[vdr] Re: LAN cable length / Streaming Plug-In

[Maarten van den Berg <maarten@ultratux.net>]

On Friday 04 April 2003 15:19, Jörg Knitter wrote:
> Maarten van den Berg wrote
>
> > The trick with cat5 is not the [absense of] shielding; it's making
> double
> > sure that at NO place the twists are 'broken'. Even a short stretch
> > of untwisted wire or a loop where the two wires are not snugly
> > together
> but
> > somewhat apart, can (will!) wreak havoc on your connection. This is
> why
> > you should be extra-extra careful when crimping RJ45 plugs; a stretch
> of
> > just 1 (one) centimeter untwisted wire protruding from the plug will
> > kill(!) your good signal, (depending on the level of "noise" around).
>
> Your information is very helpful to me. I just wonder why people told
> me that they even telephone cables and ISDN wall sockets also work...

It all depends on your setting. If you live in a little village you might 
get away with running 10 meters cat1 that happily runs 100base-TX.

On the other hand, if you live next to some factory or plant in a big city 
you may see so much interference that even the position (orientation) 
your cable is in makes a difference (though this would be rather weird)

> I have already spent money in the wrong wall sockets, do you really
> believe that only exchanging them against CAT5 sockets would solve all
> my 20m+10m (wall sockets connection)+20m problems? Do you really think
> that this causes so much loss of signal quality? Getting less then 5
> MBit/s from 100 MBit/s just because of some shielding weakness looks a
> little unbelievable to me, but if this is really true...

That is simple to verify; your're obviously running linux, right ? ;-)  
So, look at the lines from 'ifconfig' (as root) for the ethernetcard.  
These look somewhat like this:

eth0      Link encap:Ethernet  HWaddr 00:E0:7D:B6:2C:6B
          inet addr:10.0.0.1  Bcast:10.0.0.255  Mask:255.255.255.0
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:218423148 errors:0 dropped:0 overruns:0 frame:0
          TX packets:374519874 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:100
          RX bytes:4056969579 (3869.0 Mb)  TX bytes:4168451818 (3975.3 Mb)
          Interrupt:9 Base address:0x1f00

Now the lines you should watch are the ones starting with "RX packets..." 
"TX packets..." and "collisions...". As you see, in my case the numbers 
printed behind "errors", "dropped", "overruns" and "collisions" are zero. 
If you have bad cabling, these will definitely not be zero. You would 
probably see either errors, dropped packets, collisions, or all of them. 
If you do, you must determine if they represent a big enough part of the 
total traffic (the totals printed behind "packets" here). If you have 
like 20 errors from 20000000 packets sent, then all is well. But if you 
happen to get like 1% or 2% errors you definitely have a problem with the 
wires, the cards, or something else. 
And anything over 5% of errors (or collisions, etc.) means that you have 
barely a connection left and you're lucky to get traffic through at all.
Of course, all those numbers will be zero if you just booted, so you have 
to do some tryouts, possibly even with different cable setups, and look 
or count afterwards to see if you had any errors.  

Good luck !

> Additionally, turning your house into a full features multimedia home
> via Ethernet looks quite complicated...

No, it isn't (I did that) but it indeed does help if you have experience 
crimping and such.  Everything you do for the fist time is hard. ;-)
The golden rules are: 
* Avoid using two cables where you can use just one.
* Never, never allow cable to be untwisted unless absolutely neccessary.
* Practice first. Wire some plugs or sockets, you'll get better quick.

> Thanks again (and to all other users in this ML giving feedback)

No problem, glad if I can help :)

Maarten

> Jörg

-- 
This statement is either false or a paradox.


--
Info:
To unsubscribe send a mail to listar@linuxtv.org with "unsubscribe vdr" as subject.