******* IBIS-connector specification updates ***** 1-30-98 , KC, GP - Added naming to [Begin_Cn_Pin_Map] and [Begin_Cn_Phy_Map] keywords - Added [Begin_Cn_Pin_Map] and [Begin_Cn_Phy_Map] names to the [Begin_Cn_Model] keyword usage to allow re-use of the pin and physical maps and to create consistency with sample models that had already been created. - Made SGR optional for SLM models - Change the [End_Cn_Model] keyword usage to have the name of the model optionally follow it instead of required. 1-26-99 to 1-29-99 versions 25 to 30, KC, GP - Cleaned up the Swath method so we think it will work. - Removed the Swath_no_sim keywords - ran spell checking - Added numerous usage rules (many hundreds lines of changes) - Added the auto_map feature to allow variable size connector models. 12-15-98 ver 24, KC - Modified the examples in Begin_Cn_Model - Added descriptions of the Cn_Stub and Cn_Section keywords. - Increased the number of stubs allowed on a line from 3 to 10 - Removed the Begin_Cn_Matrix and End_Cn_Matrix keywords as they were redundant with the Begin_Cn_Section - Add and modified the examples of Begin_Cn_Model to included the above changes. 12-11-98 ver 24, KC ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ [Begin_Cn_Model] * Modified the examples for stubs in the [Begin_Cn_Model] [End_Cn_Model] keyword. | | VALID | <- (1 or more series matrixes) -> | Pin >------------------------------------< Pin | |stub1 (1 to 3 stub matrices) | | VALID | | Pin >------------------------< Pin | |stub1 |stub1 | |stub2 | |stub3 | | INVALID (more than 3 series stubs) | | |stub1 | Pin >----------------------< Pin | |stub1 | |stub2 | |stub3 | |stub4 | | INVALID (parallel) | | |stub1 | Pin >----------------------< Pin | |stub1 | |stub2 [Begin_Cn_Pin_Map] * Added a clarification to the [Begin_Cn_Pin_Map], [End_Cn_Pin_Map] keywords as follows: | Description: | Describes the index into the matrix for which a given pin applies. Also added the following usage rule to above | Simulators should consider pin names with case in-sensitive compares. To | make human comparison easy it is recommended that all uses of the pin name | in the file utilize the same case. Also changed the Required flag to Yes because we no longer support the old SLM syntax and we now do everything using matrices so this is required. Required: Yes Also added the following notice regarding special signal names and usage. |NOTICE: The special Signal names PWR, GND, and PWRGND may be used to | indicate that this pin should be connected to either a Power pin like | +5V, a ground pin, or (either a power or ground pin). This information | may be used by the simulator or PCB layout tool to provided a warning | that the connector is not following an established industry standard | layout. | This power ground pin information SHALL NOT be used by the simulator | for any purpose which effects a simulation. |============================================================================== [Begin_Cn_Map] * Added a clarification to the [Begin_Cn_Map], [End_Cn_Map] keywords as follows: | Description: | Describes the physical layout of the connector for use with Swath operator | or general documentation. Also added the following usage rule to above | Simulators should consider pin names with case in-sensitive compares. To | make human comparison easy it is recommended that all uses of the pin name | in the file utilize the same case. * The group requested the keyword PWRGND be added to allow the simulator to warn the user if the connector is not being used with required pins. Added PWRGND to reserved word list * The group requested that SLM/MLM flag be added back to spec so we could determine when the signal to ground ratio was needed. | Usage Rules: | Model name is required and must be less than 20 characters. | | Model type is required and must be either SLM or MLM | SLM indicates this model is a "single line model" and | all values must be specified using the Diagonal_matrix either | to define the entire connector or using the swath method. | Use of Sparse_matrix, Banded_matrix, and Full_matrix is an error | in an SLM connector model | The Signal-Ground-Ratio or "SGR" value is required if Model type is SLM. | For example: 3:1 would indicate there is one (power or ground pin) | for every 3 signal pins. The SGR is the value that was assumed | when the model was created. The simulator may use this value to | provide a warning if the model is being used with a high or lower SGR. | If the simulator provides such a warning a mechanism should be provided | to allow a user defined amount of tolerance. For example some areas may have | 4:1 and others 2:1 with the nominal value being 3:1. | It is recommended that the simulator make an average SGR display available | to the user if the Model type is SLM. | | MLM indicates this model is a "multi line model" and is | a coupled or partially coupled model. | Any combination of Diagonal_matrix, Sparse_matrix, Banded_matrix, or | Full_matrix may be used in an MLM model. | | NOTICE: Simulators SHALL over-ride all (power or ground) pins in an SLM connector | model and replace their entries with 0 ohm resistors as follows: | R = 0.0 ohms, | L = 0.0 nH | C = 0.0 pF | Connector model creator's SHALL create SLM models with the knowledge that | the simulators shall automatically replace all ground or power locations in | the connector with 0 ohm resistors. Added SLM/MLM Model type and SGR to the [Begin_Cn_Model] keyword as follows: [Begin_Cn_Model] modelname SLM 3:1 [End_Cn_Model] modelname Or [Begin_Cn_Model] modelname MLM [End_Cn_Model] modelname 12-3-98 ver 23,KC ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * Added the key word Diagonal_matrix to provide a smaller version of the Banded_matrix with 0 width since it is used frequently. * Removed the SLM and MLM model types. It was decided by the group that a Diagonal_matrix could be used to describe an SLM model without specifically indicating it is an SLM model. This simplifies the connector description and reduces the file size. * Changed Cn_swath_location to Begin_Cn_Swath_NoSim_Locations, End_Cn_Swath_NoSim_Locations * Added "Cn_Horizontal_Swath, Cn_Vertical_Swath, Cn_Grid_Swath" sub parameters to the [Begin_Cn_Swath] keyword 11-17-98 ver 22 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * Updated this document * Updated example to include A, B, C FULL and Banded Matrices. * Adjusted "Single Matrix: as a zero width banded matrix. Somebody please double check this... For a single line model... "Bandwidth = 0"... right? * Suggest we first define swath as a Horizontal ONLY... AFTER we get this hammered out... Let's then start the other potential swath options (i.e. for PGA sockets). We will need a different example to represent other "swath options. * Installed cn_ prefix in swath keywords, [Begin_Cn_Pin_Map], other.. * Worked swath in the example... included "keywords" at bottom of specification * Changed [Default_System_Z] 50 to [Cn_Z] to better reflect the proposed usage Comments and Yet to be done... * Any feedback from the committee with regards to not including the "sparse" matrix? * Define matrix values.. The solver will need to know what type of matrix is generated.. Suggest that we limit the type to a single variety (i.e. for inductance... use a PARTIAL INDUCTANCE matrix) Comments? ~~11-01-98 ver 21 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * Implemented IBIS Look and Feel * Incorporated 2.1 Package Model * other?? 10-30-98 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * Changed Begin_Connector_Model to Begin_Cn_Family_Model Begin_Model to Begin_Cn_Model to remove the confusion factor between the two names * Added ASCII ART to Begin_Cn_Model / End_Cn_Model * Cn_Section - Starts a series Section * Cn_Stub - Starts stub section.. The end is to be assumed to be connected to an open circuit. * Added Legal disclaimer boiler plate * Agreed to break update notes out of the specification to improve clarity in specification. It will go in separate file on ftp site. for example: update_19.txt * Agreed to break models out in separate files and reference the file names in the specification so if the spec. and files are separated at least someone reading the specification will know to look for the example files. * Renamed exmpl_19.txt to exmpl_19.cnn IS "cnn" used anywhere else .ibiscnn did not save to the Novel server where I am achiving these updates. 10-14-98: Changes from previous version: ************************************************************************* * Still not addresses Swath (Address at meeting) * "Long" keywords... (Address at meeting) * Cn_ Issue (Address at meeting) * [Notes]: 4 lines max * Added better stub definition * Removed misc. "humor" related statements * Added ASCII to jpeg drawing .. removed GIF format per Compuserve concerns. * Begin_ Section name length defined as 36 alphanumeric characters * Renumbered Begin_Connector_Model_List note. * Added "Single_matrix" to matrix type to account for single Line Models * Added "Section_A" ONLY in example.... IF correct, then we can add B and C.. Will need to change the Connector Model after B anc C have been added * Added ROW ro begin connector map... required because length could exceed 80 * IBIS_Connector_Model_Version description changed to reflect the version of the IBIS spec that this file conforms * [File Rev] is now the version of IBIS connector model as defined by the Manufacturer * added "/" where required in the example 10-14-98: Changes from previous version: ************************************************************************* * Added ability to "Stub" with in each connector model. **** The following items were change in response to dividing the model into an "example" and a "specification". When division was done... It sure seemed like the following changes would make sense to keep with the "keep it simple" goals.... I realize that with some of these changes, I might have increased some of the burden on the simulator companies.... However, as a model creator... I think this greatly simplifies the model creation process. I hope that the increased burden on the simulator companies is not too severe.... My limited experience with writing code would suggest that it's not too bad... but I would definitely like to encourage others to review this stuff. * Removed Uncoupled model type... This kind of model can be represented by standard matrix... ... representation using the "Banded_Matrix" type. Variable lengths not required for the connectors. * Separated Example from the Spec. * Removed Bandwidth from Resistance Matrix. We are not addressing Frequency variable resistance in this specification... ... as such I don't think this is needed at this time. * Removed "File Revision" because "[IBIS_Connector_Model_Ver]" should be enough.. Right?? * Remove Number of Cascaded sections... I think the simulator would be smart enough to use the... ..."Begin / End Key words to parse the section information * Changed [Begin/End_Mated Model] to [Begin/End_Model] . [Begin/End_Model_List] takes... ... care model "mating" description. * Removed [Begin_Coupled_Cascade_Section] /[End_Uncoupled_Cascade_Section] The sections... ... used and their respective matrix representations will determine this parameters * Removed [Begin_Unmated_Side_A_Model] /[End_Unmated_Side_A_Model] [Begin/End_Model_List] ... ...takes care model "mating" description. * Revised "SLM" definition * Need: ** SIZE LIMIT for Description ** SIZE LIMIT for Disclaimer ** SIZE LIMIT for [Begin/End_Model] name ** Geometries such that EM matrices can be generated and installed into the example 9-18-98: Changes from previous version: ************************************************************************* ** Swath matrix has been unwrapped... The same number of PINS are modeled for each Section ** IBIS Line length limit is 80 characters ** For the connector Map Every Row must have same number of entries.... use NA where pins are missing... ** Keep pin names in the swath definition ** Swath requires that a model is defined wide enough where edges of the swath are no longer an issue ** Initially, we as a group, will not address any thing but left.... center.... right... once we get this down we might incorporate other concepts * removed now obsolete "swath level" * Removed Swath option if it's an option.... swath information will be supplied... if swath is not supplied it shouldn't be used * It's been confirmed that swathing wil be handled by simulator... the spec/model will prove "swath rules" that can be performed on the given matrices * Added Keywords... [Begin_Pin_Map], [End_Pin_Map] *** Deep Thoughts... * I think we talked about the "variable length" from Single line model. Didn't we decide that we should remove this option? 9-14-98: Changes from previous version: ************************************************************************* * Per our last conference call, the goal of this revision is to refine the Swath concept per the conference call discussion. * Also "Swath description" names were changed.. * First, we discussed the option of "unwrapping" the swath matrix...This is thought important because an unwrapped swath could apply the Swath function through multiple cascaded section models... This would including single line sections * In order to unwrap the swath matrix we MUST assume that... ALL sections have the same number of pins The following keywords were added to further refine the SWATH concept. Added Keywords: [Begin_Model Description] Starts the description of the MODEL [End_Model Description] Ends the description of the MODEL The MODEL description would provide potential swath size limits [Begin_Connector_Description] Starts the description of the physical CONNECTOR [End_Connector_Description] Ends the description of the physical CONNECTOR The CONNECTOR description would provide swathing limits (as far as number of pins) The following are used in the MODEL and CONNECTOR descriptions [Max_Number_of_Conductors] The largest amount pins the connector will have [Min_Number_of_Conductors] The least amount most pins the connector will have [Columns_of_Pins] This might not work for all connectors... but would be required for connector modes that would use the swath command [Rows_of_Pins] This might not work for all connectors... but would be required for connector modes that would use the swath command 9-03-98: Changes from previous version: ************************************************************************* - Added Swath Matrix Concept / Proposal - In real life "Swath" will probably need more than 3 columns... however to keep the example easy and consistent... I will assume that the 6 pins represent 2 rows x 3 columns - Added Multi-Section link scheme - Redefined Begin Connector Model List to include connector Name/number. My thought is each model is made up sections and that each section could be coupled or not coupled - Proposed separate sections (unwrapped matrix from a specific model) such that multiple models can access one or more sections. Also, This makes it easier to invoke the "swath" concept.... Assuming that every section in this model will reference the same relative connector pins Should we "unwrap" each matrix, too????? - Incorporated capacitive pads/vias on "Begin_Pin_Side B": - added keywords to the bottom of the keyword list below Future points of discussion (i.e. "deep thoughts") ************************************************************************* - Should we make it a requirement that a simulator treat each section as a "symmetrical filter". By this I mean, for example... a single section model, that is coupled OR un-couple should show the same electrical response to an applied AC signal traveling from left to right OR right to left. - The way this specification is shaping up.... this could be an interconnect model... not just a connector model. For instance.. I could have (a PCI bus): (representing a trace between bus driver and connector * an uncoupled single section transmission line (~0.5" in length) to (represents nonloaded connector) * an unmated coupled section to (represents the traces between connectors) * an uncoupled single section transmission line (~0.5" in length) to (represents nonloaded connector) * an unmated coupled section to (represents the traces between connectors) * an uncoupled single section transmission line (~0.5" in length) to (represents loaded connector ... video card maybe) * a mated coupled multi section connector model to (represents the traces between connector and terminator) * an uncoupled single section transmission line (~0.5" in length) Only thing required would be to add key words that would cascade models (i.e. [Begin_Cascade_Model, End_Cascade_Model], etc.?) Notes: ************************************************************************* 1) We propose this be a separate appendix to the main IBIS specification and a completely separate file. This should make the specification much easier to understand for model developers only interested in making a connector models. 2) The default comment character for IBIS is the pipe character