The non-overlapping channels are 1, 6, and 11. The radios on channel8 and 11 will somewhat interefere with each other. I would move the Linksys access points from 8 to 6.
Look for microwave oven, 2.4GHz cordless phones, wireless security cameras, 2.4Ghz wireless mouse, Bluetooth phone base near the access points, microwave plastic pre-heater, 2.4Ghz sulphur lamp, Zigbee networks, municipal wireless, and other 2.4GHz wireless networks. There have also been reports of wireless doorbells, radio remote controls, and high power commerical VHF/UHF radios and paging xmitters causing problems.
Changed from what to what?
If the intereference is coming from a non-802.11 source, Netstumbler and the various passive sniffers will not detect anything. You might try using Kismet on Linux to do the sniffing as it will support most NDIS5 compatible cards. The easiest way is to use a "Live CD" such as:'s a huge download but worth it if you're planning to need 802.11 hacking and sniffing tools. However, my guess(tm) is that it's not 802.11 related. The consistant 10 minutes sure sounds like a production microwave oven. You'll need a spectrum analyzer and someone experienced to interpret the display to find the source.
What does "this would happen" mean? Did the laptop stay connected or not when the PDT6800 lost the connection?
Is the back wall of the warehouse tranparent to RF? Or is the warehouse made of steel and/or contrete? (Is it coming from inside or outside?)