I like your sense of humor man, you got have some in dealing with this things. LOL.
Sometimes humour is all I have to keep me from going crazy.
You mentioned that you had older HW running A modes with out issue. One test I would do is try Single mode A or N if you can and see if it exhibits the same issues on 5Ghz. Try playing with channels.
One thing you could try, go get a different device, try a router, or access point something different. I might bet that device would exhibit the same problem. Just a suggestion.
It's a pain to find something that supports 5.2GHz. Bleagh. Found two, picked up one, trying it now.
Keep us posted on what you see on InSSIDer.
There are 13 other networks in my area according to inSSIDer in the 2.4GHz range, a plurality of networks with 40MHz blocks can be found on channel 1 (five of them), another cluster on channel 6 (four of these), one on channel 10; two 80MHz blocks exist on channels 6+10 and one on channels 1+5. I am running the only networks showing up in the 5.2GHz band at all.
I have tried a few configurations, using the units alone or in various groupings, and the results are the same. When used alone, they are placed in about the same place; when used together, I place them several feet apart but in the same area as when using them alone. Uplink is always autosensing 10/100/1000 switch, with them all using the expected speed to the switch (1000Mb/s for the 802.11n boxes, 100Mb/s for the 802.11a box), all over Cat5e cabling. I do not have any of the units set to drop wireless on loss of uplink. All units are powered by their discrete power converters (no Power over Ethernet) and all units plus the switch are plugged into an APC SmartUPS 1000XL.
I have tried max transmit power, -3dB, and -6dB. Signal at the clients seems to start to drop noticeably starting at -6dB, but none of these settings has any effect on the problems.
I have three client systems, all laptops, usually powered from line and often docked, connecting to these devices. I may have a Broadcom WiFi card around here somewhere but it is not mounted and the drivers are a pain (since it seems only Dell officially used that particular card, the drivers have to be tricked to install to anything but a certain model Dell laptop that I have never owned). All three clients are using Intel WiFi cards, which is what my limited testing has found to be most solid:
Lenovo ThinkPad W700, Intel 5300AGN, Windows 7 Ultimate, native WiFi controls
Toshiba Tecra M3, Intel 2915ABG, Windows XP Pro, native WiFi controls
IBM ThinkPad A31p, Intel 2915ABG, Windows XP Pro, Intel WiFi controls
Signal column is 'about average value' at client with the base station at full output.
Unit Mode Band Signal Free (<=20%) Loaded (>20%)
----------- ----- ------------- ------ ------------ -------------
DWL-7100AP a 5.2GHz -50dB no problems no problems
DWL-7100AP b,g 2.4GHz 0dB no problems no problems
DWL-7100AP a,b,g 2.4GHz,5.2GHz (above) no problems no problems
----------- ----- ------------- ------ ------------ -------------
TL-WR1043ND b 2.4GHz -40dB no problems no problems
TL-WR1043ND g 2.4GHz -40dB no problems no problems
TL-WR1043ND b,g 2.4GHz -40dB no problems no problems
TL-WR1043ND n 2.4GHz -40dB no problems no problems
TL-WR1043ND b,g,n 2.4GHz -40dB no problems no problems
----------- ----- ------------- ------ ------------ -------------
DAP-2590 a 5.2GHz -50dB problem 1 problem 2
DAP-2590 n 5.2GHz -50dB problem 1 problem 2
DAP-2590 a,n 5.2GHz -50dB problem 1 problem 2
DAP-2590 b,g 2.4GHz -35dB no problems no problems
DAP-2590 n 2.4GHz -35dB no problems no problems
DAP-2590 b,g,n 2.4GHz -35dB no problems no problems
----------- ----- ------------- ------ ------------ -------------
inSSIDer is confused by the DWL-7100AP in dual-band mode: it alternates it between showing up in 2.4GHz and 5.2GHz, but clients attached to it show no such confusion or any traffic interruption that would seem related. Maybe this has to do with it claiming the same MAC address in both bands (despite different SSIDs)?
Problem 1: Radio resets randomly, usually between 1 and 8 hours between resets, disconnecting all clients and requiring intervention (clients can not reconnect automatically and must be forced to dissociate before they can reconnect). When the radio resets, it pops back up on an apparently randomly chose channel.
Problem 2: Radio appears to keep working but control interface seems to become confused -- log disappears (page 1 of zero) and WiFi stats mostly stick at zero (maybe other symptoms I have not yet noticed). This seems to happen after several hours. The box must be restarted before the logs show up again (and any logs that did exist are lost).
On the DAP-2590, firmware 1.13 seems to show more credible stats for the wireless clients (particularly signal strength, which is often 80% to 90% but sometimes varies outside of this range for short intervals, and reduces if I reduce power at the client), but it seems flaky about connecting both 802.11n and 802.11a clients at the same time. While firmware 1.20 does not seem to have problems with having both 802.11n and 802.11a clients, it usually shows what I think are pretty bogus signal levels (almost invariably 100%, and does not seem to reduce until I set the client power very low). I do know the ThinkPad A31p antennae were designed for 2.4GHz radios, but the 5.2GHz card works fine with it at slight cost of signal strength on both client and access point end, so basically it should probably never show up as 100% at the AP (sits around 60% according to the DWL-7100AP).
The DAP-2590 signal strength according to inSSIDer is pretty stable in 5.2GHz mode, sitting around -50dB, but in 2.4GHz mode it hops about, from -85dB to -10dB or better, often swinging wildly between samples, though it seems to tend toward -35dB. Maybe this has to do with all the clutter in 2.4GHz that does not appear in 5.2GHz, but it's a touch unnerving and yet another reason I would prefer to stay in the 5.2GHz range. A number of the other networks in the 2.4GHz band have similar behaviour, but my DWL-7100AP seems pretty steady at around -50dB in the 5.2GHz band (but curiously shows up at 0dB in the 2.4GHz band).
Despite the 'improved' signal level, performance in 2.4GHz is awful. The DAP-2590 usually transfers data at between 20% and 60% of the claimed speed, and in the 2.4GHz band more often the 802.11n client claims speeds in the range of 11Mb/s to 40Mb/s (though sometimes as high as 130Mb/s, even though inSSIDer claims it is always using an 80MHz channel). In 5.2GHz it consistently provides nominal speeds of 200Mb/s and up, usually toward 300Mb/s. Perhaps due to the 2.4GHz clutter?
In the 2.4GHz band, with no traffic other than basic ambient fluff, pings between wireless and ethernet take somewhere between 4ms and 5000ms (maybe averaging about 500ms), and about half of them are dropped. More to do with 2.4GHz clutter? The TL-WR1043ND and DWL-7011AP do not exhibit this at 2.4GHz (or the DWL-7100AP at 5.2GHz), though; both of them show a tighter time range (3ms to 100ms, averaging about 15ms) and drop less than 1%. The DAP-2590 provides better (but still not the best of the three) results at 5.2GHz.
Interestingly, I have not seen it reboot the radio in 2.4GHz... I wonder if they test it at 5.2GHz in the lab...?
So after all this, the tech support claim that it is resetting in 5.2GHz mode because of 2.4GHz interference, despite supposedly only operating on a single band, is sounding more bogus. I have had 802.11a (5.2GHz) connections to the DWL-7100AP stay up for weeks. Seems to be little point in mentioning I have had 802.11n (2.4GHz) connections to the TL-WR1043ND also stay up for weeks.
But despite the DAP-2590 apparently working in 2.4GHz, it is not working where I wanted it to work, as I specifically purchased it to have 802.1n in the 5.2GHz band.
I need to return the DAP-2590 by early December.
If I can find something that does
not (apparently randomly) drop all connections when running at 5.2GHz (and otherwise works well and does what I want), I will probably keep it. Otherwise I'll skip 802.11n (go back to old faithful 802.11a) or wait for something that does work the way I want. As for the DAP-2590, I'm not willing to pay commercial grade prices for beta grade product, so will probably send it back in a few days, with minimal additional tinkering, if nothing particularly inspiring comes along.