IBIS Macromodel Task Group Meeting date: 08 December 2015 Members (asterisk for those attending): ANSYS: * Dan Dvorscak * Curtis Clark Avago (LSI): Xingdong Dai * Bob Miller Cadence Design Systems: Ambrish Varma Brad Brim Kumar Keshavan Ken Willis Cisco: Seungyong (Brian) Baek eASIC: * David Banas Marc Kowalski Ericsson: Anders Ekholm GlobalFoundries: Steve Parker Intel: * Michael Mirmak Keysight Technologies: Fangyi Rao * Radek Biernacki Maxim Integrated Products: Hassan Rafat Mentor Graphics: * John Angulo * Arpad Muranyi Micron Technology: * Randy Wolff Justin Butterfield QLogic Corp.: James Zhou Andy Joy SiSoft: * Walter Katz Todd Westerhoff * Mike LaBonte Synopsys: Rita Horner Teraspeed Consulting Group: Scott McMorrow Teraspeed Labs: * Bob Ross TI: Alfred Chong The meeting was led by Arpad Muranyi. ------------------------------------------------------------------------ Opens: - Arpad reviewed upcoming ATM meeting dates, noting last week's decision to cancel meetings on December 22nd and 29th. He also proposed that we cancel the meeting on January 19th because DesignCon and the associated IBIS summit occur that week. The group decided to cancel the meeting on January 19th, 2016. -------------------------- Call for patent disclosure: - None. ------------- Review of ARs: - None. ------------------------- Review of Meeting Minutes: - Arpad: Does anyone have any comments or corrections? [none] - Mike L.: Motion to approve the minutes. - Michael M.: Second. - Arpad: Anyone opposed? [none] ------------- New Discussion: Discussion of language corrections regarding "ground": - Walter: [Reviewing "GND BIRD" presentation from 30 Jun 2015, to which Walter had added one new paragraph on page 4 (below)] - New paragraph: "GND is used in many places in this document as the signal_name of Pins that have Model Name GND. GND in many "buffer example schematics" use this name GND as a signal name such as VCC, VDD, VSS. The string GND in this document shall refer to either a Model Name in the Pin list or a signal name on one or more pins that coincidently also have Model Name GND. GND shall never be interpreted as the reference node in many SPICE simulators that is often called Node 0. Since IBIS defines signal_name as a component Data Book Name, we must assume that all Pins that have Model Name POWER or GND and have the same signal_name are assumed to be connected together on the components board or module. Each buffer can have one or two ground terminals (pdref and/or gcref) that are local nodes of the of the power distribution circuit of Pins that has Model Name GND. [Pin Mapping] is required to determine which signal_name is the local ground nodes of each Buffer." - Discussion: Walter stated that if we could agree on the new paragraph, then he thought things would be straightforward. Walter reiterated that Section 3 "GENERAL SYNTAX RULES AND GUIDELINES" in the spec stated that GND "must not be used" for anything other than the reserved model name used with ground [Pin]s, and that this statement had been violated repeatedly in the spec. He felt his new paragraph captured the ways GND had actually been used. Walter further explained his paragraph, and summarized by stating that in the IBIS Spec, GND is used as a reserved model name or as a signal name. Walter stated that his preference was to replace GND with VSS wherever it was used as a signal name, but he recalled that others had objected to that. - Arpad: I partially agree, but the spec does not prevent anyone from using GND as a signal name on a [Pin] that has a different [Model] name than GND. - Walter: Nothing prevents it, but in the spec it's only used one of two ways. - Radek: GND was also used in the context of a node in a schematic figure. - These had been replaced by ground symbols [mistakenly] when new figures were created. - Walter: In each of those instances, where GND is used "Vcc" is also used. - Vcc is a signal name. - In every case GND probably references a [Pin] where GND is a signal name. - The intent is that it's a node in the schematic with a low impedance connection to a [Pin] with signal name GND. - Bob R.: I think that's as far as we should go. - State that it's a reserved model name or a signal name. - Mike L. - Doesn't this mean that every place GND is used in the document we should say GND model or GND signal? - Walter - Our task in this review is to make sure it is clear in each context that it's a model or a signal. - Discussion: Walter stated that since IBIS defined a [Pin]'s signal name as the data book signal name, we can rely on two [Pin]s with the same signal name being connected. He asked if anyone disagreed that if two pins in the data book had the same signal name that they were in fact connected somewhere in the package? Arpad said he wasn't sure. Mike L. said it might not be specific enough, and that two pins with the same name meant they couldn't be assumed not connected, but might not guarantee they were connected. - Arpad - What is meant by "assumed to be connected"? - Everything is connected. The question is where and how. - Walter - We agreed that "connected" means a low impedance path between them. - Arpad: I could see two different groups of VCCH coming from totally different VRMs. - Walter: If they wanted to partition VDD, they would do things like VDDA, VDDB. - This is done a lot with memory chips where you have VDD and VDDQ. - If we can't agree that [Pin]s with the same signal name are connected, then this attempt to clarify GND is an impossible task. - Arpad: Why does everything hinge on this statement? - Walter: If you use GND or VDD on a schematic, it just means it's connected to one of the [Pin]s with signal name GND or VDD. - That's what [Pin Mapping] does in essence. It says, for this particular buffer instance, its pu_ref is signal name VDD. - Bob R.: Global statement should mention that GND or VDD could be bus labels. - Walter: Each buffer can have one or two ground terminals (pd_ref, gc_ref). - Bob R.: Theoretically, any of the reference terminals could be used. - Walter: As a practical matter, almost every buffer has just one, pd_ref or gc_ref. - There are unusual cases where it's not true. - These connections to the buffer are local nodes of the power distribution circuit of [Pin]s that have model name GND. - [Pin Mapping] is required to tell us which signal name or bus label is the local ground node of each buffer. - Without [Pin Mapping] we don't know how to connect C_comp because we don't know which [Pin] C_comp has to be connected to though the power distribution network. - Arpad: There is a special case in which [Pin Mapping] is not needed. - If the [Component] has only one Power and or one GND [Pin], it is clear. - Walter: Yes there is a special case. - If the [Component] only has one signal name associated with [Model] name GND, it could be multiple [Pin]s, and one signal name associated with [Model] name Power then we would know the local references for every buffer. - Discussion: Arpad asked if we had to worry about when [Voltage Range] was specified in lieu of the four reference values [Pullup_ref], etc. Walter said that those values merely specified the voltages found at the reference nodes while I/V and v(t) were measured. Bob R. agreed. - Walter: If we accept this, then we just have to: - Go through the spec and look for GND. - Revert schematics where GND was improperly converted to a ground symbol. - Deal with A_gnd and a few other things. - Walter: The point is that every device has its local ground. - All the I/V and v(t) are relative to that local ground. - Anything but 0 for [Pulldown Reference] or [GND Clamp Reference] is a meaningless concept. - Radek: That is not entirely true. - If you have a [Pulldown Reference] that is non-zero, then the v(t) tables include that offset. - I/V characteristics are the same, but for v(t) tables that is not true. - We really have to define what that local reference, I/O reference, is. - Original IBIS never did this. - Aside from [Pin Mapping] there was never any relationship between local ground for a specific buffer with respect to any ground pins. - Walter: When you measure, what is the voltage at the I/O referenced to? - The only thing that makes sense to reference it to is the local ground. - Input - gc_ref node. - Output - pd_ref node. - I/O - one or the other. - Once you have power aware simulations, this has to be defined. - Discussion: Bob R. objected to Walter's proposal to specify that gc_ref or pd_ref nodes were the local ground reference. He only wanted to state that it might be any one of the local reference nodes, and provided ECL as a counter example. Walter said we could add special cases. Radek said that if we didn't specify it then we ran the risk of inconsistent interpretation. Arpad suggested that we come up with a way to specify what the reference was, and stated that he was uncomfortable with Walter's assertion that GND in the schematic always referred to a GND signal associated with a ground [Pin]. Walter responded by reviewing various figures in the spec (ISSO figures among others) and pointing out that where GND was used the figures also used signal names like VDD. - Arpad: How should we move forward? - Do we have agreement, even on the problem statement? - Bob R.: We have identified exceptions to the "must not be used." - Bus labels, signal names, are perhaps the exceptions. - I would vote against it now though. It has too much stuff in it. - Some is contradictory to some existing [Model]s. - Separate the notion embedded in this that local ground must be tied to one of the pd_ref or gc_ref nodes. - Arpad: We will continue the discussion next week. - Thank you all for joining. ------------- Next meeting: 15 December 2015 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives