tim@cit-vax.Caltech.Edu (Timothy L. Kay) (09/16/89)
I am hooking some Newbury drives to an Iris 80GTB, and I am running into just a bit of trouble. When I have four 380 MB drives hooked up, everything works great. If I hook up a fifth, I am overloading the bus somehow, I think. When it boots, I get the message unexpected transfer phase: state=4b phase=20 resetting scsi: hardware error It seems to happen only when that fifth drive is on the ribbon cable. I have moved the drives to different connectors on the cable to verify that there are no bad connectors. Also, when I am in the five drive configuration, I can use all five drives, but the bus gets errors infrequently. Anybody have any ideas what could be wrong? Is there a document that tells me about maximum cable lengths, more information about termination, etc.? Thanks in advance, Tim
olson@anchor.sgi.com (Dave Olson) (09/18/89)
tim@cit-vax.Caltech.Edu (Timothy L. Kay) writes: >I am hooking some Newbury drives to an Iris 80GTB, and I am running >into just a bit of trouble. When I have four 380 MB drives hooked >up, everything works great. If I hook up a fifth, I am overloading >the bus somehow, I think. When it boots, I get the message > unexpected transfer phase: state=4b phase=20 > resetting scsi: hardware error The most likely problem here is a SCSI bus problem. Check to be sure that ONLY the drive furthest from the CPU is terminated, that stubs off the main ribbon cable are no more than 4" long, that the overall cable length is less than 20 feet (including cabling inside the driver tower, and cabling from the CPU board to the tower). Finally, if all these are OK, if your system is more than about 6 months old, you are probably out of luck unless you want to spend the bucks to upgrade to a new power supply/drive stack. A number of the earlier 4D machines were cabled in violation of the SCSI spec, and changes made in manufacturing made it progessively worse. You are lucky to get as many as 4 drives working reliably if this is your problem... Dave Olson It's important to keep an open mind, but not so open that your brains fall out. -- Stephen A. Kallis, Jr.
terryk@encore.Encore.COM (Terence Kelleher) (09/18/89)
Another Possibility is a collision of SCSI target IDs on the SCSI bus. If two targets or a target and a host are both set for the same IDs, there will be confusion and bad states will be seen. Terry Kelleher