================================================================================ IBIS INTERCONNECT TASK GROUP http://www.ibis.org/interconnect_wip/ Mailing list: ibis-interconnect@freelists.org Archives at http://www.freelists.org/archive/ibis-interconn/ ================================================================================ Attendees from March 21, 2018 Meeting (* means attended at least using audio) ANSYS Curtis Clark* Cadence Design Systems Bradley Brim Intel Corp. Michael Mirmak* Keysight Technologies Radek Biernacki Mentor, A Siemens Business Arpad Muranyi* Micron Technology Justin Butterfield*, Randy Wolff* SiSoft Walter Katz*, Mike LaBonte* Teraspeed Labs Bob Ross* Michael Mirmak convened the meeting. No patents were declared. Justin Butterfield took minutes. Review of Minutes: - Michael called for review of the minutes from the March 16 meeting. Mike LaBonte moved to approve the minutes. Randy Wolff seconded. The minutes were approved without objection. Review of ARs: - Walter Katz to create an example with a local reference using A_gnd at the [Model] level. - Michael noted this is done and was discussed in ATM. - Arpad Muranyi to create an example with a local reference using A_gnd at the [Component] level. - Arpad reported he has prepared a few slides, which he would like to discuss on the topic. Opens: - None. Summary of ATM: Arpad stated that Walter went over his A_gnd examples. He suggested we review Walter's examples along with the pictures that he sent. Michael asked if there were any changes to the examples. Arpad stated that they made some changes in the ATM meeting, and these were noted in red in an additional email sent by Walter. Walter agreed it would be worthwhile to go through Arpad's pictures. A_gnd Referencing Discussion: Arpad stated the question was if A_gnd is local to [Component] or [Model]. Michael noted we are trying to understand the appropriate level of hierarchy, or if it is required to define where A_gnd is local to. Arpad stated he took Walter's examples and represented them in graphic form. In the first examples, the power is connected from Pin to Pad/Buffer. He tried to use the same names as Walter, but he did not have a way to represent the Pin Mapping graphically. For the buffers, he wasn't sure if the buffer was a classic I-V curve buffer or a BIRD158 buffer, but he assumed an I-V buffer. In the second example, Walter added a W-element, which carries the power and has a node 0 type reference. In the third example, the ground based model uses a W-element to connect only the power, VDDQ and VDD, and the reference is connected to node 0. Walter commented that the ground based model should have a ground symbol on the VSS and VSSQ at the pin side. Arpad agreed and stated he will fix the ground based drawing and send out the updated slides [AR]. Bob Ross noted that the W-element does not have to use node 0 on both sides. Arpad replied that is true, but he would like to keep the examples simple. Michael asked what the small 0's in the drawing are supposed to represent. Arpad replied that these are node 0. Arpad stated for Walter's Set1 and Set2 examples, he has the reference nodes of the W-elements connected to the VSSQ on both the pin side and pad/buffer side. Walter noted that this is not correct. For both Set1 and Set2, the assumption is that some place else there is a connection from Buf_VSSQ to VSSQ at the pin side, and this connection supplies the current. Arpad asked about the subcircuit definition from Walter's email. Walter replied the one reference for the W-element is the reference for the model. Arpad asked if he should have done the thing same as the Set2 subcircuit. Walter noted both subcircuits should reference Buf_VSSQ. The connection between VSSQ and Buf_VSSQ would be handled in a separate PDN model. He stated, in practice, W-elements reference node 0, and we are not going to resolve that issue. Walter noted in his Set3 you have two different Interconnect Models, and there is only one buffer per Interconnect Model. In Set3, you could use Pin Mapping to connect the references to Buf_VSSQ or Buf_VSS, but in Set4 with one Interconnect Model for two buffers, it is not clear how to connect the reference. Arpad stated he has his own examples using both I-V curve buffer models and BIRD158 buffer models. He added the triangle symbol to make the A_gnd clear. For the I-V curve buffer models, he included the PC_ref, PU_ref, PD_ref, and GC_ref terminals. The first case has all five buffer terminals connected through the W-element. He has the reference connected to A_gnd. There are three A_gnds in the [Component]. He proposed that the A_gnd nodes are separate for each [Component], but A_gnd would be the same node inside one [Component]. The second example uses one big W-element with one A_gnd. In the third example there is no power, and A_gnd is the same for the buffers and W-element reference, and the EDA tool can make the connection. On the next slide, Arpad changed the references of these examples to a "DFN_Exposed_pad" pin. The model maker could spell out the connection to this DFN_Exposed_pad, and the reference is connected to a specific pin. For the BIRD158, you would have to connect the left side to A_gnd. On the right side package model, the model maker used the DFN_exposed_pad. Arpad stated that if the reference of the W-element is connected to DFN_Exposed_pad for BIRD158, this is a problem. Walter stated we have demonstrated that using A_gnd is a mess for the EDA tool simulator to figure out. A_gnd should be the simulator reference node. If model makers do not want to do this, they can connect the reference to a specific terminal. Bob stated that BIRD158 is separate issue, and a simple solution can be applied. Arpad asked what Walter is proposing. Walter replied he would like for A_gnd to mean node 0. Michael asked if A_gnd would be a universal reference node. Walter replied, yes it would. Arpad thought that this would be too restrictive, as this would force every S-parameter to use node 0 as the reference. Walter noted that the user is not forced to use A_gnd. Walter noted for BIRD158 there is no current going to through the reference, so the connection does not matter. Bob noted for BIRD158 you would not connect the supply power through the package, but you could use a specific reference. Michael asked Arpad if BIRD158 is not a consideration if he has the same concerns. Arpad replied that if there is a way to treat A_gnd in a generic way, he would much prefer to do that. Next Meeting: The next meeting will be March 28. Arpad moved to adjourn. Bob seconded. The meeting adjourned without objection. Task List BIRD189.5 editorial additions/changes to be completed: