USB OTG port does not appear to be working.

[DAP]

I bought a mini-A to full sized A adaptor so I could use my USB OTG port.

First, I'm really annoyed with Amazon since what was delivered was a mini-B to full sized A adaptor. The picture clearly showed a mini-A connector. The text did not mention what type of connector A or B, just that it was mini. But the text also claimed that it was a 90 degree connector which it was not (and the picture did not show any 90 degree connectors).

In any case, if the mini B connector is wired correctly, it should work fine.

I verified that the sense line is tied to ground like it should be.

I plugged in a keyboard to the adaptor, and plugged the adaptor into the Pandoras USB OTG port... Nothing. No power to the device, no response from keypresses, does not show up in 'lsusb'.

Incidentally, with nothing plugged in, lsusb only shows one device. Shouldn't it report two? One for the full sized USB root hub, and one for the USB OTG root hub?

I also bought a two port high speed USB unpowered hub.

I plugged the hub into the full sized USB port on the Pandora, it shows up on 'lsusb'

I plugged in the keyboard to the hub, it works.

I did a lsusb -v and verified that the keyboard only requires 100 mA.

Why would my USB OTG port not work?

 

[Neelix]

There was another thread about this recently.  I'd suggest following the sequence in this post, to see if that gets it working for you.

- Neelix

 

[Linux-SWAT]

Also, not all adapter works, be sure to get a nokia one.

Check this:

http://pandorawiki.org/USB_compatibility_list#OTG_.28On_The_Go.29_Adapters

 

[Galaxis]

Usually, the magical trick to making the OTG port work is to enable USB Networking (which happens to load a few of the required drivers) in addition to the USB host. I just have that on autostart. 

 

[hmc]

My experience is that as soon as the necessary drivers are loaded and the USB Host is enabled, almost all adapters work. I have a bunch of different adapters here, including some really cheap ones from China, and they all work well.

You need to enable USB Host and USB Networking, though. Otherwise it won't work.

 

[ingoreis]

Dont forget to Upgrade your System over Wlan.

There was a Bug in USB OTG 1-2 Weeks ago and its solved.

Upgrading your Pandora OS should help and dont forget to activate "USB Networking" in the Boot Services.

But Renember that the OTG Port deliver for many Devices not enough energy.

Open a Terminal,

put in your OTG Adapter in with the Machine you want to use,

type in the Terminal "Dmesg"

and see the last Messages.

With my Wireless Microsoft Mouse work it perfectly.

 

[DAP]

Dont forget to Upgrade your System over Wlan.

There was a Bug in USB OTG 1-2 Weeks ago and its solved.

Upgrading your Pandora OS should help and dont forget to activate "USB Networking" in the Boot Services.

But Renember that the OTG Port deliver for many Devices not enough energy.

Open a Terminal,

put in your OTG Adapter in with the Machine you want to use,

type in the Terminal "Dmesg"

and see the last Messages.

With my Wireless Microsoft Mouse work it perfectly.

I updated. This was not sufficient.

I enabled USB networking. (rebooted) Now the keyboard works.

Certainly not plug & play, but functional.

Thank you.

 

[ingoreis]

Nice to read that it worked.


I am glad when i can help.

 

[notaz]

I've worked on kernel OTG driver over the weekend, could anyone help testing this with various OTG-to-host adapters?

 

[moxie]

I've only got one, but I can test with that? As long as you tell me how you want stuff tested, I'm on.

 

[Linux-SWAT]

I have the nokia and also a cheap one which never worked.

 

[EvilDragon]

I also have a few ones

 

[comradekingu]

I have 10 (!) that just came in, from ebay. Bulking up on OTG. Happy to test for you.

 

[Neelix]

I also now have an adapter cable, and a variety of low powered USB devices to test with.

Incidentally I recently started using USB networking and I've noticed a tendency for the gadget drivers (not only g_crc, but also g_audio)  to sometimes crash the entire unit when the attached host device is disconnected.  (or on being unloaded in the case of g_audio)   While you are looking at the gadget drivers is there any chance you could take a look at that?

- Neelix

 

[ibisum]

I have tons of stuff to test with.

 

[ingoreis]

I have tons of stuff to test with.

Dont forget thet the USB OTG deliver not so much power as the Big USB Host port.

Can we maybe mod it or write a Driver for more Energy Output?(only a Idea)

The Energy reaches here for using a Microsoft Wireless mobile Mouse 4000,

and i tested it with my DVB-T Stick Hauppauge WinTV Aero.

Mine USB Sticks did not work here,because to low power.

 

[Binky]

Are the +5V lines the same between the two ports? ie.e could they be wired together?

I'm guessing no.

 

[Loonie]

One thing I've noticed is I don't need to use the rmmod/modprobe g_cdc workaround anymore, the Logitech Nano is always recognized and functional when I connect it. I'm not sure if anything you've done has affected this, but it is undeniably way more reliable. Disconnecting, reconnecting, standby, most of the things that messed up the port before no longer do so. Only two things confuse it, and only one of them to any detriment

If the unit is in standby when the mouse is connected, it brings the unit out of standby, but the mouse doesn't work. However, disconnecting and reconnecting it brings it back.

The only thing that screws it up is if I disconnect the mouse while the Pandora is on standby, in which case it powers back up, but will not detect the mouse again until I do the g_cdc command.

 

[urjaman]

Are the +5V lines the same between the two ports? ie.e could they be wired together?

I'm guessing no.

RTFM (or maybe the Hacking Manuall) . But you guess right, the OTG Host 5V supply is actually a totally different "4.8V nominal" 100mA supply.( Things that I'd expect could work on the OTG port are single low power devices, like a mouse or a keyboard... The main benefit is working USB 1.1 without a hub.)

 

[notaz]

Ok here is the kernel packaged into ipk (1st step for planned kernel updates through "update pandora OS" as that can't update 3.2 kernel currently):

http://notaz.gp2x.de/tmp/omap3-pandora-kernel2_3.2-pandora-r1.5_omap3-pandora.ipk

You need:

- to back up /boot/uImage-3 before starting to revert to older kernel later if needed.

- to know how to install IPKs

- have SuperZaxxon Final or later installed (recommend fully updated SuperZaxxon 1.53 with "update pandora OS" applied).

- at least ~20MB of NAND space

- be ready to reflash if things go wrong (or do manual kernel recovery)

What needs to be tested:

- general functionality (works or not)

- role switching device->host, host->device (use as USB device, then unplug and connect some keyboard through host adapter or something and vice versa)

- suspend

Notes

- you should give it at least a second between (un)plugging devices so that interrupts, state machines and things finish their work. I don't want to go through all the race conditions of the driver and try to fix them, that would take months of work I think 1 second is reasonable compromise here.

- if you are running custom OS: it still can't suspend with g_* driver + USB cable active, the device misbehaves so the driver won't even allow it now, so rmmod g_* modules before suspend (default SuperZaxxon should do that for you).