After testing different WIFI modes B/G/N I have found the esp devices work best in:
B : Full 20mhz
I have limited the esp devices to B mode by simply making my router B mode only.
I don't think there is a WX module command in the parallax esp firmware... probably a good suggestion.
Then users don't need to limit their whole wifi to B.
I use B because it is much better in staying connected to the esp, going around obstacles that interfere,
no longer interfere, the telnet putty connection and webpages don't stop responding or drop.
Rarely do users need their esp's connected to the internet,
so i suggest getting a local wifi device for your esp and setting it to B only.
Also seriously consider how fast a esp can really communicate at, and can it really even reach B bandwidth?
I should do a test between B and N and see if the power use is any different.
Heres what the esp data sheet says about the wifi in the different modes.
802.11 b: +20 dBm
802.11 g: +17 dBm
802.11 n: +14 dBm
802.11 b: -91 dbm (11 Mbps)
802.11 g: -75 dbm (54 Mbps)
802.11 n: -72 dbm (MCS7)
Im pretty sure these things aren't like your regular B/G/N device, they have their limits.
I personally find this to be true, b actually works better.
To make this REALLY useful it needs to be selectable in the firmware.
I have 3 of the parallax esp devices inside my railroad engine controller, and they all respond a magnitude better in B mode.
I use the esp devices to make 4+ network connects at once, telnet, webpage(2), programming.
Many do not make this many connections repeatedly, or I haven't seen any project that does what my railroad engine controller does.
So its a pretty good test of these esp devices, I even have intermittent power due to dirty track,
so I also have a bad power condition to test under and B wins hands down.
My obstacle is aluminum bars that blocks the wifi around the back corner, N will drop every time the engine corners, but in B mode it never drops the telnet connection.
It will sometimes drop in b mode on the corner if I am doing telnet, and webpage, but not like N did.
I also found that my specific code with the esp's doesn't like more than 1 connection per 2 seconds.
So I keep my webpage java refresh of data from the esp to 2 seconds.
And If I make multiple connections I will stagger their connection attempts by 1 second so they alternately access the esp.
Its important to keep your webpage to esp connections in series, its VERY easy to overload the esp with requests.
RE: new/ modules dropping connections, modules not connecting, modules loosing their program/ota.
I would like to mention to everyone who uses a esp module which is soldered smt to a carrier board, like parallx's WX esp8266 module.
The one parallax uses is soldered to a carrier board.
When this is solder pasted, and then "cooked" most of the solder looks ok, but it is not always.
I have ordered over 10 of the sip or dip modules from parallax, and have noticed one major issue that has appeared a few times now.
Simply put the ground for the module isn't bonded well or possibly cold joint, when I get a new module and it has issues staying connected, issues taking programs,...
I fire up a fine tipped solder gun and some fine solder, and give each smt pin a touchup. You can tell when you hit the ground it takes a long while to heat up, even with a nice precision gun, and you will see why its probably the issue.
It might be wise to get someone to touch up the ground smt pad on those production modules!