I'm having an issue with our USB0 on our custom board. It specifically seems to be related to the D- line and possible contention on the bus (but i'm not sure). We are running linux and using the RNDIS gadget. When the OMAP device is connected to a PC through a hub, the PC recognizes the device. When it is directly connected to the PC, it fails to recognize it and we get "Unknown Device" in the Windows 7 Device Manager.
The culprit seems to be when the OMAP device is connected to a PC, the USB bus get's a valid J state (D+ is high, D- is low). Then when it transitions to the K state the D- Line doesn't reach a valid state. It looks like a bunch of noise on the D- line but it is not pulled up high enough by the OMAP. This eventually leads to the device not being recognized. Running the bus through a hub makes things work. The hub doesn't seem to care about the D- line issue on the upstream side, and shows the approrpiate level downstream towards the PC.
I have a breakout board I use to measure these signal, and when I disconnect the D- line (so it doesn't go to the PC), it goes to the proper level. This tells me the OMAP is trying to drive it to the proper state, but it's running into contention some how.
Here's pictures of what I'm seeing...This is a capture of the data lines where the issue is:
The yellow trace is D+, the Green Trace is D-
Here is a capture of a different board design which this board is based on, where it works:
The yellow trace is D+, the Green Trace is D-
This is the fixture I'm using to probe the data lines. Note the resistors are not in the circuit. The jumpers are set for a straight pass through to the probes (in the SMA Connectors). By pulling the jumper on the green (D-) line I get a plot that looks similar to the second one labeled "Data lines look clean."
