Single mode, dual fiber optical link downsides

I'm about to get my first articles of a duplex fiber link, ECL in and out at maybe 1M to 2 GHz. It's for testing missiles or something.

I'm using a Cisco 10G 850nm multimode SFP module, which is probably good for a few hundred meters at my speed. It will run through several gymbal joints that have "fiberoptic slip rings"

What's strange is that Amazon wants about $20 for the SFP. Digikey wants $94. The list price is $1200.

How fast do you plan to go? Need DC coupling?

I would strongly advocate for dual fibers (one in each direction) unless you have a specific reason to not use dual fibers.

Single fiber usually means that you need Bi-Directional optics which are more rare and more expensive. They also transmit and receive on different wavelengths. So sparing parts is more complicated because the ends aren't like for like.

I bought four of those a few years ago on eBay for about $10 in total. Hardly anyone uses 850nm multi-mode anymore which is why they are so cheap. Single mode fibre is now usually cheaper than multimode.

John

I see multi-mode / 850 nm / short reach used all the time, including new installations. I've seen a LOT of Fibre Channel make *HEAVY* use of 850 nm.

It all has to do with the distance that you need to go. 850 nm / multi-mode / short reach is almost always cheaper and earlier to appear than single-mode counterparts at faster and faster speeds.

If you need something to go from a switch to equipment in the same cabinet / row / moderate sized data center, then 850 nm / multi-mode / short reach is a perfectly viable candidate.

I've done some 100 Gbps over multi-mode within the last few months.

I think the customer's optical slip ring things only work multimode.

It's sort of mind bogling to make a duplex fiberoptic swivel joint, and pipe light through 3 axes of twirling gymbals.

The only slip ring that I've seen was for a single fiber. Hence why we /had/ /to/ use Bi-Di optics over a single fiber.

The device we were using was single-mode. But I (naively) assume that there are multi-mode counterparts.

See page six:

Schleifring uses the Dove Prism approach. See US Patent 5,157,745 to Gregory Ames of the US Navy. Also see Speer and Koch, "The diversity of fiber optic rotary connectors", SPIE Vol 839, Components for Fiber Optic Applications II, 1987, pages 122-129.

Joe Gwinn

Depending on the application, you can do continuous rotation using the Indian dancer's trick. Try it with a can of soda: hold it by the bottom, up near your ear. Rotate it towards you as you pass it under your armpit and back up next to your ear. Unlimited continuous rotation without a rotating joint.

Cheers

Phil Hobbs

If you have a hollow shaft and really good bearings, you can do that for a single channel--it just has to stay within a micron or two of the rotation axis. MM allows a lot more slop.

Some fibre slip rings intended for solid shafts use one or two MM fibres on the stationary side, illuminating a split fibre bundle that fans out into a ring around the outside of the rotating side, so that at least one Rx fibre is illuminated at all times. That requires a larger detector, which is slower. Alternatively you can do it the other way round, which is faster but wastes beaucoup laser power. It's easier at smaller radii.

One could think of weirder things, e.g. a circular optical waveguide whose core is exposed on one side, and a sliding tap using some coupling fluid and frustrated TIR. Dunno if anybody has done that. It would produce some entertaining data synchronization problems at higher speeds, but probably not as bad as a sliding metal contact.

Cheers

Phil Hobbs

The designs I have seen usually put a collimating lens on the source fibre and pass an expanded beam across the gap between the rotating parts, and then focus it onto the receiving fibre with another lens. That allows a lot more translational slop, though only limited angular slop (which is probably fine if there is more than one rows of balls / rollers in the bearing races on the axis).

Some of the multi-channel ones with geared dove prisms look quite mechanically complicated and expensive, and are probably better avoided by using some some sort of WDM or digital multiplexing. Still, there seem to be a lot of fibre optic slip rings available, with up to 16 channels from several suppliers, in a choice of single or multi-mode versions.

BITD expanded-beam connectors were fairly common, but IIRC Amphenol is the only company making them these days.

A Dove prism rotates the image at twice its own speed, interestingly without doing a great lot to the polarization. (It will go mildly elliptical, but a single TIR bounce off an air/low index glass causes way less retardation than a Fresnel rhomb, which is a pretty good achromatic quarter-wave plate.

Cheers

Phil Hobbs

My friend does not put connectors on; rather he buys pigtails and fusion-splices them onto the run.

He hates to splice multi-mode so badly, we moved to single mode just to keep him happy. When you have someone with the right tools, just do what he wants.

We order what we need from Fiberopticcableshop.