.

9. Schedule next meeting

  • October 2.
    • Teresa expects be out.

10. Reminders

  • None.

11. Any Other Business

  • None.

12. List New Action Items

  • None.

13. Adjourn

  • Bill (Eklow) moved to adjourn, seconded by Brad.
  • Meeting adjourned at 12:10 PM EDT

Respectfully submitted,
Ian McIntosh