[mythtv-users] OT: USB inductance (or other) problems with external peripherals

[Stephen Worthington]

On Mon, 14 Oct 2013 09:30:15 +0200, you wrote:

>Hi,
>
>Totally off-topic this one. Sorry to be making noise on the list.
>
>   I'm currently working on a multimedia console for the end of my sitting room. the console contains everything for the front fo my entertainment system, from a TV-lifted TV, to tuners, decoders amplifiers, sub and centre channel speakers, Hauppage pvr and,. mythtv peripherals.
>
>There wasn't room for my combined FE/BE in the bench, su I ripped out the DVD-rom, the imon lcd, three esata connections and a usb hub and built them into the bench, running the cables (about 1,5M) through quite crowded conduits down to the cellar where the computer is hanging from the ceiling. The WAF improved as the wife hates ANY fan-noise. The power to the imon-lcd, power reset and power led are running through two sets of isdn cables with rj45 connectors at each end.
>
>I have two USB 2.0 buses. One is reserved for the hd-pvr running in a dedicated conduit, and the othe r is reserved for a powered hub. In addition I have a 1.1 bus serving a BT dongle and a 1.1 bus to the imon lcd and IR receiver.
>
>Imagine my disappointment when only about 10% of my imon keystrokes are getting through. The rest end up in dmesg as "unknown keystroke".
>
>I have tried changing the bus for the imon, but none of them give me reliable keystrokes. I have two main theories for the failure:
>
>1. Maxing out the usb connections with relatively long cables is stressing the USB controller and giving me garbage. I'd be better off restricting myself to a USB 2.0 bus to the HD-PVR and running everything else through other USB 2.0 bus and the powered hub.
>
>2. The problem is the crowded conduit. I'm getting inductance problems which are causing noise.
>I need to split up the cable runs with max one or two cables per condiut ( a pain in the *** as I'd need to rip the cellar ceiling apart (again).
>
>I' m way under the max cable length of 3-5M for USB, so whatever is causing it, my two guesses to the problem itself are interference and voltage drop.
>
>Does anyone know about the electromechamnics of what I'm doing and can they shed some light on where I should start changing things?
>
>Worst case is I'd have to abandon the whole approach and do it the proper way, spending the money and building a small footprint and SILENT dedicated frontend into the bench.
>
>Cheers
>
>Marius

Just a few thoughts of things that have bitten me in the past.

There is serious potential for ground loops - is the PC on a basement
power circuit?  And are the USB hub and the amplifier the audio cables
plug into plugged into the sitting room mains?  Combine that with the
earth paths through the aerials into the tuners, and you could be
picking up a lot of hum.  And there could even be serious voltage
differences between the basement and sitting room circuit's earths.
Put a meter between an earth coming up from the basement and the
sitting room earth to check for that.

Long cables are also aerials, and can pick up all sorts of strange
signals when not properly shielded - not just from nearby cables or
equipment, but any transmitter they are the right wavelength to
receive.  Powerful nearby radio stations are easy to receive on a long
cable of a matching wavelength and their signal can swamp other
signals.

My (tiny) experience with a long USB cable is that the voltage drop on
one is severe - you need to use cables with voltage regenerators to
get the device at the far end with enough voltage to work properly. Or
use a self-powered device at the far end, or a powered hub.  Of
course, there are hubs and hubs - some USB devices do not like to be
on any hub, and some hubs work much better than others.

The imon display presumably has a backlight, so is drawing a fair bit
of power.  So check the voltage it is receiving at the end of its
ISDN/RJ45 power cable while it is drawing that power - it might be
enough for the backlight, but not enough for its USB transmitter to
pump out a valid signal back through a long cable.

Problems like this are usually fixable if you know what the cause is -
so you may need to be using an oscilloscope to look at your signals to
diagnose the problem.