AM335x USB Host mode VBUS question

bobby boucher

Prodigy 210 points

Other Parts Discussed in Thread: AM3352

We designed a board using a USB-Ethernet chip on board. We use USB0 in host mode on an AM3352.

Since the USB is internal only, we didn't see the need for a power switch connected to VBUS and DRVVBUS.

How should we handle VBUS? Can it be directly connected to a fixed 5V supply?

thanks

over 8 years ago

0 Bin Liu over 8 years ago

TI__Guru*** 125366 points

Bobby,

In your use case - USB in host-only mode, you can remove the power switch, and connect USB0_VBUS to 5V. But you will need a kernel patch to get USB working in this case. What SDK kernel do you use? Then I can point you to the right patch.

0 bobby boucher over 8 years ago in reply to Bin Liu

Prodigy 210 points

We are using Linux kernel 3.2, AM335x PSP 04.06.00.11

0 Bin Liu over 8 years ago in reply to bobby boucher

TI__Guru*** 125366 points

Bobby,

For kernel 3.2, please use the patch mentioned in e2e.ti.com/.../1163908 to arch/arm/mach-omap2/omap_phy_internal.c.

0 bobby boucher over 8 years ago in reply to Bin Liu

Prodigy 210 points

so, I have

USB0_ID grounded
USB0_CE unconnected
USB0_DRVVBUS unconnected
USB0_VBUS connected to +5V
USB0_D+ and USB0_D- connected to device

I applied the patch

But my USB device (usb to ethernet chip) is still not detected. Any idea?

root@1234:~ devmem 0x44E10620
0x3C106004
root@1234:~ devmem 0x44E10628
0x3C186004

cat /proc/driver/musb_hdrc.0
Status: MHDRC, Mode=Host (Power=e0, DevCtl=19)
OTG state: a_idle; inactive
Options: ?dma?, otg (peripheral+host), [eps=16]
Peripheral address: 00
Root port status: 00000100
CPPI: txcr=0 txsrc=0 txena=0; rxcr=0 rxsrc=280de80 rxena=0

0 bobby boucher over 8 years ago in reply to Bin Liu

Prodigy 210 points

0 Bin Liu over 8 years ago in reply to bobby boucher

TI__Guru*** 125366 points

Bobby,

The USB0 state looks good, but did you connect the USB0_VBUS line to the USB_Ethernet device VBUS pin too?

0 bobby boucher over 8 years ago in reply to Bin Liu

Prodigy 210 points

No, the usb-ethernet chip is powered using a 3.3V supply. Is that an issue?
The chip seems powered fine because a 100 Mbps link is negotiated successfully.

0 -DK- over 8 years ago in reply to bobby boucher

TI__Mastermind 27720 points

bobby boucher said:
No, the usb-ethernet chip is powered using a 3.3V supply. Is that an issue?
The chip seems powered fine because a 100 Mbps link is negotiated successfully.

Most embedded USB devices like this have a VBUS_PRESENT pin or the like. Does this one, and it is connected correctly? This pin tells the device when to activate or remove its pullup.

0 bobby boucher over 8 years ago in reply to -DK-

Prodigy 210 points

There is indeed a VBUS_DET pin on the LAN9500AI. It is connected to the same +3.3V.

0 Bin Liu over 8 years ago in reply to bobby boucher

TI__Guru*** 125366 points

The device VBUS_DET pin should be connected to USB0_VBUS pin. 3.3v is too low for it.

0 peaves over 8 years ago in reply to Bin Liu

TI__Guru 55325 points

Bin, your comment may not be true. Some devices expect an external voltage divider to reduce VBUS from 5 volts to 3.3 volts.

The expected input voltage for this function needs to be confirmed from the device's data sheet.

Regards,
Paul

0 bobby boucher over 8 years ago in reply to peaves

Prodigy 210 points

Yeah, according to the LAN9500Ai datasheet, while the input is 5V-tolerant, they recommend 3.3V on VBUS_DET, so I don't think this is the issue.

0 Bin Liu over 8 years ago in reply to bobby boucher

TI__Guru*** 125366 points

Bobby,

Then can you please probe the VBUS0_DP line with a scope? It should be high after the 3.3v is applied to VBUS_DET pin, during the system boots up.

0 bobby boucher over 8 years ago in reply to Bin Liu

Prodigy 210 points

Interesting, it is not high. It should remains high as long as negotiation with the host haven't started, right?

0 Bin Liu over 8 years ago in reply to bobby boucher

TI__Guru*** 125366 points

bobby boucher said:
Interesting, it is not high. It should remains high as long as negotiation with the host haven't started, right?

That's right.

0 -DK- over 8 years ago in reply to Bin Liu

TI__Mastermind 27720 points

Bin Liu said:

bobby boucher

Interesting, it is not high. It should remains high as long as negotiation with the host haven't started, right?

That's right.

Not necessarily. In the absence of VBUS (or VBUS_DET in this case), a USB device must disconnect its pullup within 10s.

0 -DK- over 8 years ago in reply to -DK-

TI__Mastermind 27720 points

I took a glance at this datasheet...there are a ton of device config options (TXCLK, PORT_SWAP, etc.), many of which would break USB if not done correctly. Might be worth going over them again.

0 bobby boucher over 8 years ago in reply to -DK-

Prodigy 210 points

VBUS_DET is connected to 3.3V, therefore I expected the chip to activate its pull-up on D+.

TXCLK and PORT_SWAP are both disconnected. I am still investigating.

0 Bin Liu over 8 years ago in reply to -DK-

TI__Guru*** 125366 points

Dave,

-DK- said:
Not necessarily. In the absence of VBUS (or VBUS_DET in this case), a USB device must disconnect its pullup within 10s.

The Device's VBUS_DET is tied to v3.3 power rail, it is always present, so D+ pullup should be there until the speed negotiation finished.

0 bobby boucher over 8 years ago in reply to Bin Liu

Prodigy 210 points

Thanks for your help. The patch was indeed needed. We also had another, unrelated problem regarding the EEPROM configuration but it is now working.

0 Bin Liu over 8 years ago in reply to bobby boucher

TI__Guru*** 125366 points

Bobby,

I am glad the issue is solved

Do you mind to explain what was causing the device did not drive D+ high?

0 bobby boucher over 8 years ago in reply to Bin Liu

Prodigy 210 points

The EEPROM used contained unitialized values (0xFF) except for the MAC address. Some of these values must have affected the USB communication.
After mounting the pull-up on the EEP_DISABLE pin, it started to work, since the LAN9500 no longer loaded its configuration from the EEPROM.

0 Colman Lai over 7 years ago in reply to Bin Liu

Intellectual 830 points

I have the same problem with the USB VBUS pin, but I am using kernel 4.1.18 installed from ti-processor-sdk-am335x-evm-02.00.21.11-Linux-x86-Install.bin. Do you have any patch for this?

Regards,

Colman

Processors

Processors forum

AM335x USB Host mode VBUS question