USB mode does not return to b_idle

Peter,

Please run the follow diagnostic script on your board and post its output log.

4604.chkusb.sh.txt

chkusb.sh Version 0.2.2
Linux am335x-evm 3.14.26 #9 Tue Jan 5 11:34:32 CST 2016 armv7l GNU/Linux
USB is initialized
usb@47401000: otg, okay
usb@47401800: host, okay

Gadget Kernel Config: g_serial is enabled
gadget driver loaded: DRIVER=g_serial

The list of USB gadget drivers installed:
/lib/modules/3.14.26/kernel/drivers/usb/gadget/:
g_serial.ko
libcomposite.ko
u_serial.ko
usb_f_acm.ko
usb_f_obex.ko
usb_f_serial.ko

Grounding VBUS (T18) caused the mode to return to b_idle; it was still "b_peripheral" after sitting disconnected for about an hour.

When I have the debug filesystem loaded I can issue:

  cat /sys/kernel/debug/musb-hdrc.0.auto/regdump | grep DevCtl

and get 99 when it is disconnected and 91 immediately after disconnecting the cable.

Peter Steinberg said:

Grounding VBUS (T18) caused the mode to return to b_idle; it was still "b_peripheral" after sitting disconnected for about an hour.

This is indeed a VBUS discharge problem. I have forwarded this thread to our hw expert, he will comment here.

Peter

Using a scope, can you measure how long it takes for VBUS to discharge after the cable is unplugged?

I notice in your schematic, that there is a GND and GND_CHASSIS, are they tied together somewhere? 

Can you remove the 100nF capacitor on VBUS?

Thanks

David

Thanks,
after handing it over to our hardware person it looks like a design problem.

The hardware mod seems to resolve the issue.

Hello David,

We have the similar problem with b_idle state. We use a LP3869x LDO for supply USB devices.

We see a 2.5V level at the USB0_VBUS after disconnecting the USB cable and

cat /proc/driver/musb_hdrc.0 | grep "OTG state"
reports that device in "b_peripheral, inactive state".
We tried to remove all capacitors from USB0_VBUS line, but VBUS is 2.5V anyway and everything works only after grounding USB0_VBUS.

Ivan

I see there are two GNDs in your design, which GND do you use to ground the VBUS? How are the two grounds connected to each other?

Thanks
David

Hi David,

There are two different symbols for only one GND. Apparently, these parts of the schema are inherited from older projects.

I removed a LDO, but there is 2.5V on the VBUS anyway.

This is the full path of USB lines from am335x to USB female connector:

Thanks

Ivan

When you removed the LDO, are you referring to the DA3 LDO? Does 2.5V eventually goes to 0V?

Are you in host mode or peripheral mode? If you are in host mode, do you drive DRVBUS signal low when USB cable is removed?

Please double check your layout. In the tool I used, if the GND symbol is different, it will be treated as two separate ground plane, not a single ground plane.

Thanks

David 

Hi David,

Yes, I removed the DA3 LDO, but USB0_VBUS was 2.5V anyway.

The device stays at the "peripheral inactive" state, does not return to b_idle. DRVVBUS is low.

I connected this pads with a wire on the board but it is not helped me.

Thanks

Ivan

Can we try a simple experiment, if you connect the GND of C133 to the GND of DA3 with a wire, does VBUS go low when USB is not connected?

Thanks

David 

I think that it is not necessary =)

Ivan

Do you ever see 2.5V going down to 0V or does it stay at 2.5V?

Can you please check the SESSION bit in the DevCtl register? Does it set to low when USB cable is removed?

Thanks

David