The System JTAG Working Group

Minutes of Weekly Meeting, 2008-12-08

Meeting called to order at 8:20 EST

1. Roll Call

Carl Walker
Eric Cormack
Brad Van Treuren
Peter Horwood
Ian McIntosh
Heiko Ehrenberg
Tim Pender (joined at 8:42am EST)

Excused:
Carl Nielsen

2. Review and approve previous minutes:

11/26 Minutes: Approved, motion by Eric, second by Brad

3. Review old action items

Adam proposed we cover the following at the next meeting:

Establish consensus on goals and constraints
What are we trying to achieve?
What restrictions are we faced with?

Establish whether TRST needs to be addressed as requirements in the ATCA specification if it is not going to be managed globally (All)
Adam review ATCA standard document for FRU's states
Patrick contact Cadence for EDA support person.
All to consider what data items are missing from Data Elements diagram
[All] do we feel SJTAG is requiring a new test language to obtain the information needed for diagnostics or is STAPL/SVF sufficient?
see also Gunnar's presentation, in particular the new information he'd be looking for in a test language (http://files.sjtag.org/Ericsson-Nov2006/STAPL-Ideas.pdf)
[Ian] I propose to defer any further discussion on data and languages until the New Year when we can look at it with fresh minds, but please keep thinking about the subject.
[All] does the new proposed meeting time for Mondays: 4pm GMT (http://wwp.greenwichmeantime.com/) work for you?
do we still want to hold Wednesday meetings?

4. Discussion Topics

White Paper section status

Volume 1 - Overview:
[Ian] I think we've concluded that the Overview is as complete as we'll get it for now?
[Brad] I agree.
Volume 2 - Use Cases:
[Heiko] There are no updates to report.
Volume 3 - Hardware:
[Carl W] We both have information to input: We've been trying to get a session togther to divide up the work between us. I expect it'll happen next week.
Volume 4 - Languages:
[Ian] I don't think we'll be able to progress this until we can get further into the whole data and languages subject, and at the moment no-one has been assigned to this section.
Volume 5 - Business:
[Ian] I wasn't sure what was intended here, but I exepected that Value Propositions from the Use Cases would form part of this.
[Brad] Yes, the Value Propositions will be part of justifying each use case. We should also reference Tim's 2003 BTW paper as it puts dollar values on the benefit of remote updates. We should grab anything that shows dollar values to support our arguments.
[Ian] Heiko, you have some value proposition material in Volume 2: Should we move that to Volume 5 in due course?
[Heiko] There is some. I guess it can move OK.

EDA participation

[Ian] I've chased up my contacts at Mentor Graphics, but haven't managed to get any response from them. It's really a bit disappointing. I got a couple of other names at NTF: Thomas Rinderknecht and Paul Reuter at Mentor and Ed Malloy at Cadence, but I haven't tried following those up.
[Eric] Synopsis will keep dropping into the website to monitor what's happening but they say that none of their customers are asking for SJTAG support, so it's not a priority for them.
[Brad] Thomas Rinderknecht is on the 1687 group - he's really a device person.
[Ian] So maybe not really in a position help to us.

Revised meeting time (proposal)

[Ian] The proposal from last week was for Mondays at 11:00am Eastern. Patrick has since indicated that he'd prefer 10:30. It would also suit me better to be a little earlier, but I don't have a strong opinion on that. 11:00 is as late as I'd like to go. Adam has a bit of a problem with 11:00 am Monday as it backs up with another meeting, meaning he'd have to miss the latter part of our meeting. Moving to either a Thursday or a Friday would help him or alternatively start at 10:30. Friday just doesn't work for me though. I e-mailed Anthony Sparks over the weekend to ask if 10:30 was too early for him, but I haven't heard back yet.
[Brad] The 8:15 start was realy to help out China, but they've pretty much dropped away now.
[Heiko] 10:30am EST on Monday would work for me.
[Brad] I'm wondering if you would propose another day in the week; Thursday at 11am would work for me, for example, and it may be better for the west coast.
[Heiko] Thursday isn't so good for me as I wouldn't be able to make so many meetings due to travelling. Same for most days through the week.
[Brad] With Heiko as Vice-chair you need to consider that.
[Ian] I think we need to postpone a decision on this until next week. I'll take an action to circulate a note on the options we've discussed here, so we can come to a conclusion next week.

Follow-up survey

[Ian] We had propsed to carry out a follow-up survey "later in the year" - That's obviously going to be next year now. But what are we actually hoping to learn from the survey? How will we benefit from the information?
The previous survey had the aim of identifying the topics that the broader community felt were important for SJTAG, prompted by a comment from Jim Webster.
The old survey is at http://survey.sjtag.org/jul2008/stopped_index.html if anyone want to refresh their memory.
During the exercise of preparing that survey there were questions that I felt I wanted to ask to gain a better understanding of how people were currently employing BSCAN (not necessarily SJTAG). Section 2 ended up just asking if the respondent used JTAG at board and/or at system level to keep that survey simple. But questions like that are only relevant to End Users not Tool Vendors, Consultants, Educators, etc., so maybe we need questionnaire sections for each category of person?
[Brad] I agree there's a need to partition the survey for test engineers vs tool vendors vs management. Maybe end users isn't a good enough category - I think we'd say we have different types of end user, depending on what role you're in. There was also a link to the survey results?
[Ian] Yes it is [admin URL removed]
[Brad] OK, I was really looking at the categories we had in there.
[Ian] I'd been thinking that the Roles and Functions we had in that survey weren't much use to us this time, but now I see that the Roles we identified then are a lot better than I remembered.
[Ian] We had a "refer-a-friend" option previously. Given that we'll only be sending this survey to people who opted-in last time, do we still want that feature?
[Brad] I think that's still a worthwhile feature.
[Ian] As for the questions, I guess I've mainly been considering the End User case, not management, etc., perhaps asking people if they have applied each of the use cases.
[Brad] We should ask people to rank their top five use cases.
[Ian] You mean ask the management people?
[Brad] No, generally. Within our company there are people who use SJTAG for different use cases; for different reasons.
[Tim] Update is very important for us.
[Ian] Also in my business.
[Brad] Accessibility is the main reason driving the use of remote updates.
[Tim] An aside: There was a mention of test coverage earlier - Is there any kind of standard for how test coverage is measured?
[Brad] Test coverage standardisation may be outside of our scope.
[Ian] Tool vendors have likely applied their own rules to this and they are probably not consistent with each other.
[Brad] With JTAG it's quite deterministic what nets are tested.
[Ian] Problem might be how you treat nets that are held in a fixed state during test so you don't know if they can be driven to the opposite state. Do you call them 50% tested or untested?
[Brad] You need to look at the categorisation of nets too. I've seen boards recently where over 80% of the nets are SERDES with no BSCAN cells on the pins, so only 10% of the nets on the board are testable. If there's less than 25% of testable nets on a board, then it's probably not worth investing in an embedded BSCAN test.
[Tim] So that would involve some FPGA based functional test?
[Brad] Well, yes - JTAG enabled testing.
At the TTSC Chair's meeting at ITC there was an initiative to look at SERDES testing.
[Tim] Would you have to configure the devices to prevent damage, driving TTL into LVDS for example?
[Brad] RocketIO on Xilinx Virtex 5 is on dedicated pins, so that's not so much a problem, but some configuration is often needed, to set up impedance matching on SRIO for example.
[Tim] And there are DSP dedicated lines.
[Brad] Yes, to get the bandwidth we need, we're finding we're using DSPs more.
[Carl W.] I agree.
[Brad] DSP farms, FPGA farms all coming together to provide a flexible interface. Testability is not keeping up.
[Ian] Delay Testing is now often being used at device level aid testing, but that doesn't benefit board or higher level test.
[Brad] Root Cause Analysis is also going to be a big application area. Mike Lydon at ITC spoke on fail data back to device vendors even before the part was de-populated from the board. Some parts now have mutiple temperature or voltage sensors to collect data that can help understand the failure.

Working Group Operating Procedures

Deferred until 12/15 due to lack of time.

5. Schedule next meeting

Monday, December 15th, 2008, 8:15am EST

January 2009 schedule
[Ian] I guess we can't actually set the January schedule until we've resolved the meeting time and day. But in outline, do people think that the week beginning 5th January is too early for the first meeting of next year?
[Eric, Brad] No, that's OK.
[Ian] OK, tentatively, and on the assumption that we have Monday meetings, that will give us 5th, 12th, 19th and 26th January. We can firm up next week.

6. Any other business

[Ian] One thing I should have mentioned a couple of weeks ago was that Jim Webster has advised us that he is now taking retirement and is leaving the group.

7. Review new action items

Ian: Circulate the three options for revised meeting times to allow a decision to be made next week.
All: state your preference for new meeting schedule
All: review/comment on "Operating Procedures of the SJTAG Working Group"

8. Adjourn

Meeting adjourned at 9:22 EST
- Motion by Heiko, second by Carl W.

Many thanks to Heiko for help with these notes.

Respectfully submitted,
Ian