How to set up a spare Linksys WRT54Gv2 as a wired access point network extender

Discussion in 'Linux Networking' started by Liam O'Connor, Mar 8, 2014.

  1. Now that's interesting.
    I had almost exactly the same experience.

    The Play Station 3 is only about 15 feet from the spare Linksys
    WRT54G v5 router, and it was the only client that was ungraceful
    about the new network.

    It was on the "old" SSID of FOOBAR, but my kids said it wasn't
    on the network when I added the spare router as a wired repeater.

    So, I instinctively rebooted the PS3, and that's when it started
    giving the (8071002) DNS error. I doublechecked to see if I could
    set the DNS server to 8.8.8.8 on the spare router, but, I was
    surprised there is no way to do that given its configuration
    (DHCP server turned off, Gateway mode, etc.).

    So then I unplugged backward and rebooted forward (typical
    network reboot procedure). Still it gave the 8071002 DNS error.

    So, Googling I found what appeared to be every kid on the planet
    with the same 8071002 DNS error, most of which simply said to
    tell it to readjust the settings. So, that's what I did:
    Settings->Network Settings->Internet Connection Settings->Adjust
    settings->Easy

    This made the 8071002 DNS error go away. But the kids said they
    could now log into the PSN but it would disconnect them randomly.

    I was perplexed because, as with you, the Sony PS3 is only about
    15 feet from the spare router. In addition, I don't know of *any*
    other sources of interference within 40 feet of that PS3.

    Googling, I found the random disconnect also a typical error,
    although not as prevalent as the 8071002 DNS error. The most
    common solution was to disable the media server:
    Settings->Network Settings->Media Server Connection->Disabled

    At the same time I disabled the Media Server on the PS3 (I had
    to look up what that even was), I plugged a patch cable in from
    the spare router, running loosely along the floor, to the PS3).

    The kids haven't complained.
    So, that's how I left it.
    Until the wife finds out (about the loose wire) ... :)
     
    Liam O'Connor, Mar 9, 2014
    #21
    1. Advertisements

  2. Liam O'Connor

    Mark Lloyd Guest

    On 03/08/2014 08:07 PM, Scott Hemphill wrote:

    [snip]
    AFAIK, it's not shielding but differential signals. The current is each
    wire is opposite. The wires are twisted so any noise affects both
    equally and they're opposite so the effect is canceled out.
    Normally, only the first two of those pairs are used for ethernet. I
    suppose the odd pairing in the middle is to prevent problems caused by
    someone mistakenly plugging a phone cable into an ethernet jack (the
    shapes will center it).

    BTW, to make my own ethernet cables, I put the pairs in
    orange-green-blue-brown order (closest to red-green-blue) and put solid
    colors on the lower numbered pin. That's to make it easy to remember.

    --
    Mark Lloyd
    http://notstupid.us

    "Might there have been fewer crimes in the name of Jesus, and more mercy
    in the name of Judas Iscariot?" [Thomas Pynchon]
     
    Mark Lloyd, Mar 9, 2014
    #22
    1. Advertisements

  3. It didn't with phones.

    It didn't with cat 5 or older.

    It certainly DOES with cat 5e or newer.

    The twist per inch rates of each pair DIFFER in the new cabling.
    Those twist rate differences are used by the port hardware to test the
    line integrity during the handshake process.

    So it DOES matter which set of which pair of wires is on which set of
    which pair of pins.

    IF you really do want Gigabit per second performance from your network
    hooks.
     
    DecadentLinuxUserNumeroUno, Mar 9, 2014
    #23
  4. The crosstalk takes place at that connector. The amount of
    non-twisted wire in the termination itself even matters. There is a
    spec for assemblers, and arriving at a properly constructed cat 6
    termination is not merely stuffing them in and crimping.

    After the pairs are matched up to the right pins, the length of the
    'combed straight' set of wires *must* be trimmed back such that it
    barely makes it all the way into the connector.

    Otherwise, it will not perform out to full spec.

    The best connectorization for cat6 includes a SHIELDED termination
    shroud and shielded cable.

    But yeah... it will not halt a connection at someone's home. It just
    will not pass the data at the full rate. Essentially un-noticed since
    that is not where the bottleneck is.
     
    DecadentLinuxUserNumeroUno, Mar 9, 2014
    #24
  5. Not at all...
    The twist rate differs on *any* cable with multiple twisted pairs.

    And there is no easy way to detect which pair has a greater or
    lesser twist than any other pair. The twist rate is not used by
    the port hardware to test line integrity.
    It does not matter in any technical way.

    The coloring is there for easy identification, and of course as
    one would expect there are several different standard ways to
    color code wires. The Blue, Orange, Green, Brown, Slate color
    series are Bell Standard, developed in the US decades before
    EtherNet existed.
     
    Floyd L. Davidson, Mar 9, 2014
    #25
  6. Crosstalk takes place all along a cable. "Near end crosstalk"
    is certainly the most significant though, but that is due to
    higher signal levels on the transmit pairs and lower signal
    levels on the receive pairs than at other points along the
    cable.

    Certainly incorrectly done connectors would add to that.
    There should not be more than 13mm of untwisted wire at the
    connector.
    Shielded Twisted Pair has it's uses, but generally speaking is
    not better than UTP. The length that can be used is reduced,
    due to the added capacitance. STP is used in more severe
    environments.

    For example, in a telecommunications center where there are
    cable racks with literally thousands of pair of wires, long runs
    between different rows of racks will uses shielded pairs, while
    short runs within the same row of racks will use unshielded
    pairs. That is partialy due to the way grounding systems are
    designed, and if for some reason the ground at one end of a row
    is provided differently tha the ground at the other end, then
    STP would be used instead of UTP.
    That is not correct. The point is that the any crosstalk would
    be so rare as to be insignificant.
     
    Floyd L. Davidson, Mar 9, 2014
    #26
  7. Cat 6 has VERY specific twist rates and they differ in the same wire
    very specifically. It is not just some lame manufacturing window the
    maker of the wire needs to comply with to make the claim of it being a
    certain wire. They are specifically (with cat5e and up) two different
    twist rates and belong at different locations on the GbE port and
    therefore also on the connector that gets you into that port.

    Sure, you may get a handshake with your sub orthodox jumper, but why
    differ from what the entire rest of the world does?

    And no, your jumper will not work correctly on a real network link
    that fully taxes the link as a part of its built in reliability testing.

    A bad link anywhere in our gateways gets found pretty much
    instantaneously. Subbing in a sub par jumper gets noted in the
    functional testing, so it must make a difference. Think Nyquist.

    BTW, Floyd.. I consider you as one of the most intelligent folks on
    Usenet and always have for a couple decades now. Back when a.e.e. was
    more popular.
     
    DecadentLinuxUserNumeroUno, Mar 9, 2014
    #27
  8. It makes exactly *no* difference which pair has which twist rate
    or which pair carries which data.

    Twist rates are also defined specifically in other cables, but
    in fact for Cat 5 cable they are defined as:

    Pair Turns per meter
    Blue 64.8
    Orange 56.2
    Green 65.2
    Brown 51.7

    Note that per TIA/EIA-A and TIA/EIA-B the data will be on the
    Orange and Green pairs. Green has the highest twist and Orange is
    the next to the lowest, clearly indicating that twist is not
    significant to the data rate.

    If maximum difference were important, the twist rates for Brown
    and Blue would have been used for Orange and Green. If the hightest
    twist rate was significant, the twist rate for Blue would have been
    assigned to Orange instead. But in fact *any* twist rate above 50 per
    meter is enough.
    The hardware cannot detect which twist rate is used on a pair.
    How does Nyquist relate to cable pair twist?

    The different twist rates are meant to avoid two pairs laying
    too close to each other over any significant distance.

    Using different twist rates has been standard practice with
    multipair twisted pair cable essentially since Alexander Bell
    discovered the advantages of twisted pair cables. (Don't think
    Nyquist, think about open wire line!)
    Intelligence may be overrated, and I certainly don't claim to
    have much. What I do have is a few decades of experience with
    testing comm cabling.
     
    Floyd L. Davidson, Mar 10, 2014
    #28
  9. Liam O'Connor

    Dan Purgert Guest

    Yep, T568A and B. The only difference is where the green and orange
    pairs are.
    Yes, yes, yes. It's kinda like home electrical wiring ... you *could*
    use any color wire you wanted (since "the copper doesn't care"), but the
    standard is:

    Black = Hot
    White = Neutral
    Green (or bare) = Ground.

    110V 3-Way wiring adds in Red (secondary hot)

    220V or higher adds in other "Hot" conductors (I think Yellow, possibly
    Red).

    Granted, this is a "for safety" standard as compared to the "for sanity"
    telecomms standard ... but the idea is the same -- you want to make the
    guy checking it out 2 years down the road to not have to guess at WTF
    you were up to.
    Mainly, it's a "sanity" thing. Also, if you look at them side by side
    (or as close as I can get in text), it might be easier:


    568A: ................. 568B:

    1. Lt. Orange (TX) .... 1. Lt. Green (TX)
    2. Orange (TX) ........ 2. Green (TX)
    3. Lt. Green (RX)...... 3. Lt. Orange (RX)
    4. Blue ............... 4. Blue
    5. Lt. Blue ........... 5. Lt. Blue
    6. Green .............. 6. Orange (RX)
    7. Lt. Brown .......... 7. Lt. Brown
    8. Brown .............. 8. Brown

    As you can (hopefully) see, wiring one end of a cable as 568A and the
    other 568B will create a crossover cable (that is, the Transmit (TX)
    pair on one end is the Receive (RX) pair on the other end). This isn't
    a huge deal since most ports these days are auto-sensing (i.e. they
    figure out which pairs the other end is transmitting on, and set that as
    RX, then TX as the other pair), but it was necessary in the early days
    of Ethernet networking, since most devices were not auto-sensing.

    -Dan
     
    Dan Purgert, Mar 10, 2014
    #29
  10. This is quite interesting!
     
    Liam O'Connor, Mar 10, 2014
    #30
  11. But of course that is not unique to "the new cabling". Multipair
    twisted pair cable has all been made that way for literally
    decades, long before the advent of Ethernet.
     
    Floyd L. Davidson, Mar 10, 2014
    #31

  12. Apparently not to some.

    I can guarantee you it IS of interest to the hardware you are hooking
    into. Some of it even cares one way or the other.
     
    DecadentLinuxUserNumeroUno, Mar 11, 2014
    #32
  13. It is of interest to virtually everyone. It was discovered by
    Alexander Graham Bell himself.
    Show us a cite where *anyone* that is credible demonstrates that
    as a fact.

    I guarantee you the hardware cannot determine the difference in
    twist rate between 50 per meter and 70 per meter. Which is to
    say that it cannot determine which pair is which in a CAT 5
    cable.
     
    Floyd L. Davidson, Mar 11, 2014
    #33
  14. Apparently not.
    You mean the jackass who stole the phone from Meucci?

    Doubtful he came upi with anything himself. He and his employer were
    pure thieves.

    Yeah... I have every confidence that he knew what latency was... NOT!
     
    DecadentLinuxUserNumeroUno, Mar 12, 2014
    #34
  15. I already said that CAT 5 SUCKS!

    Where the **** were you?

    Oh and YES, the HARDWARE DOES test the line and some CAN tell you if
    you are miswired, and nearly all can tell you how far down a line it is
    broken in a bad wire circumstance. Those tests are made more reliably
    by the twist variances.

    Guess some folks haven't been keeping up.
     
    DecadentLinuxUserNumeroUno, Mar 12, 2014
    #35
  16. What does latency have to do with this discussion?
     
    Floyd L. Davidson, Mar 12, 2014
    #36
  17. CAT5 cable does not "suck".
    An EtherNet interface can't even detect a split pair, and it
    certainly cannot determine if the brown pair is exchanged with
    the blue pair.

    So just find a cite to explain anything about how that is done.

    While a normal EtherNet interface does not have that type of
    built in test capability, there are examples of test equipment
    available, that can be manually applied to a line, which will
    test for all variety of wiring errors. Using a Time Domain
    Reflectometer it is possible to locate specific faults such as
    the impedance bump from a broken wire (or for that matter from
    too sharp a bend in the wire).
    Lets go back to the basic question, and restate the facts: The
    specific color and or pair number in a CAT5, CAT5E, or CAT6
    cable is not significant. If different pairs are wired to the
    appropriate pins for a particular data pair the connecting
    interface will still function exactly as it should.

    What that means in practical terms is that a cable can be wired
    as T568A or T568B (or another "standard") as long as both ends
    use the same standard.
     
    Floyd L. Davidson, Mar 12, 2014
    #37
  18. Liam O'Connor

    Eef Hartman Guest

    But the OP was talking about a CAT 6 (six, not five) cable.
    CATegory 6 cabling is for higher network speeds - like 1 GB/s -
    (so called 1000BASE-T). It has a 250 MHz bandwidth as compared to
    only 100 for CAT 5

    PS: in the (still higher speed) CAT 7 cables the individual twisted
    pairs are shielded INSIDE the cable against each other.
    This permits frequencies up to 600 MHz
     
    Eef Hartman, Mar 12, 2014
    #38
  19. His initial statements did not specify any category, and the
    discussion previous had been about CAT 5. When challenged
    he specified CAT5e and later cables.

    The point is *still* the same: the interface hardware cannot
    determine what the twist rate is, does not test which pairs are
    used for which data, and will work perfectly well regardless of
    which cable pair is assigned to which data. And that is by
    design, not accident.
    CAT7 is not UTP, and is outside the range this discussion has
    covered so far.

    Regardless, the target speeds for each category level has not
    been under contention, and anyone can use Google to get more
    information if desired.
     
    Floyd L. Davidson, Mar 12, 2014
    #39

  20. The different characteristic of the two types of tp included in the
    cable. It allows the port hardware to perform the testing I mentioned.
     
    DecadentLinuxUserNumeroUno, Mar 13, 2014
    #40
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.