desimone@tom.columbia.edu.UUCP (02/06/87)
Hi...I'd like to know if anyone out there has a good method for testing modems for peformance. When it's time to buy a modem I usually don't do much more than get some demos, plug them in, and see how each handles. I'd like to have a more structured approach that would really give them a good once-over. Any suggestions? Thanks... Salvatore
berger@clio.UUCP (02/09/87)
Do you have access to any specialized test equipment? It's not easy to test the important characteristics of a modem unless you can control the characteristics of the phone line.
wdb@quando.UUCP (02/11/87)
/***** quando:comp.dcom.mod / columbia!desimone / 6:19 am Feb 6, 1987*/ Hi...I'd like to know if anyone out there has a good method for testing modems for peformance. When it's time to buy a modem I usually in one of the last BYTE-issues (Nov or Dec 1986) was a test of about 20 modems, which were tested also in the technical aspects of line noise etc. perhaps you'll find soemthing about your problem there... with best regards Wolf ------------- Usenet: wdb@quando.uucp ..!unido!quando!wdb
zhahai@gaia.UUCP (02/24/87)
In article <22500002@quando.UUCP> wdb@quando.UUCP writes: >in one of the last BYTE-issues (Nov or Dec 1986) was a test of >about 20 modems, which were tested also in the technical aspects >of line noise etc. perhaps you'll find soemthing about your >problem there... >with best regards Wolf The Byte modem article was interesting, but my recent experiences with 2400 bps modems make me suspect one "flaw" in particular. I have been discovering that many 2400 bps modems are very particular about which other modems they connect well with - and these "predjudices" are more or less consistent over many calls on the same lines. None of the modems I have tried here consistently give good connections with all of the 2400 modems I wish to call (this is all local calls, by the way, not long distance) with some working better with some systems and some with others. Thus the Byte tests, which tested many modems and many simulated and controlled line conditions, but which always used the same target modem at the other end (an AT&T model considered by the author to be very good) may have test as much for good match with that particular target modem as for general quality. I assume there is some correlation with general quality, but I really don't know how much. (interestingly another AT&T model came out best in the tests - either because it was truly the best technically, or because it was the most compatible with the fixed test modem...). I begin to think that what we need are the controlled line impairments of the Byte tests (assumeing they are good reflections of real impairments, which I cannot judge), and then do a connection matrix - a lot of work. This would be a full matrix, not just a triangle, because it really seems to matter quite a bit which side initiate the call. I sure would like to see a few such matrices (reflecting different line conditions and giving error rates) using the common 2400 modems (USR, Hayes, CTS, Basic Time, Vadic, Anchor, THE, AT&T, etc.). Actually, I have found the same thing with 1200 (a Hayes 1200 which works well with most connections has mucho noise with a Zoom Modem which works well with most connections, etc) - but the problems are much less frequent. The black magic aspect reminds me of working with a lot of systems with not- quite well aligned floppy disk drives. A royal pain. Unfortunately, we have little choice since modems don't get tuned - and we often can't get other systems to replace their modems <especially since some other user would probably suddenly get problems...> ARRGH. -- Zhahai Stewart {hao | nbires}!gaia!zhahai