X3T9.2/87-99 To: John Lohmeyer June 30, 1987 From: Ralph Schultz Subject: Terminator Power Several years ago, I brought to the committee a proposal which was born out of the confusion in the implementation of SCSI Terminator Power. This proposal was discussed, and resulted in the present wording in SCSI-2. At a recent working group meeting, the subject again raised its ugly head as a host adapter vendor objected to the mandatory requirement that the initiator supply terminator power to the TERMPWR pin. Controller manufacturers apparently have not taken advantage of the wording in SCSI-2. Many of the controllers still have terminator packs mounted on the controller board. Some vendors still provide a jumper to permit the target to power the TERMPWR pin. And still others do not power the terminators from the TERMPWR pin, but rather connect the packs to the controller +5v supply. In light of the persistence of the problem and the apparent lack of implementation to the agreed-to revision, I should like to re-propose the original implementation of Terminator Power for SCSI-2. It seems to me that the best implementation is to require that both initiators and targets be capable of supplying terminator power to the TERMPWR pin. The intended implementation of this would be that both initiators and targets have provision (i.e. jumper) to supply +5v to the TERMPWR pin through a diode (or similar device). The controller power requirements would be XX ma. @ 5v if the jumper were not installed and (XX + 800) Ma if the jumper were connected. (Incidentally, the jumper could take the form of installing a diode into the PC board). Likewise, the power requirements for the initiator would be ZZ Ma if supplying TERMPWR and (ZZ - 800) Ma if the jumper were not connected. With this scheme, the TERMPWR configuration for a particular system would be connected at installation by the system integrator or recommended to the user by the supplier. If the spirit of this proposal were adopted, then subjects like excessive current limiting would be moot. Likewise, problems with maximum current from the +5v backboard pin would be compensated by powering from someplace else. Issues like system integrity with some device powered down would also be moot, as the integrity is defined by how the system is configured. To implement this proposal, I would like to recommend the following changes to Section 4.4.3 (page 4-12) of SCSI-2 Rev. 1: Delete the first and third sentence of the first paragraph. Add the following first sentence: "All SCSI devices shall have the capability of supplying terminator power to the TERMPWR and TERMPWRB pin(s). This capability shall be individually configurable on each device." Change the IMPLEMENTORS NOTE to read: "IMPLEMENTORS NOTE: The selection of which SCSI devices actually supply terminator power is determined when the system utilizing these devices is configured. The configuration may take the form of connecting a jumper on the initiator or target, although the standard does not specify the means of implementation. It is recommended that the initiator(s) be configured to supply power in order to preserve powered terminators in the event a target device is powered down. The configuration should insure that the maximum current available on the TERMPWR and TERMPWRB pins is limited to less than 5 amps."