1. Opening Remarks John Lohmeyer called the meeting to order at 9:00 am, Monday, December 4, 1989. He announced that this was the SCSI-2 cable impedance testing working group meeting. The plenary meeting will commence at 1:00 pm. John introduced Bob Bellino, the chair of the cable impedance working group. The chair requested those in attendance to introduce themselves. Bob Bellino appointed Larry Lamers to record the minutes of the meeting. Bob Bellino informed the group that he had contacted Sun Micro, DEC, Unisys, HP, and IBM at three locations for cable testing. Also four terminator houses have been contacted about construction techniques: Aura Harbor, SCSI Tritec, Moldcon, Labage. 2. Cable Outside Diameters An a maximum insulation diameter of 0.035 inches per conductor and a maximum cable diameter of 0.45 inches was agreed to at the previous meeting. This agreement was not changed at this meeting. 3. DEC testing Bob Bellino presented the results of testing done at DEC Connect Labs. This indicates that DEC prefers solid polyethylene because of its parameter consistency and repeatability. The cable does not meet the minimum 90 ohms (it is 82 ohms). 4. Characteristics impedance requirement Bob Bellino stated that the 90 ohms characteristic impedance requirement for single-ended cables is impractical based on his discussions with system integrators and terminator houses. Most terminator houses prefer to work with solid insulation. Bill Spence questioned how Bob came to this conclusion. Bob Bellino stated that on checking the origin of the 90 ohm number, that it was set as an objective, and was not believed to be easily attainable. Bob cited several systems integrators who had tested and approved cables for use on SCSI systems that did not meet the 90 ohm requirement. The question arose of where did the 90 ohm requirement originated. Bob Snively answered that it came about from practical implementations on older SCSI systems and evolved into a standard. Also it was easily measured. Gary Murdock asked what has been done to get the impedance higher. Bob Bellino responded that changing from polyethylene insulation to foamed insulation could increase impedance. However, polyethylene has a .95 specific gravity, foamed TFE has a 2.2 specific gravity and the cost per pound is higher; thus foamed insulation is significantly higher in cost. Also molding foamed insulation is difficult because of its lower temperature rating. 30 AWG wire could be used to increase impedance, but using mixed sizes 28 & 30 AWG wire is a problem for termination houses. Solid polyethylene insulation has approved by Unisys, IBM, Sun Micro, and DEC, even though those cables do not meet the 90 ohm requirement. Bob Bellino stated that foamed insulation can be used if a metal can is used instead of pre-mold. This is more costly, and more difficult to produce. Termination houses require banded marking which is very difficult on foamed insulation. John Gibson of Berk-tec stated that Berk-tec does not have difficulty band marking foamed cable. Wills Xu stated that he had submitted cables to Sun. The cable that did not work was the composite 28 & 30 gage cable. Bob Snively stated that another vendors composite cable works, but it had a lower skew specification. Bob Bellino asked if the task group wanted to recommend to the plenary that the 90 ohm impedance requirement be reduced for single-ended cables. The consensus of the group was not to reduce the 90 ohm requirement until further study was done on what parameters determined a working cable. 5. Test Procedures A test procedure using a time domain reflectometer was agreed to at the previous meeting. Bob Bellino stated that there is concurrence on the test procedures among the following five vendors of cables: C&M, Belden, Montrose, Berk-tek, and Madison. Jim Fiala of 3M stated that a TDR did not necessarily measure impedance in an appropriate manner. Bill Spence requested that the test method be included in the SCSI document as an appendix. 6. Transfer Rate Gary Murdoc stated that the cable should be specified for 10 mega-transfers to provide for growth in data transfer rate.Bob Snively stated that the Sun Micro testing was not done at 10 mega-transfers, but the waveforms at 5 mega- transfers look like it may achieve 10 mega-transfers. However, some cables with good impedances exhibit anomalous behavior. This may be a noise problem related to the cable construction technique. This is still being investigated. 7. Kurt Chan's letter Kurt Chan stated that there appears to be a cause effect relationship based on the testing. The correlation needs further study and the test methods need to be established. Kurt requested that the measurement of cable delay/skew, crosstalk, and mutual and intrinsic impedance and inductance on SCSI cables be addressed by the group. John Gibson accepted an action item to draft a proposal for crosstalk measurements at near end 8. System Testing Gary Murdock questioned whether the a specification could be developed for a cable that would work in all 'legal' system configurations. He suggested that a systems test be specified. The question arose of who would do the testing, and what the interaction factors of protocol chips and the performance limitations of systems would have on the results. The consensus is that a specification for a cable is what the objective of the working group is and it should proceed in that directions. Bob Snively stated that Sun Micro will continue their work to determine the what other relevant factors may be needed to specify a usable cable for SCSI. 9. Transmission Line Nomographs Jim Fiala of 3M presented a transmission-line impedance nomograph, way to have a single parameter to specification cable. These work if the correct test methods are used. The line drivers see the capacitance, specification max capacitance per foot, and the max deviation of capacitance. Bill Spence challenged this, stating that the drivers see the impedance, and capacitance just slows down the propagation. Then the skew and impedance fall out. The material is the other factor that determines the cable. He stated that the TDR does not do a good job of measuring impedance. Bob Snively stated that the problem with nomographs is that the 100 to 1 range of frequencies blurs the line on the nomograph considerably. 10. New 3M flat cable Bill Spence presented preliminary test results of a new 3M cable at 24 meters in a single-ended environment (see X3T9.289-148). The point is that single- ended cable is viable. This new cable met the 90 ohm requirement and worked at 24 meters. Gary Murdock raised the question of whether the ground shifts and common mode problems could be handled at 24 meters. 11. Color code Should there be a recommended color code for cable? Bob Bellino stated that the coloring could affect the electrical properties if carbon black is used in the pigments. 12. Recommendations 12.1. No change to 90 ohm requirement at this time 12.2. Continue search for relevant parameters 13. Next Meeting January 8, 1990 at 9:00 am at Red Lion in San Jose. Larry Lamers will take over as chair of the group.