USB 3G modem (Novatel MC547) resetting constantly: port x disabled by hub (EMI?)

I'm running OpenWrt (kernel 2.6.37) on an embedded router (Ralink RT3052 SoC) with an USB OTG host (using the DWC OTG driver). I have numerous 3G USB modems (and USB flash drives) that all work perfectly with this board. However, I have one new modem, the Novatel Ovation MC547 (42 Mbps), that is unusable because it keeps disconnecting/resetting as soon as inserted:

usb 1-1: new high speed USB device using dwc_otg and address 2
scsi0 : usb-storage 1-1:1.0
hub 1-0:1.0: port 1 disabled by hub (EMI?), re-enabling...
usb 1-1: USB disconnect, address 2
usb 1-1: new high speed USB device using dwc_otg and address 3
scsi1 : usb-storage 1-1:1.0
hub 1-0:1.0: port 1 disabled by hub (EMI?), re-enabling...
usb 1-1: USB disconnect, address 3
usb 1-1: new high speed USB device using dwc_otg and address 4
scsi2 : usb-storage 1-1:1.0
hub 1-0:1.0: port 1 disabled by hub (EMI?), re-enabling...
usb 1-1: USB disconnect, address 4
usb 1-1: new high speed USB device using dwc_otg and address 5
scsi3 : usb-storage 1-1:1.0
hub 1-0:1.0: port 1 disabled by hub (EMI?), re-enabling...
usb 1-1: USB disconnect, address 5
usb 1-1: new high speed USB device using dwc_otg and address 6
scsi4 : usb-storage 1-1:1.0
hub 1-0:1.0: port 1 disabled by hub (EMI?), re-enabling...
usb 1-1: USB disconnect, address 6

This modem is supported by the Option serial driver in recent kernels. It works fine in my Linux PC (Fedora 13) and my Windows 7 laptop. It also works in my Fonera 2.0n router (another Ralink RT3052 board by a different manufacturer).

I have tried everything I can think of to solve this problem, including updating the drivers and experimenting with drivers/usb/core/quirks.c.

Interestingly enough, it doesn't reset when connected to the problem router via an external, non-powered USB hub. The modem switches (via usb_modeswitch) and the serial ports are exposed... but they don't work

I'm all out of ideas here. I'm concerned it may be a hardware problem with the router's USB host... but it works fine with all other devices I can throw at it.

Any help and/or suggestions are greatly appreciated.

Regards,
Layne

That's a pain in the <expletive deleted>, to quote a certain American President.

Usual thing is overcurrent. If you have a 4:1 hub outside the box, for example, the curerent the m/b port sees is all the other currents added together. People festoon these things with leds (20mA each) to make things worse. So the port sees too much current, and switches off.

Give that modem it's own port; if it acts up, try a different port on your m/b.

I am unable to get this device to work with usb_modeswitch.

I am using Ubuntu 10.10 and the device is being recognized, but judging from the messages, it is not switching properly:

Aug 15 11:08:34 gandalf kernel: [1547128.971274] usb 2-4: new high speed USB device using ehci_hcd and address 2
Aug 15 11:08:34 gandalf kernel: [1547129.190024] Initializing USB Mass Storage driver...
Aug 15 11:08:34 gandalf kernel: [1547129.190349] scsi4 : usb-storage 2-4:1.0
Aug 15 11:08:34 gandalf kernel: [1547129.190431] usbcore: registered new interface driver usb-storage
Aug 15 11:08:34 gandalf kernel: [1547129.190433] USB Mass Storage support registered.
Aug 15 11:08:35 gandalf kernel: [1547130.191122] scsi 4:0:0:0: CD-ROM Novatel Mass Storage 2.31 PQ: 0 ANSI: 2
Aug 15 11:08:35 gandalf kernel: [1547130.193107] sr1: scsi-1 drive
Aug 15 11:08:35 gandalf kernel: [1547130.194416] sr 4:0:0:0: Attached scsi generic sg2 type 5
Aug 15 20:53:19 gandalf kernel: [1582213.675817] usb 2-4: USB disconnect, address 2
Aug 15 20:53:20 gandalf kernel: [1582215.140024] usb 2-4: new high speed USB device using ehci_hcd and address 3
Aug 15 20:53:21 gandalf kernel: [1582215.448688] usbcore: registered new interface driver usbserial
Aug 15 20:53:21 gandalf kernel: [1582215.448700] USB Serial support registered for generic
Aug 15 20:53:21 gandalf kernel: [1582215.448733] usbcore: registered new interface driver usbserial_generic
Aug 15 20:53:21 gandalf kernel: [1582215.448735] usbserial: USB Serial Driver core
Aug 15 20:53:21 gandalf kernel: [1582215.484328] USB Serial support registered for GSM modem (1-port)
Aug 15 20:53:21 gandalf kernel: [1582215.484621] usbcore: registered new interface driver option
Aug 15 20:53:21 gandalf kernel: [1582215.484624] option: v0.7.2:USB Driver for GSM modems
Aug 16 07:53:00 gandalf rsyslogd: [origin software="rsyslogd" swVersion="4.2.0" x-pid="866" x-info=""] rsyslogd was HUPed, type 'lightweight'.
Aug 16 07:53:00 gandalf rsyslogd: [origin software="rsyslogd" swVersion="4.2.0" x-pid="866" x-info=""] rsyslogd was HUPed, type 'lightweight'.
Aug 17 08:03:03 gandalf rsyslogd: [origin software="rsyslogd" swVersion="4.2.0" x-pid="866" x-info=""] rsyslogd was HUPed, type 'lightweight'.
Aug 17 13:59:44 gandalf kernel: [1730198.881400] usb 2-4: USB disconnect, address 3
Aug 17 13:59:50 gandalf kernel: [1730205.010028] usb 2-4: new high speed USB device using ehci_hcd and address 4
Aug 17 13:59:50 gandalf kernel: [1730205.163717] scsi5 : usb-storage 2-4:1.0
Aug 17 13:59:51 gandalf kernel: [1730206.161235] scsi 5:0:0:0: CD-ROM Novatel Mass Storage 2.31 PQ: 0 ANSI: 2
Aug 17 13:59:51 gandalf kernel: [1730206.163220] sr1: scsi-1 drive
Aug 17 13:59:51 gandalf kernel: [1730206.164530] sr 5:0:0:0: Attached scsi generic sg2 type 5
Aug 17 13:59:56 gandalf kernel: [1730210.763975] usb 2-4: USB disconnect, address 4
Aug 17 14:00:20 gandalf kernel: [1730234.561278] usb 2-4: new high speed USB device using ehci_hcd and address 5
Aug 17 14:00:20 gandalf kernel: [1730234.714755] scsi6 : usb-storage 2-4:1.0
Aug 17 14:00:21 gandalf kernel: [1730235.711156] scsi 6:0:0:0: CD-ROM Novatel Mass Storage 2.31 PQ: 0 ANSI: 2
Aug 17 14:00:21 gandalf kernel: [1730235.713392] sr1: scsi-1 drive
Aug 17 14:00:21 gandalf kernel: [1730235.714697] sr 6:0:0:0: Attached scsi generic sg2 type 5

Can you tell me what usb_modeswitch file that you are using in /etc/usb_modeswitch.d?

I am hoping that I don't run into the problem that you had, but it seems that you are getting a bit further than I did. Any help would be appreciated.

Thanks,
Bob

I created a custom usb_modeswitch config from a similar device of same make (just changed the filename). However, I think it may already be in the latest usb_modeswitch data by now.

As for my the reset issue with the MC547, it was a hardware problem... not handling the inrush current. Upgrading the USB host capacitor fixed it.

Regards,
Layne

Probably too much smoothing inside the mc547 for the port specs. Glad you're sorted.

Thanks for the quick reply. I found that all that was required was to manually eject the cdrom that was being mounted. Once I do that, the modeswitch appears to happen as it should.

So the modeswitch does work, but it requires manual ejection of the cdrom that is mounted.