PTIM Workgroup Outcome


Bookmark and Share

PTIM Working Group Breakout Session

April 1st, 2009 - SEMI World Headquarters – San Jose, CA

Attendees:

Ray DiGiacomo

Kevin Hoffman

Thijs Waardenburg

Chas Mikolajewski

Nils Snekkeviik

Ted Gutfeldt

Todd Lappin

Paul Rowbottom

Thomas Herbst

Burnie West

Edward Smith

Luke Schreier

Notes:

Meeting started with introductions and an agenda focused on bringing past and new participants of PTIM efforts up to speed. PTIM is a carryover from the STC working group effort begun in early 2007. Format for delivery was a powerpoint presentation summarizing past activities.

Comments on the presentation:

  • Slide 7 on the scope of PTIM: concern expressed about focusing PTIM on “lower-performance” applications.
  • Point: Shouldn’t it be focused on higher-performance capability that isn’t currently available in ATE? Bringing new, cutting edge capability into a PTIM format would make PTIM very viable and more likely to be successful.
  • Counterpoint: If only the highest-end instrumentation is implemented in PTIM, it might make the development cost of a PTIM device prohibitive compared to the relatively low volume expected from sales of the device.
  • Point: PTIM won’t be successful if it’s only reproducing capability already found in ATE.
  • Counterpoint: A PTIM use case might be adding simple measurements that are not readily available (in a cost-effective manner) on the current ATE platform in use. Lowering the barrier to entry would also create a larger volume of available PTIM offerings.
  • Decision: leave scope of PTIM undefined.
  • Slide 11 on the choice of leveraging existing standards: debate over whether the market for semiconductor-specific PTIM platform is unfeasible.
  • Point: If PTIM is defined right and the development cost is sufficiently low, an independent developer with the right expertise wouldn’t need to sell hundreds to make a profit. Forcing the developer to conform to a standard like PXI or LXI could add significant development overhead and cost.
  • Counterpoint: Finding a way for PTIM to leverage PXI or LXI will open the door to thousands of existing products and dozens of vendors as solution providers in the space. It would also provide alternate industry channels for PTIM devices aside from semiconductor (as both PXI and LXI sell across many verticals).
  • Question: Is success of PTIM based on new capability it brings to ATE or available products to choose from?
  • Question: Shouldn’t USB also be included in this list of instrument standards?
  • Decision: none sought past discussion.
  • Slide 12 on gaps of existing standards: why shouldn’t instrument vendors be forced to absorb some of the cost in making a PTIM? By keeping with existing standards, it will only be ATE vendors and customers that pay.
  • Datapoint: Advantest developed a PXI carrier module for the T2000 that was largely unsuccessful.
  • Slide 17 on removing performance board option: Although Q-Star has developed something that could fit a definition of PTIM, their effort could be considered significantly different from the other three approaches, so it makes sense to eliminate performance board PTIM from ongoing discussions.
  • Slide 18 on software (local vs. bridged): debate over which option would be more feasible for PTIM.
  • Point: Local control keeps processing independent of host ATE PC and therefore has less integration challenges. It also keeps it operating-system independent.
  • Counterpoint: Bridged control leverages native software driver of the instrument, making the experience of using it more “off-the-shelf”.
  • Open question: which is faster – Ethernet or PCI Express for data transfer? Gigabit Ethernet is ~125 MB/s theoretical while PCI Express Gen 1 x4 is 1 GB/s theoretical. With Ethernet, though, local control is assumed and therefore pre-processing would likely occur, making it more of a “results” bus instead of a “data” bus. This would make the burden of transferring large packets of data smaller.
  • Point: Ethernet is easier to integrate with off-the-shelf components that PCI or PCI Express.
  • Counterpoint: For higher-speed Ethernet, processing power is required to manage traffic. Ethernet (and a local processor) also require software development to packetize and send data over the cable to the host.
  • Open question: is there any reason you’d limit PTIM to one or the other? Why couldn’t both be possible?
  • Decision: none sought past discussion. Follow up with more detail in a future meeting.
  • Slide 20 on STC PTIM Charter: debate of wording
  • Using the word “result” in the last sentence of the first paragraph implies PTIM effort will make existing instruments PTIM compatible. A better word might be “enable”, meaning the effort will define a class of products that may or may not already exist.
  • Also, to say the instrumentation should be broadly available off-the-shelf is too limiting. It should be broadly available or easily designed.

During discussion of the presentation, the STC WG’s goal of producing a “guildelines” document as opposed to a “specifications” document was debated. Group sentiment was that a guidelines document does not define the problem enough to be usable. A specifications document will clearly define compliance and maximize adoption with less ambiguity. No conclusion or vote was taken on this topic.

There was a plan to review the previous progress on the guidelines document, but time expired before this was possible.

Conclusions/Next Steps:

  • The scope of PTIM should be opened up to any type of measurement capability. It doesn’t make any sense to limit it.
  • Next meeting should review capability of PXI, LXI, and other existing standards to get all group members on the same page with regards to the capabilities of each platform.
  • Tentative plan is to next meet at Semicon West in July
  • Email to follow meeting to confirm individual members desire to participate in future efforts.