Ok let’s go to the source. In pure VoIP or ISDN environment ecko is possible only as a result of the poor design of UA (endpoint, terminal, soft/hardphone) with a feedback from a speaker to a mic. In this case YOU are causing the problem. Other story when your voice is getting converted to 2-wire line inside PSTN. It uses special transformers which separate analog signals going in oposite directions. This is the major source of the problem. Besides that if you have a problem with one specific extension of one specific PBX should it indicate the problem with that extension. Could you give me a link to the article which claims that ecko doesn’t exist in analog PBXs? I’d be interesting in contacting the author and discuss some aspects of the theory of telecommunications.
I totaly understand that it’s much easyer to make you feel guilty becuse you use “noname” brand PBX (Asterisk in eyes of management) and they installed and use for years some kind of Avaya and just don’t wont to think that it could have a problem too.
Back in days I was working on experimental implementation of drivers for one of ISDN cards and had exactly the same problem, it was extremely sensentive to ecko from the PSTN and was working just fine in pure ISND environment. The only fix was to implement EC and autogain algorithms in the driver.