<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<br>
<div class="moz-cite-prefix">On 10/15/13 5:26 AM, warpme wrote:<br>
</div>
<blockquote cite="mid:4dda65dd.50d27b38.525d0a59.7d944@o2.pl"
type="cite">
<p>Marius,<br>
<br>
I think we can consider following hypothesis:<br>
<br>
1.USB transfer errors due RJ45 impedance mismatch compared to
USB specs.<br>
2.Non sufficient power supply due RJ45 cable resistance.<br>
3.USB transfer errors due RJ45 cable cross-talks (with other
cables).<br>
4.USB data signals levels biasing due ground loops.<br>
<br>
Ad1: USB cables are called 'USB Cables' not only because they
have USB connectors but also because they have correct wave
impedance. <br>
Impedance depends on frequency and there is high impedance
difference because of ISDH freq ( order of 200kHz) and USB1.1
(multiple MHz). <br>
So RJ45 will have wrong impedance for USB because of:
1\different spectrum of frequencies and also because 2\it is
electrically designed for different impedance (ISDN requires
120ohm) while USBequires 90ohm.<br>
<br>
Ad2: USB requires given resistance on +Vusb and GND not higher
than. <br>
RJ45 cable never was designed to carry high currencies required
to power-up low-voltage devices like USB.<br>
<br>
Ad3: USB cable has practically always D+ and D- as twisted pair.
Not only for achieving given impedance - but also for reducing
sensitivity for external electro-magnetic fields which can cause
induction of distortion signals in data path.<br>
<br>
Ad4: this is always problem when You have multiple power
sources...<br>
<br>
Order of hypothesis I presented is IMHO from most probable to
least probable.<br>
For verification of them I propose however to go with following
order of tests:<br>
<br>
Start with p2: use voltmeter on end USB device and check
voltage. <br>
If voltage is OK (not lower than 4.75V) - make sure there is no
ripples by spikes of load. <br>
Ideally will be to use oscilloscope. In You don't have one -
simple add 470-2200uF electrolytic capacitor to filter ripples.<br>
<br>
When 2 not helps: test p4. Simply power-up ALL HW from single AC
source. You can use temporarily AC power extender cable i.e..<br>
<br>
When this not help time is to eliminate p1+p3: try use
dedicated, good quality of USB cable. <br>
I would look in comp.store for special USB reach extending
cable. You may choose at least USB2 supporting cable as they
have to meet higher impedance requirements and consider to
choose famous vendor (HAMA, etc)<br>
<br>
PS: I personally will start immediately with good USB cable as
I'm pretty sure Your problem is with wave impedance mismatch....<br>
<br>
Good luck! </p>
<br>
<fieldset class="mimeAttachmentHeader"></fieldset>
<br>
<pre wrap="">_______________________________________________
</pre>
</blockquote>
It isn't always impedance causing a problem, especially with USB.
USB uses a lower frequency than gigabit ethernet. <br>
But, one can suffer from ringing on long runs, secondary to
inductance in the line.<br>
That is why many long length USB cables come with built in Schmitt
triggers as buffers.<br>
</body>
</html>