On Mon, 3 Feb 2003, Ken Hohhof wrote: > > > And sometime, their pop account are locked and their are asked to > retype > > > password but it is not valid.
It's essential to make sure the users are aware that the mail client stores their passwords and will ask for it to be reentered if there are _any_ problems, in case it was a password error (as well as to avoid slamming the server with repeated requests if something's wrong at client or server end, but users won't understand that part) > We see this as a common complaint with users who connect via dialup modem. I did too - but I solved 90% or more of user modem problems by getting them to ensure fallback is enabled and to find the telephone on the line which is pulling the line down after a few minutes. This is a constant problem for dialup providers. Basically lots of phones out there draw line power to maintain their last-dialled-number memory or speed dial presets, instead of using batteries (or the batteries go flat and the phone uses line power to maintain things, so the users are completely unaware of it until there's an icky gooey mess in the battery compartment). The problem with line powered memory is that on long modem calls, the capacitors feeding the ram discharge to the point where the phone will attempt to pull power from the line (foward voltage conduction threshold on the rectifier diodes). At that point, the line impedence changes and starts varying wildly. That has the effect of scrambling the carefully negotiated echo cancelling settings the modems set up at the start of the call. _IF_ there's fallback enabled, the modems will renegotiate and continue at a lower speed. It will take several rounds before the line stabilises, so several models of modems often give up after a number of retrains. Even cheap modems will survive this kind of problem if fallbacl/fallforward is enabled. The _hard_ part about diagnising this and proving it is that as soon as the user's modem goes offline, the line voltage rises back to 48v and the faulty phone immediately charges up, giving several more minutes of calling before the whole thing repeats itself. The only way I've found to conclusively show this kind of shit to users is to go onsite with a modem which has online mearsurements of signal/noise ratios (I used a U-1496 for a long time for this. Bulky, but effective). It was easy to show users what a good line looked like and what happened when the line was turning to shit because of a faulty phone _and_ how it instantly got better when the bad phone was unplugged. Following that, it's a simple matter of showing things getting bad/good as the phone is plugged/unplugged. Most users toss the phones (it's invariably a $5 cheapie bedside phone they bought years ago). It's also fixable by putting a privacy adaptor between the bad phone and the line - these isolate the phone from the line when voltage drops to the nominal 6-12V looped level from 48V idle level (at simplest, a pair of back to back 12V zener diodes will work). AB
