henry@utzoo.UUCP (Henry Spencer) (10/15/86)
Here's a quick summary of the comments people have made on my question. > Many of those smart modems send a string saying they've connected, > and at what speed. ... An awful lot of people have suggested this one. Give me credit for a bit of intelligence, folks; our modems are (*flame on*) professional quality equipment, not these dimwit-friendly junkboxes (*flame off*) and don't do such stupid things. When they connect, they raise Carrier Detect, period. Martin Minow suggested that badly-balanced lines can echo to the host to some degree; apparently DEC has patented the use of this phenomenon for testing some types of equipment! Several people suggested that the modem's speed-sensing transitions, or the beginning of carrier from the originating end, might be seen by the host as a framing error. Apparently old 300-baud modems in particular often are sloppy about oscillator startup. This sounded like a possibility, although the Model 100 modem is relatively new and I would be a bit surprised if the relatively fancy modems on our machine passed startup transients on to the host. Finally, as you've probably all seen, Brad Silva has definitely pinned the problem down to the Model 100's quite slow screen-handling doing a lot of XON/XOFF handshaking, which of course shows up as framing errors until the host finds the right speed. -- Henry Spencer @ U of Toronto Zoology {allegra,ihnp4,decvax,pyramid}!utzoo!henry
larry@kitty.UUCP (Larry Lippman) (10/17/86)
In article <7229@utzoo.UUCP>, henry@utzoo.UUCP (Henry Spencer) writes: > ... > Martin Minow suggested that badly-balanced lines can echo to the host to > some degree; apparently DEC has patented the use of this phenomenon for > testing some types of equipment! > ... Almost everybody - including myself - exceeds the RS-232C distance specification of 50 feet (actually, the spec says 2500 pF of capacitance, which usually equates to 50 feet of cable). Exceeding this specification can be an invitation to spurious character echo, especially if "homemade" cables are incorrectly wired. The most common wiring error that I have seen involves the use of multi-conductor cable having wire in pairs; some people think that TD (BA) and RD (BB) should be in the same pair - this is the WORST thing that anyone can do, since these leads are hardly a balanced pair! If you're going to use twisted pair cable, pair TD and RD each with ground if there are enough conductors; if not, at least pair them with some lead that is "static" in state - like DSR (CC) or DTR (CD). If you're really determined to "push" a RS-232C line to a long length, the best way is to use multi-conductor cable having individually shielded (i.e., coaxial) wires. An even worse echo problem can result through the use of RS-422 as a limited-distance modem. RS-422 lines should have a termination resistor. Not all RS-422 interfaces have this resistor, because they assume that the _other_ end will have the termination resistor. So, if you connect two RS-422 devices together, and neither one provides the termination resistor - chances are there will be trouble over any significant cable length. ==> Larry Lippman @ Recognition Research Corp., Clarence, New York ==> UUCP: {allegra|decvax|rocksanne|rocksvax|watmath}!sunybcs!kitty!larry ==> VOICE: 716/688-1231 {hplabs|ihnp4|seismo|utzoo}!/ ==> FAX: 716/741-9635 {G1,G2,G3} "Have you hugged your cat today?"