On Fri, 2006-02-24 at 11:54 +1030, Tim wrote: > Tim: > >> I've never heard of IDE using termination (despite understanding why > Mike McCarty: > > I suppose you mean ATA. Even the old MFM drives used termination, > > as do floppies. > >> transmission lines need proper termination). There are no termination > >> options on IDE drives, they seem to rely on tolerance (cable length > > Yes, there are. > Show me one where you select/enable/disable termination in some way or > another. I've never seen an IDE drive with any *options* for > controlling termination. Keyword being *options* (as in something that > you control). Not all termination needs to be optional. A lot of SCSI controllers handle it automatically (drives are another matter, although some single-ended SCSI drives did pull it off). > I won't argue that they don't need termination, but I'll say that it's > not an option, the drives and hosts are designed with what they need > as-is. Which goes some way to explaining why you can't use long cables, > you can't terminate drives optimally for how you've cabled them (one or > two drives), what they have has to work whether there's an extra drive > in the middle of the cable or not. Just because it's "not an option" it doesn't mean it's not there. It can easily be active termination which is tolerant of over termination or some other form of compromise, overdamped, termination which doesn't severely impact a short cable with few devices at the frequencies in question. > Ages ago I waded through the specs IDE/ATA, whatever you want to call > it, and don't recall user-configurable termination being discussed, > either. Nobody said anything about "user-configurable termination". Not all termination has to be "user-configurable". Especially with short cables that are reasonably damped. With a drive on one end and a controller on the other, you can be slightly underdamped and have no ill effect (having done this, designing some proprietary mirroring controllers years ago). Then, a second drive in the middle will introduce some reflection and result in the cable being slightly over damped but, still, the ringing is within acceptable limits. It's actually pretty easy to do with active termination and a limit of three devices (two drives and a controller) on a reasonably short cable. It can get out of hand if you put the lone drive in the middle and THEN you have a complete OPEN at the far end. That then forms a stub resonator that can ring like the bells of St Mary. Sometimes it will work (if you are at a reasonable spot in the middle of the cable) and sometimes it won't. With slight underdamping of the cable by the two device, you stand a pretty good shot of it working, just no guarantees. Generally, with two devices, one at each end of a short cable, that's the easy part of getting cables and signals to behave and you can be really tolerant of mismatches in that case. For IDE/ATA the case you need to optimize for would be the three device case, where you have a disturber in the middle. Get that right and the two device at the ends case is a no-brainer even without "user-configurable" terminations. This isn't SCSI where you can have 3 meters of cable with 16 devices on the cable. Mike -- Michael H. Warfield (AI4NB) | (770) 985-6132 | mhw@xxxxxxxxxxxx /\/\|=mhw=|\/\/ | (678) 463-0932 | http://www.wittsend.com/mhw/ NIC whois: MHW9 | An optimist believes we live in the best of all PGP Key: 0xDF1DD471 | possible worlds. A pessimist is sure of it!
Attachment:
signature.asc
Description: This is a digitally signed message part
