It actually looks like a limitation in OSX and the type of drivers that are used.  A quick Google shows that there are others having USB issues with all items in general once they get past 8 or 9.  Doesnt look like OSX supports that many devices connected at once.

You dont need all those arguments, they are pointless and probably confusing bfgminer.  Just use:

c:\tools\bfgminer3.1.4\bfgminer.exe -o pool -u username -p password -G -S all

That will not use your GPU and will scan all of your ports for USB devices and use them.  If  you want to add the argument for the log file still you can.

What is your command line/config file? Have you tried a powered hub?

The only other suggestion I would say is try dialing down your intensity and/or your thread concurrency.  Maybe the card is being pushed too far.  Just because one person is able to get those settings working with their 6990, doesnt mean you will be able to.  All about the luck of the draw with the quality of the chips.

Also, when mining ltc, you need to be using scrypt and you are most worried about thread concurrency, intensity, and gpu memclock.  For btc mining, no scrypt, and you are worried about gpu engine and intensity.  So ideally you will need two different lines depending on which you are mining.  For example:

Radeon 5870, assuming it is the second video card in my case.

For LTC: --scrypt -o <pool> -u username -p pass -d 1 --gpu-memclock 1200 -w 256 -I 15 --thread-concurrency 7164 --auto-gpu --auto-fan --temp-target 75 --temp-overheat 85

For BTC -o <pool> -u username -p pass --gpu-engine 800-950 --gpu-memclock 300 -I 12 --auto-fan --auto-gpu --temp-target 75 --temp-overheat 85

You actually are overclocking in those settings, just not overclocking the GPU.  You are scrypt mining which uses more VRAM than it does GPU power.  Unlike your GPU, your VRAM on your cards do not have a active cooling system, so you have to make sure you have good case airflow to keep them cool.  A nice case that you can stick a couple of 120mm fans on the side to blow across the video cards would be ideal.

I also noticed you are not using any memory clock arguments. You might want to tweak those as well.  I dont know what the optimum settings for your cards, but you can easily look them up.  Then just add the argument to your command line, i.e. --gpu-memclock 1200

Lastly, you appear to be using two different command lines to connect to the same pool, using the same settings.  You can use just one command line for this.  Just take one of those lines, and remove the -d 0 or the -d 1 in one of the lines.  These arguments tell cgminer to only use one of the cards (either the one is slot 0 or slot 1), but if you are using them both at the same time, doing the same thing, then you can have it run both.  Since they are the same cards, the worksize settings, memclock settings, and everything else do not need to be changed and cgminer will apply those settings to all cards.

What temps are being reported in cgminer?  I would take my temp readings from there and not Afterburner.  In fact, uninstall Afterburner and dont even use it, cgminer will do all the tweaking you will need right from it.  Just learn a little command line or how to format a config file and you are good to go.

Can you post your config file or command line?

You can, but you have to run your QT client for whatever coin you are mining on a Windows PC, then point your minepeon to connect to it.

You will have to run your QT client on a Windows PC, and then in your config file for that client you tell it to allow rpc connections from another PC (other than your localhost).  Then you point your minepeon to that Windows machine.

I have solo mined with a QT client running on my network, and then my BAMT rig pointed towards that Windows machine.

He appears to be solo mining so there really isnt "accepted" shares.  When he finds a block, his accepted shares are of that of how many coins were in that block.

I think Catalyst will tell you, but not sure (dont use it).

Here is the AMD page.

http://developer.amd.com/tools-and-sdks/heterogeneous-computing/amd-accelerated-parallel-processing-app-sdk/downloads/download-archive/

Plugging them directly into the laptop?  If so, probably the reason they arent running.  Laptop ports are notoriously under powered.

What about your SDK?  You have to have an SDK 2.6 is needed (I believe, could be wrong, could be older) for scrypt mining.

If you are using the latest version of bfg, I dont think you need "--icarus-options 115200:1:1 --icarus-timing 3.0=100", you can delete that.

What version of bfgminer are you using?

What about cooling?  These things get hot and you need a good air moving fan to blow across them.  Get you a USB fan, or, just take a box fan and point it at them. It could be the reason why they begin to run, then crash (especially the ones in the hubs that are lined up next to each other, it creates a heat dome).

Are you running this off the USB port from the motherboard? Or are you using a powered hub?

I was kind of thinking the same thing. What extra benefit are you trying to get?

Is that your full command line (because I dont see anything in there about a declared thread concurrancy)?  What GPU(s), OS, driver/SDK are you using?

Well I updated and seems to be running okay for now.  I do have the folder on the desktop.  I will have it moved and see if the behavior continues.

Thanks for the help.