Here its working fine now. Its all switches these days. I actually looked for a hub to see if I could replicate the issue and found a couple from the 90's.
My issue though as documented here began gradually and was very random for a couple of years back on my network of multiple unmanaged switches.
From what I can see the issue happened with only switches for HAI users. (probably unmanaged).
Most folks that I know do not put managed switches on their home network and typically use the ports on their combo firewall router AP plus maybe another small unmanaged switch. I don't really know how many pro installers or even if pro installers today install managed switches in their residential installations (maybe for the 10K home they do; but probably not the 5K or less home).
The "fix" of putting the HAI OPII on its own network behind a firewall has fixed my issues.
Relating to this issue specfically It appears to be present with my unmanaged switches. I did see a broadcast storm the other day.
Paying attention somewhat I have seen that my xAP network (UDP broadcasts) of many hardware to xAP pieces is very chatty. (almost too chatty). Aside from the HAI combo sensors everywhere I also have combo 1-wire sensors literally in every room of this home and outside in multiple locations. IE: when looking at temps / humidity on the second floor I am looking at some 5 1-wire combo sensors plus one in the attic. I have only two combo HAI sensors on the second floor.
I am streaming much video / audio these days but I do not think that is the issue. IE: separate NAS boxes for video/audio streaming storage, MythTV/Squeezebox server with 5 Tuners, HDHomerun with 3 tuners and PlayOnTV/Tversity box and little Aopen DE's running XBMC these days all over the place plus up to 20 touchscreens that connect to the HAI panel (mothership), Homeseer (mothership), ZM (mothership) plus now some 8 IP cameras ....funny thing is that I have satellite, OTA and cable TV and do not watch TV anyways (that is me though).
I ordered a new managed Gb switch anyways simply because I have not paid any attention (other than making my little subnet DNS / IP spreadsheet) to how large my network has become. I did get rid of my managed cisco switch a couple of years ago to move towards smaller with similiar port capacity unmanaged Gb switches. I have a "mini" sized 24 port Gb unmanaged switch inside of my Leviton 42" panel that appears to be working fine these days. On the "server" rack some 30 feet away I have another 24 port Gb unmanaged switch. In my home office I have 4 network runs to the basement going to the patch panel. That said my desk has some 4 touchscreens that I play with so added another 8 port Gb switch (to play with). I also added another small Gb switch just for the garage devices (some 8 today). There are a few more "pockets" of Gb runs to Gb mini switches in the house today because there is never enough cable run.
I am still in process of "building" my POE network and decided on using a midstream POE 24 port managed Tycon midstream injector (managed kind of switch). I am not there though and still using multiple smaller POE switches (now have multiple vendor, multiple port et al small POE switches). My goal here though is to have all of the IP cameras and touchscreens using POE switches. The newest "experimental" touchscreens (that run Linux, Android and XP) are using POE power splitters and are doing well with these but do draw some up to 1.2 amps or so of power.
So the issue and the solution here for this specfic issue of the network disconnects, time sync and comm issues; "little engine that could" are all interelated and due the legacy network interface on the HAI Panel with a solution of a managed switch.
That said I don't see the issue cropping up with a small home network on 1-2 unmanaged switches and some 10-20 devices on the network.
