2
\$\begingroup\$

I'm trying to understand what would prevent a 1 GbE non-optical slip-ring from transmitting 10GbE, 40GbE or even 100/200/400GbE. They all have 'metallic'/copper variants (backplane or short cable) and I can only find up to 1GbE industrial slip-rings in a quick market search.

I'm a slip-ring and 'cabling' noob, I don't have the right lexicon to start with... But if it is 'just' about losses, the slip ring size can't be much more than 7m (the limit on some of those copper modes) and can be reamplified/relayed at input/output of the slip-ring, right?

I'm missing something obvious but I don't know what.

\$\endgroup\$
4
  • 1
    \$\begingroup\$ Is this a rhetorical question, or are you actually in need of passing a 1gig link through a rotating assembly? How fast does it rotate? \$\endgroup\$ Jul 27 at 5:51
  • \$\begingroup\$ Passing a 1GbE through a rotating assembly slip-ring is quite standard these days, but I need to put up an array of very high def/bandwidth sensors, so I need a far higher datarate (up to 2Tbit/s if I go full-spec, can be split in smaller individual streams though of course). Up to 60 rpm. \$\endgroup\$ Jul 27 at 7:48
  • 1
    \$\begingroup\$ Is the high bandwidth requirement unidirectional? \$\endgroup\$ Jul 27 at 8:47
  • \$\begingroup\$ Very good question. No indeed it's unidirectional. Does it change something? \$\endgroup\$ Jul 27 at 9:25
10
\$\begingroup\$

Maximum ethernet bandwidth is about maintaining the characteristic 100Ω (transmission line) impedance of the 10Gbe cable from device to device. As long as that can be done, you could use many different kinds of copper transmission media.

However, this is probably not going to be easy through a slip ring, as by the looks of twinax cable the impedance is tightly controlled. When designing transmission lines any capacitance or inductance that does not look like a 100Ω transmission line will cause reflection and attenuation and cause the signal to degrade and the bandwidth reduced (or no bandwidth at all).

\$\endgroup\$
8
  • 1
    \$\begingroup\$ Some tolerance, the tolerance is buried in some ethernet spec somewhere. \$\endgroup\$
    – Voltage Spike
    Jul 26 at 19:50
  • 1
    \$\begingroup\$ This says 15%, that seems really high to me. xmultiple.com/xwebsite-forum24.htm \$\endgroup\$
    – Voltage Spike
    Jul 26 at 19:51
  • 2
    \$\begingroup\$ I think the hardest part would be avoiding reflections when linking up two different sets of copper. \$\endgroup\$
    – Voltage Spike
    Jul 26 at 19:58
  • 1
    \$\begingroup\$ Change of materials, change of transmission line characteristics (Cat cable has twisted pair differential lines) \$\endgroup\$
    – Voltage Spike
    Jul 26 at 20:02
  • 1
    \$\begingroup\$ @VoltageSpike I suspect the 15% figure is assuming a uniform cable, and the specs on reflection/impedance discontinuities are a lot tighter. \$\endgroup\$
    – mbrig
    Jul 27 at 5:49
3
\$\begingroup\$

LTN's SC020 slip ring claims to support 10GigE

As for why this is unusual - and higher speeds even moreso - you might enjoy reading "Critical parameters for high speed data on slip rings" which doesn't reach the speeds you aspire to, but should give you an overview of the issues involved.

In short, at high data rates a wire behaves less like a hosepipe for electrons, and more like a radio antenna, where you get crosstalk between the different conductors. There's no point in having eight conductors in a cable if the moment you send a signal down one, it makes all the others unusable! Because sliprings have to be ring-shaped, they're limited in what they can do to mitigate crosstalk - especially if you want a large bore diameter.

There's also limited demand - while gigabit ethernet has been standard on essentially every computer made since 2005, I know industrial slipring users who think 1Mbps CANBus is a bit too fast for their tastes. And 10-gigabit ethernet has barely made it outside the data centre. Many slip ring manufacturers have probably never even been asked for a 10GigE product.

\$\endgroup\$
1
  • \$\begingroup\$ Thanks for the insights and link. Almost looks like one should use very short range wireless transmission instead of hoping to manage going through with copper/wire-based protocols. \$\endgroup\$ Jul 27 at 19:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.