7
\$\begingroup\$

I have several devices and some long cables on my PC's USB buses and every now and then I get a message like hub 6-0:1.0: port 2 disabled by hub (EMI?), re-enabling... on my Linux machine. Especially the devices on my desk (keyboard, mouse, 2 Arduino's) are being disconnected and reconnected. These devices respond to electric shutters for the windows, but also from passing motorbikes.

What can I do to improve immunity to EMI?

Tried ferrite cores on the mains cable of the shutters and some other equipment, but doesn't really improve much.

Bus 006 Device 025: ID 1a40:0101 Terminus Technology Inc. 4-Port HUB
Bus 006 Device 026: ID 0403:6001 Future Technology Devices International, Ltd FT232 USB-Serial (UART) IC
Bus 006 Device 027: ID 03f0:0324 Hewlett-Packard SK-2885 keyboard
Bus 006 Device 028: ID 046d:c046 Logitech, Inc. RX1000 Laser Mouse
  • Moved the keyboard off the hub, errors still appear.
  • Moved the mouse off the hub, errors still appear.
  • Moved the Arduino off the hub, errors still appear.

Eventually replaced the long extension cable with a shorter one and haven't had any errors for over a day. The longer cable is required to reach my desk though without cables dangling halfway in the air.

\$\endgroup\$
  • 7
    \$\begingroup\$ I'd guess that it's more likely vibration causing problems than EMI. If you've got a dry joint on one bus line, it only takes a couple of microseconds of disconnection for the USB controller to crash. I can't see how passing motorbikes would cause anywhere near enough EMI to knock out your devices. \$\endgroup\$ – Polynomial May 7 '12 at 21:28
  • \$\begingroup\$ What's connected to port 2? Is it always port 2? \$\endgroup\$ – Kevin Vermeer May 7 '12 at 21:29
  • \$\begingroup\$ The keyboard. But every now and then it locks up that whole hub. So keyboard, mouse and Arduino. \$\endgroup\$ – jippie May 7 '12 at 21:43
  • \$\begingroup\$ Does it happen with the keyboard disconnected? \$\endgroup\$ – Thorn May 7 '12 at 21:48
  • \$\begingroup\$ Moved my keyboard to another USB-port directly on my PC, but I cannot reliably reproduce the issue. So these tests take some time. \$\endgroup\$ – jippie May 7 '12 at 21:49
4
\$\begingroup\$

I took a rather thin USB extension cord, ran it 4 times through a ferrite core (one that comes in two halves in a piece of plastic and can be closed with a click) and ever since I haven't had any errors on my PC.

\$\endgroup\$
  • \$\begingroup\$ So, that was EMI? I don't think it is coming from motorbikes or such. Have you tried looping your oscilloscope's ground wire and connecting it to the probe's input, basically measuring the inductor that is created by the ground wire? It can tell something.. \$\endgroup\$ – abdullah kahraman Jun 15 '12 at 20:43
  • \$\begingroup\$ The logfiles in my PC claimed it was EMI. Only after running the cable through the ferrite core it disappeared. Haven't seen the messages for a month. \$\endgroup\$ – jippie Jun 15 '12 at 20:45
  • \$\begingroup\$ Yes, it probably is EMI since ferrite core solved the problem. To tell the truth, I know nothing about the subject. But I don't think they are coming from motorbikes passing by. I would create my little receiver with a scope probe and search the source of EMI, like searching a treasure underground :) \$\endgroup\$ – abdullah kahraman Jun 15 '12 at 20:48
  • \$\begingroup\$ I agree with @abdullahkahraman here, there is no way it is motorbikes. There is something wrong if it is sensitive to that. Much more investigation is needed to know what it really is. This is more of a superuser answer;. \$\endgroup\$ – Kortuk Jun 15 '12 at 21:39
  • \$\begingroup\$ @Kortuk how come it is a superuser answer? Does superuser SE answerers know anything about EMI? \$\endgroup\$ – abdullah kahraman Jun 16 '12 at 8:03
2
\$\begingroup\$

Shielded cables. But that isn't the problem if it is due to passing vehicles (as @Polynomial said in a comment above).

USB is differential. Radiated EMI would effect both lines essentially the same, so it won't have much of an effect. That suggests you have conducted EMI problems. However any decent power supply won't pass enough EMI through to be a problem.

\$\endgroup\$
  • \$\begingroup\$ The PC is behind an APC UPS, which claims to do a lot of filtering. \$\endgroup\$ – jippie May 7 '12 at 21:54
  • \$\begingroup\$ Only a UPS with a active filter. \$\endgroup\$ – Standard Sandun May 8 '12 at 0:56
1
\$\begingroup\$

Have been working a little with EMI issues and have done some studies.
A motor bike can absolutely create some serious EMI emissions, very broad band emission, nasty stuff. Especially if it's an old model or if it is home "tuned". If you want that super spark and you have a long cord to the spark plug, you have created a good loop antenna. And running 15000 volts from a capacitor trough that loop creates big spikes. No doubt.
Just listen to an AM radio when those guys are driving by the block gives you the answer.

\$\endgroup\$
1
\$\begingroup\$

I'm late to the party, but I've been pursuing this issue (port X disabled by hub (EMI?), re-enabling...) for quite some time. It's especially aggravating because this syslog message is often followed by the computer shutting down. I think I've tracked down the cause (for me).

This computer is my media server and I have it my my master bedroom closet, which is adjacent to the master bedroom bath. The bath has two independent exhaust fans. Today I determined that simply turning off either of these fans triggers the "port X disabled by hub (EMI?), re-enabling..." syslog message. Turning on either fan doesn't cause an issue.

I can't confirm that motorbikes cause the issue, but I can confirm that switching off a small inductive load (fan motor) generates enough EMI to trigger it.

I'm going to buy a snubber, or build my own, and see if that resolves the problem.

\$\endgroup\$
0
\$\begingroup\$

I have experienced similar issues with some USB-based diagnostics kit. The problem we have found is it is more often then not the hub within the computer that is the problem.

From what we can deduce, if we are connected to our motor-drives and we start switching, EVEN if the controller is gnd-referenced then there is disconnect issues and lockup at some point. If we use laptops and essentially float the laptop then there isn't any issues, likewise if we use an optical fibre USB link then there isn't an issue. Some form of gnd-loop between the PC -> USB device -> UUT causes the PC, using cheap, commodity parts to have problems.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.