Extending DX link beyond 100m

Forums Forums dLive Forums dLive General Discussions Extending DX link beyond 100m

This topic contains 9 replies, has 4 voices, and was last updated by Profile photo of msteel msteel 1 year ago.

Viewing 10 posts - 1 through 10 (of 10 total)
  • Author
  • #97932
    Profile photo of msteel

    We expect to have a situation where we may want to extend a DX168 with a cable run longer than 100m but less than 200m from the CDM64. Unfortunately (or fortunately?) this would not be a redundant link.

    From here https://support.allen-heath.com/Knowledgebase/Article/View/gigaace-and-dx-networking-vlans-and-fibre-optics it sounds like there are probably a number of inexpensive 10/100/1000 ethernet switches that should do this job. Probably I will want a managed switch so I can turn off “green” features, etc. correct?

    Are there any particular models that are known to be reliable (or unreliable) for this application? Thanks.

    Profile photo of SteffenR

    you don’t have to use a managed switch and only 100Mbit switches will work, no GBit
    or a switch where you can set the port speed…

    from the document

    DX Expanders

    DX links are Fast Ethernet point-to-point connections (100BASE-TX, IEEE 802.3u), Layer 2 compliant.
    Layer 2 network switches and media converters can be used, provided they support Fast Ethernet (100Base-TX) connections.

    I would suggest a network repeater with POE then you don’t need power at the switch position

    Profile photo of msteel

    Thanks for the reply. In the case of a pure DX-DX connection, are there problems with it negotiating to 100mBit correctly in some cases?
    I think I remember reading that since SLink ports use the link speed to determine whether to use DX or GigaACE, they may not always correctly negotiate with a switch; but I would not expect that issue to come up with a pure DX device on both ends.

    I will investigate the ethernet repeater but since in our case we will have power available at the midpoint the POE won’t strictly be necessary.

    Profile photo of SteffenR

    In the case of a pure DX-DX connection, are there problems with it negotiating to 100mBit correctly in some cases?

    It only works with 100Mbit switches.

    Profile photo of Mr-B

    How about using two media converters 100Mb Ethernet to Fibre, in Multimode you should get up to 550m??

    Profile photo of SteffenR

    he asks for a one time solution (at least this is what I think) so a inexpensive Fast Ethernet switch is the solution he needs.

    But the fiber solution is worth to think about, if the solution is a fixed installation.
    This separates the electric systems from each other, so no ground loops are introduced.

    Profile photo of msteel


    We have tested two methods and both appear to work fine so far.

    First, the low cost ethernet switch. The switch I tested is the Netgear GS105E. I chose it because it has some management features. Since it is a gigabit switch I thought I might need to force ports to 100MB for instance, or set up VLANS. None of that was necessary, as I plugged it in between the CDM64 and the DX168 and it “just worked.”

    Second, fiber converters. We realized that the building had an existing run of fiber that was suitable, so we decided that using it was a preferred long term solution to the switch. This is mainly due to the ability to locate the wiring in a more protected manner. We are using a pair of TP-Link MC220L(UN) fiber converters with appropriate SFP optical modules. These came recommended by a network tech not only for their reliability but also for their low cost. I believe another member here was using this family of converters without problems.

    Both methods are using gigabit hardware, so in theory it should also work on a GigaACE link. I do not currently have any plans to do that, though.

    Eventually I would like to set up a test to measure the latency of the DX link over these methods since we may at some point have some microphones on the DX168 and others coming analog to the CDM64.

    Profile photo of msteel

    I finally made a measurement of latency of DX over fiber with the TPLink MC220L(UN) converters. My conclusion is that a pair of these converters increases the DX link latency from the published 8 samples by an additional 10-11 samples, for a total latency of 18-19 samples. A description of how I measured follows.

    DX connections:
    * CDM64_DX3<->CAT6<->Converter<->Fiber<->Converter<->CAT6<->DX168

    Analog Audio Connections:
    * DX3 Out 1 plugged up to an analog Y cable
    * First out of the Y connected to an old analog line estimated as 300ft long, patched to a CDM input socket.
    * Second out of the Y connected to an input socket on the DX168.

    Signal Routing:
    * Mix Rack oscillator set to 440Hz sine routed to an aux.
    * That aux routed to an output on the DX168
    * CDM In 59 routed to channel 59, Pad in, gain at -4, Routed to Left record bus
    * DX3 In 16 routed to channel 112, Pad in, gain at -4, Routed to Right record bus
    * Left and Right Record routed via Dante card (at 48kHz) to two tracks in Pro Tools.

    * I recorded in Pro Tools and then unmuted the oscillator.
    * I then selected from a zero crossing on the left channel to a zero crossing on the right channel and found the difference to be somewhat more than 9 samples. Best estimate is 9.25 samples. Since my Dante card is at 48kHz this corresponds to 18.5 samples on the 96kHz DX link. The fractional sample is surprising.

    * The signal path is identical up until the Y cable.
    * The signal path is equivalent from the input patch through the mix engine all the way to Pro Tools.
    * The DX link is published to introduce 8 samples of latency at 96kHz.
    * The conversion to/from fiber is expected to introduce additional latency.
    * The long analog line between the DX168 and the Mix rack theoretically could introduce propagation delay. Expected value based on DX168 output impedance of 75ohms and 300ft of cable with 30pF/ft gives 340ns or 0.03 samples at 96kHz. This should not be measurable.
    * So, the half sample is hard to account for. The analog propagation delay should not be anywhere near a half sample. At the same time the digital latency should occur in whole samples. I don’t know what to make of it.

    Profile photo of Dave

    If it’s not too much trouble, I’m curious about what the latency is with the switch instead of the fiber converters.

    Profile photo of msteel

    I’ll have to see if I can set that up again. I don’t have all the cables in place for that method anymore, but on the other hand it shouldn’t take a huge amount to do either.

Viewing 10 posts - 1 through 10 (of 10 total)

You must be logged in to reply to this topic.