On Wed, 6 Oct 2010, gene heskett wrote: > I just spent an hour drilling more holes in the wall/floor here, so I could > get those two long extension cables to a direct from the mobo port. The > ports I used are on a breakout back panel slot, and add about 15" to the > cable length, plus of course the big lump from the connector junctions at > the slot filler. Neither one works, error -71. So, move those two to a > powered hub which adds about 4 feet of cable from the hub to the port, and > both are now working. So if the mobo ports have as much signal driver power > as the 4 port hub, then its the extra 15" of cable from the mobo to the > slot filler and the (low) impedance lumps at the slot filler sockets that are > killing it. Damn if I do, and damned if I don't. Maybe... Hardware faults can be very difficult to pin down. For example: I've got a USB-IDE disk drive that doesn't work when attached to my home computer. When plugged in to my office computer (using the same cable), it works fine. When I attach a USB flash drive to that cable in place of the disk drive, it works on the home computer. And when I replace that cable with a different one, the disk drive works on the home computer. Since replacing any of the three components fixes the problem, how can you say which is at fault? > I guess the upshot of this is that as an rf broadcast engineer, I have > learned that the VSWR present on USB cables is in no way, properly > terminated. It it was, then it should be somewhat like the scsi buss, > whose original specs 35-40 years ago claimed to have a max cable length > limit of 39 meters, if it was properly terminated on both ends to absorb > the echos from 5ns signal edges. There _are_ standards specifying what the USB physical and electrical characteristics are supposed to be. If you've got the right equipment, you can check the USB terminations. > But between the engineers and the production floor was a bean counter who > bought the cheapest +-20% parts he could buy, so we have had 35+ years of > scsi having a reputation that not even sacrificing virgins would make it > work reliably. Maybe it still has that rep, I haven't used it for a tape > drive in yonks, vtapes on a hard drive are easily 50x more dependable and > 100x faster to recover from. > > I have fixed a lot of scsi systems right, and without using up any hard to > find virgins, made them truly bulletproof. Bean counters will be the death > of anything they touch, could we start a bounty system on them? ;-) There's no question that a _lot_ of USB devices are designed to be as low-cost as possible, which means cutting a bunch of corners and giving up compliance to the spec. For many companies, if the device works okay when plugged in to a Windows machine using a standard cable, that's all they care about. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
