hack a cisco switch to be 802.3af?

Discussion in 'VOIP' started by Barry von Tobel, Oct 22, 2003.

  1. I have quite a few Cisco 3524-PWR ethernet switches and I'd like to use
    them for the IEEE 802.3af Power over Ethenet standard. Can I force
    power on the port, then create a 'dongle' for the cable to put the pins
    on the right wires?

    Barry von Tobel, Oct 22, 2003
    1. Advertisements

  2. Barry von Tobel

    chris Guest

    I don't believe you can force power on. Why not just take a patch
    panel and wire the correct pins to power?
    chris, Oct 23, 2003
    1. Advertisements

  3. the Cisco switch used CDP to determine power requirements, hence the
    proprietary solution. I do beleive you can force it on tho.
    Barry von Tobel, Oct 23, 2003
  4. Barry von Tobel

    chris Guest

    That makes no sense to me. If the phone isn't powered up it can't
    very well talk CDP. Chicken before the egg problem. The CDP comes
    into play after the phone is detected via 802.3af.

    According to the Cisco docs for the 3524, you can only set 'auto' or
    'never'. Here's a Cisco link that explains how the inline power
    detection works. The Cisco 3524 method certainly sounds like 802.3af
    end span detection to me.

    Relevant portions excerpted below.

    Note: There is no such thing as an 'on' mode on any of these devices.
    This should protect customers from accidentally damaging any Ethernet
    NIC cards in devices that do not expect to receive power from the

    The following method for detecting that an IP Phone is connected to a
    10/100 Ethernet port is used by the Catalyst 6000, Catalyst 4000, and
    Catalyst 3524-PWR-XL Switches.

    The port starts the phone-discovery algorithm by sending a special
    Fast Link Pulse (FLP) signal to any device that might be connected to

    The port waits to see if the special FLP signal is forwarded back by a
    connected device. The only devices that are designed to do this are
    devices that expect to receive in-line power.

    If a 79xx IP Phone is connected to the 10/100 Ethernet port, it will
    forward the special FLP signal back to the 10/100 Ethernet port on the
    Catalyst switch. It is capable of doing this because it has a special
    relay that connects its Ethernet receive pair with its Ethernet
    transmit pair. This relay is closed when no power is being supplied to
    the phone. Once power is supplied, this relay remains in an open

    Now that the Catalyst switch has determined that it needs to power the
    port (the special FLP signal was received back from the attached IP
    Phone), the Network Management Processor (NMP) is queried to determine
    if there is any power available to power the IP Phone. Since the NMP
    does not know how much power the IP Phone will need, it uses the
    configured default power allocation. Later on it will adjust this
    allocation based on what the attached IP Phone tells the switch it
    really needs.

    The port then provides power to the IP Phone over pairs 1 and 2 as a
    common mode current.

    The port is taken out of phone-discovery mode and changed to normal
    10/100 Ethernet auto-negotiation mode.

    The instant that the switch applies power to the port, the relay
    inside the phone opens and power begins to flow to the IP Phone.

    At this point a 'wait for link' timer in the switch starts as well.
    The phone has five seconds to establish link integrity on its Ethernet
    port. If the switch doesn't detect link integrity on the port within
    five seconds, it will shut off power to the port and start the
    phone-discovery process all over again. The switch has to wait at
    least five seconds so that the switch has enough time to detect all

    If the switch detects a link within the five second window, it will
    continue to supply power to the IP Phone until it detects a link down

    Once the phone has booted up, it will send a CDP message with a Type,
    Length, Value object (TLV) that tells the switch how much power it
    really needs. The NMP sees this and adjusts the power allocation for
    the port accordingly.

    Note: Only the Catalyst 6000 switch keeps track of how much power has
    been allocated for each device. The Catalyst 4006 and the 3500XL
    switches have enough power available to supply IP Phones on every

    In-Line Power Patch Panel
    The in-line Power Patch Panel (IPPP) uses the unused Ethernet pairs to
    provide in-line power. The IPPP has four rows of RJ-45 connectors each
    with 24 ports in a row. The top two rows are the powered ports used to
    connect to the end device (for example, a 79xx IP Phone). The bottom
    two rows are used to connect to the Switch which will be providing the
    Ethernet connectivity.

    Internally, the IPPP will directly connect the Ethernet pairs from the
    bottom switch port to the corresponding phone port on the top. The
    in-line power patch panel does not interfere with pins 1, 2, 3, and 6
    in any way. It does not monitor link and does not care about
    speed/duplex, because it is completely passive.

    The phone discovery algorithm for the IPPP is similar to the method
    used on Catalyst switches as explained in the previous section. It
    relies on the fact that the phone will loop back a special signal that
    the IPPP sends on its ports. In this case, however, the unused pins 4,
    5, 7 and 8 are used to detect IP Phones. If an IP Phone is detected,
    these pins (wire pairs) are also used to provide power.

    The following method for detecting that an IP Phone is connected to a
    10/100 Ethernet port is used by the The In-line Power Patch Panel

    The IPPP starts the phone discovery sequence at port 1.

    The IPPP sends a 347kHz loopback tone out port 1. The IPPP listens for
    50ms to to determine if the loopback tone is being forwarded back by a
    device that is connected to the port. Only devices that are expecting
    to receive power on these pins will forward the loopback tone to the
    sending device (the IPPP in this case). The IPPP must detect 16
    transitions within a 50ms period of time to verify that it is sensing
    the correct loop back tone and not an anomaly.

    If the IPPP verifies that this received signal is the correct one,
    power is enabled on the port. If the signal is not correct the IPPP
    moves on to the next port and starts the process over again.

    The IPPP continuously cycles through the ports repeating the above
    steps for each port.

    Each port that is supplying power is polled for 50ms every 600ms to
    ensure there is still a device attached. This ensures that power is
    turned off at the port if the device that required it has been
    chris, Oct 24, 2003
  5. Barry von Tobel

    chris Guest

    The Cisco 3524 method certainly sounds like 802.3af end span detection to me.

    Just answered my own question. The Cisco method is different.


    I wonder if you could simply add a small relay to loopback the
    send->receive if power isn't present (ie imitate a 7960 phone).
    chris, Oct 24, 2003
    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.